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-vectorizer.h (_loop_vec_info): Add a versioning_threshold
7 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
8 (vect_loop_versioning): Take the loop versioning threshold as a
10 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
11 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13 (vect_analyze_loop_2): Compute the loop versioning threshold
14 whenever loop versioning is needed, and store it in the new
15 field rather than combining it with the cost model threshold.
16 (vect_transform_loop): Update call to vect_loop_versioning.
17 Try to combine the loop versioning and cost thresholds here.
19 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
20 Alan Hayward <alan.hayward@arm.com>
21 David Sherwood <david.sherwood@arm.com>
23 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
24 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
25 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
26 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
27 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
28 (group_compare_offset): Update accordingly.
29 (split_small_address_groups_p): Likewise.
30 (record_use): Take addr_offset as a poly_uint64 rather than
31 an unsigned HOST_WIDE_INT.
32 (strip_offset): Return the offset as a poly_uint64 rather than
33 an unsigned HOST_WIDE_INT.
34 (record_group_use, split_address_groups): Track polynomial offsets.
35 (add_iv_candidate_for_use): Likewise.
36 (addr_offset_valid_p): Take the offset as a poly_int64 rather
38 (strip_offset_1): Return the offset as a poly_int64 rather than
41 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
42 Alan Hayward <alan.hayward@arm.com>
43 David Sherwood <david.sherwood@arm.com>
45 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
46 rather than a HOST_WIDE_INT.
47 * tree.c (get_binfo_at_offset): Likewise.
49 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
50 Alan Hayward <alan.hayward@arm.com>
51 David Sherwood <david.sherwood@arm.com>
53 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
54 rather than a HOST_WIDE_INT.
55 * tree-sra.c (build_ref_for_offset): Likewise.
57 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
58 Alan Hayward <alan.hayward@arm.com>
59 David Sherwood <david.sherwood@arm.com>
61 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
63 * tree.c (mem_ref_offset): Likewise.
64 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
65 * builtins.c (get_object_alignment_2): Likewise.
66 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
67 * gimple-fold.c (get_base_constructor): Likewise.
68 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
69 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
71 * ipa-polymorphic-call.c
72 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
73 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
74 (get_ancestor_addr_info): Likewise.
75 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
77 * tree-data-ref.c (dr_analyze_innermost): Likewise.
78 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
79 * tree-eh.c (tree_could_trap_p): Likewise.
80 * tree-object-size.c (addr_object_size): Likewise.
81 * tree-ssa-address.c (copy_ref_info): Likewise.
82 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
83 (indirect_refs_may_alias_p): Likewise.
84 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
85 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
86 (non_rewritable_mem_ref_base): Likewise.
87 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
88 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
89 * varasm.c (decode_addr_const): Likewise.
91 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
92 Alan Hayward <alan.hayward@arm.com>
93 David Sherwood <david.sherwood@arm.com>
95 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
98 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
99 Alan Hayward <alan.hayward@arm.com>
100 David Sherwood <david.sherwood@arm.com>
102 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
103 as a poly_uint64 rather than an unsigned int.
104 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
106 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
107 Alan Hayward <alan.hayward@arm.com>
108 David Sherwood <david.sherwood@arm.com>
110 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
112 (decode_addr_const): Update accordingly.
114 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
115 Alan Hayward <alan.hayward@arm.com>
116 David Sherwood <david.sherwood@arm.com>
118 * tree.h (bit_field_size, bit_field_offset): New functions.
119 * hsa-gen.c (gen_hsa_addr): Use them.
120 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
121 (simplify_vector_constructor): Likewise.
122 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
123 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
124 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
125 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
128 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
129 Alan Hayward <alan.hayward@arm.com>
130 David Sherwood <david.sherwood@arm.com>
132 * expr.h (emit_group_load, emit_group_load_into_temps)
133 (emit_group_store): Take the size as a poly_int64 rather than an int.
134 * expr.c (emit_group_load_1, emit_group_load): Likewise.
135 (emit_group_load_into_temp, emit_group_store): Likewise.
137 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
138 Alan Hayward <alan.hayward@arm.com>
139 David Sherwood <david.sherwood@arm.com>
141 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
142 an unsigned int to a poly_uint64.
143 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
144 sizes as poly_uint64s rather than unsigned ints.
145 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
148 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
149 Alan Hayward <alan.hayward@arm.com>
150 David Sherwood <david.sherwood@arm.com>
152 * lra-constraints.c (emit_inc): Change inc_amount from an int
155 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
156 Alan Hayward <alan.hayward@arm.com>
157 David Sherwood <david.sherwood@arm.com>
159 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
161 (add_stack_var, stack_var_cmp, partition_stack_vars)
162 (dump_stack_var_partition): Update accordingly.
163 (alloc_stack_frame_space): Take the size as a poly_int64 rather
164 than a HOST_WIDE_INT.
165 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
166 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
167 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
168 rather than a HOST_WIDE_INT.
170 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
171 Alan Hayward <alan.hayward@arm.com>
172 David Sherwood <david.sherwood@arm.com>
174 * target.def (return_pops_args): Treat both the input and output
175 sizes as poly_int64s rather than HOST_WIDE_INTS.
176 * targhooks.h (default_return_pops_args): Update accordingly.
177 * targhooks.c (default_return_pops_args): Likewise.
178 * doc/tm.texi: Regenerate.
179 * emit-rtl.h (incoming_args): Change pops_args, size and
180 outgoing_args_size from int to poly_int64_pod.
181 * function.h (expr_status): Change x_pending_stack_adjust and
182 x_stack_pointer_delta from int to poly_int64.
183 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
184 (ARGS_SIZE_RTX): Update accordingly.
185 * calls.c (highest_outgoing_arg_in_use): Change from int to
187 (stack_usage_watermark, stored_args_watermark): New variables.
188 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
189 (emit_call_1): Change the stack_size and rounded_stack_size
190 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
192 (save_fixed_argument_area): Check stack_usage_watermark.
193 (initialize_argument_information): Change old_pending_adj from
194 a HOST_WIDE_INT * to a poly_int64_pod *.
195 (compute_argument_block_size): Return the size as a poly_int64
197 (finalize_must_preallocate): Track polynomial argument sizes.
198 (compute_argument_addresses): Likewise.
199 (internal_arg_pointer_based_exp): Track polynomial offsets.
200 (mem_overlaps_already_clobbered_arg_p): Rename to...
201 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
202 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
203 Check stored_args_used_watermark.
204 (load_register_parameters): Update accordingly.
205 (check_sibcall_argument_overlap_1): Likewise.
206 (combine_pending_stack_adjustment_and_call): Take the unadjusted
207 args size as a poly_int64 rather than an int. Return a bool
208 indicating whether the optimization was possible and return
209 the new adjustment by reference.
210 (check_sibcall_argument_overlap): Track polynomail argument sizes.
211 Update stored_args_watermark.
212 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
213 (expand_call): Likewise. Maintain stack_usage_watermark and
214 stored_args_watermark. Update calls to
215 combine_pending_stack_adjustment_and_call.
216 (emit_library_call_value_1): Handle polynomial argument sizes.
217 Call stack_region_maybe_used_p and mark_stack_region_used.
218 Maintain stack_usage_watermark.
219 (store_one_arg): Likewise. Update call to
220 mem_overlaps_already_clobbered_arg_p.
221 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
223 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
224 * config/microblaze/microblaze.c (microblaze_function_prologue):
226 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
227 TARGET_RETURN_POPS_ARGS interface.
228 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
230 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
231 * config/i386/i386.c (ix86_return_pops_args): Update for new
232 TARGET_RETURN_POPS_ARGS interface.
233 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
234 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
235 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
236 TARGET_RETURN_POPS_ARGS interface.
237 * config/vax/vax.c (vax_return_pops_args): Likewise.
238 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
239 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
240 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
241 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
242 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
243 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
244 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
245 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
246 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
247 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
248 and x_stack_pointer_delta from int to poly_int64.
249 * dojump.c (do_pending_stack_adjust): Update accordingly.
250 * explow.c (allocate_dynamic_stack_space): Handle polynomial
251 stack_pointer_deltas.
252 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
253 (pad_to_arg_alignment): Track polynomial offsets.
254 (assign_parm_find_stack_rtl): Likewise.
255 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
256 * toplev.c (output_stack_usage): Update reference to
257 current_function_pushed_stack_size.
259 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
260 Alan Hayward <alan.hayward@arm.com>
261 David Sherwood <david.sherwood@arm.com>
263 * function.c (in_arg_offset, var_offset, dynamic_offset)
264 (out_arg_offset, cfa_offset): Change from int to poly_int64.
265 (instantiate_new_reg): Return the new offset as a poly_int64_pod
266 rather than a HOST_WIDE_INT.
267 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
268 (instantiate_virtual_regs_in_insn): Likewise.
270 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
271 Alan Hayward <alan.hayward@arm.com>
272 David Sherwood <david.sherwood@arm.com>
274 * rtl.h (get_args_size, add_args_size_note): New functions.
275 (find_args_size_adjust): Return a poly_int64 rather than a
277 (fixup_args_size_notes): Likewise. Make the same change to the
278 end_args_size parameter.
279 * rtlanal.c (get_args_size, add_args_size_note): New functions.
280 * builtins.c (expand_builtin_trap): Use add_args_size_note.
281 * calls.c (emit_call_1): Likewise.
282 * explow.c (adjust_stack_1): Likewise.
283 * cfgcleanup.c (old_insns_match_p): Update use of
284 find_args_size_adjust.
285 * combine.c (distribute_notes): Track polynomial arg sizes.
286 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
287 end_true_args_size, beg_delay_args_size and end_delay_args_size
288 from HOST_WIDE_INT to poly_int64.
289 (add_cfi_args_size): Take the args_size as a poly_int64 rather
290 than a HOST_WIDE_INT.
291 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
292 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
293 polynomial arg sizes.
294 * emit-rtl.c (try_split): Use get_args_size.
295 * recog.c (peep2_attempt): Likewise.
296 * reload1.c (reload_as_needed): Likewise.
297 * expr.c (find_args_size_adjust): Return the adjustment as a
298 poly_int64 rather than a HOST_WIDE_INT.
299 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
300 to a poly_int64 and change the return type in the same way.
301 (emit_single_push_insn): Track polynomial arg sizes.
303 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
304 Alan Hayward <alan.hayward@arm.com>
305 David Sherwood <david.sherwood@arm.com>
307 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
308 from HOST_WIDE_INT to poly_int64.
309 * expr.c (push_block, emit_push_insn): Likewise.
311 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * function.h (frame_space): Change start and length from HOST_WIDE_INT
317 (get_frame_size): Return the size as a poly_int64 rather than a
319 (frame_offset_overflow): Take the offset as a poly_int64 rather
320 than a HOST_WIDE_INT.
321 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
322 (assign_stack_temp): Likewise for the size.
323 * function.c (get_frame_size): Return a poly_int64 rather than
325 (frame_offset_overflow): Take the offset as a poly_int64 rather
326 than a HOST_WIDE_INT.
327 (try_fit_stack_local): Take the start, length and size as poly_int64s
328 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
329 rather than a HOST_WIDE_INT.
330 (add_frame_space): Take the start and end as poly_int64s rather than
332 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
333 (assign_stack_temp): Likewise for the size.
334 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
336 (find_temp_slot_from_address): Handle polynomial offsets.
337 (combine_temp_slots): Likewise.
338 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
340 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
341 poly_int64 rather than a HOST_WIDE_INT.
342 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
343 than a HOST_WIDE_INT.
344 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
345 polynomial frame offsets.
346 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
347 as a poly_int64 rather than an int.
348 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
349 * config/v850/v850-protos.h (compute_frame_size): Likewise.
350 * config/v850/v850.c (compute_frame_size): Likewise.
351 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
352 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
353 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
354 * config/pa/pa.c (pa_compute_frame_size): Likewise.
355 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
356 rather than a HOST_WIDE_INT.
357 * explow.c (get_dynamic_stack_base): Likewise.
358 * final.c (final_start_function): Use the constant lower bound
359 of the frame size for -Wframe-larger-than.
360 * ira.c (do_reload): Adjust for new get_frame_size return type.
361 * lra.c (lra): Likewise.
362 * reload1.c (reload): Likewise.
363 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
364 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
365 * rtlanal.c (get_initial_register_offset): Return the offset as
366 a poly_int64 rather than a HOST_WIDE_INT.
368 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
369 Alan Hayward <alan.hayward@arm.com>
370 David Sherwood <david.sherwood@arm.com>
372 * reload1.c (elim_table): Change initial_offset, offset and
373 previous_offset from HOST_WIDE_INT to poly_int64_pod.
374 (offsets_at): Change the target array's element type from
375 HOST_WIDE_INT to poly_int64_pod.
376 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
377 (elimination_costs_in_insn, update_eliminable_offsets)
378 (verify_initial_elim_offsets, set_offsets_for_label)
379 (init_eliminable_invariants): Update after above changes.
381 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
382 Alan Hayward <alan.hayward@arm.com>
383 David Sherwood <david.sherwood@arm.com>
385 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
386 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
387 (decomposition): Change start and end from HOST_WIDE_INT
389 (decompose, immune_p): Update accordingly.
390 (find_inc_amount): Return a poly_int64 rather than an int.
391 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
394 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
395 Alan Hayward <alan.hayward@arm.com>
396 David Sherwood <david.sherwood@arm.com>
398 * tree.h (get_inner_reference): Return the bitsize and bitpos
399 as poly_int64_pods rather than HOST_WIDE_INT.
400 * fold-const.h (ptr_difference_const): Return the pointer difference
401 as a poly_int64_pod rather than a HOST_WIDE_INT.
402 * expr.c (get_inner_reference): Return the bitsize and bitpos
403 as poly_int64_pods rather than HOST_WIDE_INT.
404 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
406 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
407 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
408 (optimize_bit_field_compare): Update call to get_inner_reference.
409 (decode_field_reference): Likewise.
410 (fold_unary_loc): Track polynomial offsets and sizes.
411 (split_address_to_core_and_offset): Return the bitpos as a
412 poly_int64_pod rather than a HOST_WIDE_INT.
413 (ptr_difference_const): Likewise for the pointer difference.
414 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
415 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
416 * dbxout.c (dbxout_expand_expr): Likewise.
417 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
418 (loc_list_from_tree_1, fortran_common): Likewise.
419 * gimple-laddress.c (pass_laddress::execute): Likewise.
420 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
421 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
422 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
423 * tree-affine.c (tree_to_aff_combination): Likewise.
424 (get_inner_reference_aff): Likewise.
425 * tree-data-ref.c (split_constant_offset_1): Likewise.
426 (dr_analyze_innermost): Likewise.
427 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
428 * tree-sra.c (ipa_sra_check_caller): Likewise.
429 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
430 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
431 (instrument_bool_enum_load, instrument_object_size): Likewise.
432 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
433 to get_inner_reference.
434 * hsa-gen.c (gen_hsa_addr): Likewise.
435 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
436 * tsan.c (instrument_expr): Likewise.
437 * match.pd: Update call to ptr_difference_const.
439 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
440 Alan Hayward <alan.hayward@arm.com>
441 David Sherwood <david.sherwood@arm.com>
443 * fold-const.c (fold_comparison): Track sizes and offsets as
444 poly_int64s rather than HOST_WIDE_INTs when folding address
447 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
448 Alan Hayward <alan.hayward@arm.com>
449 David Sherwood <david.sherwood@arm.com>
451 * expr.h (get_bit_range): Return the bitstart and bitend as
452 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
453 as a poly_int64 rather than a HOST_WIDE_INT.
454 * expr.c (get_bit_range): Likewise.
455 (expand_assignment): Update call accordingly.
456 * fold-const.c (optimize_bit_field_compare): Likewise.
458 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
463 (aarch64_legitimate_address_p): Use it instead of an rtx code,
464 as an optional final parameter.
465 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
466 (aarch64_legitimate_address_p): Likewise.
467 (aarch64_print_address_internal): Take an aarch64_addr_query_type
468 instead of an rtx code.
469 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
470 (aarch64_legitimate_address_hook_p): Likewise.
471 (aarch64_print_ldpstp_address): Likewise.
472 (aarch64_print_operand_address): Likewise.
473 (aarch64_address_cost): Likewise.
474 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
475 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
476 (aarch64_mem_pair_lanes_operand): Likewise.
478 2017-12-20 Richard Biener <rguenther@suse.de>
480 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
481 update_stmt_if_modified.
483 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
485 PR tree-optimization/83491
486 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
487 before walking uses. Improve coding style and comments.
489 2017-12-20 Tom de Vries <tom@codesourcery.com>
491 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
493 2017-12-20 Jakub Jelinek <jakub@redhat.com>
496 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
498 (pass_ipa_free_fn_summary): Add small_p private data member,
499 initialize to false in the ctor.
500 (pass_ipa_free_fn_summary::clone,
501 pass_ipa_free_fn_summary::set_pass_param,
502 pass_ipa_free_fn_summary::gate): New methods.
503 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
504 | TODO_dump_symtab if small_p.
505 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
506 entry and add another instance of the pass with false parm after
508 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
509 ipa_free_fn_summary here.
511 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
513 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
515 2017-12-20 Martin Sebor <msebor@redhat.com>
518 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
519 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
520 (expand_builtin_strncmp): Same.
522 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
525 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
526 allowed debug stmts before labels.
527 (expand_gimple_basic_block): Likewise.
528 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
529 * gimple-iterator.h (gsi_after_labels): Likewise.
530 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
531 rename reused variable, and simplify using gsi_move_before.
532 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
533 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
534 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
535 (gimple_verify_flow_info, gimple_block_label): Likewise.
536 (make_blocks): Move debug markers after adjacent labels.
537 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
538 allowed debug insns outside blocks.
539 * df-scan.c (df_insn_delete): Likewise.
540 * lra-constraints.c (update_ebb_live_info): Likewise.
541 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
542 (vt_initialize, delete_vta_debug_insns): Likewise.
543 (reemit_marker_as_note): Drop BB parm. Adjust callers.
545 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
546 Alan Hayward <alan.hayward@arm.com>
547 David Sherwood <david.sherwood@arm.com>
549 * poly-int-types.h (round_down_to_byte_boundary): New macro.
550 (round_up_to_byte_boundary): Likewise.
551 * expr.h (get_bit_range): Add temporary shim.
552 * gimple-ssa-store-merging.c (store_operand_info): Change the
553 bitsize, bitpos, bitregion_start and bitregion_end fields from
554 unsigned HOST_WIDE_INT to poly_uint64.
555 (merged_store_group): Likewise load_align_base.
556 (compatible_load_p, compatible_load_p): Update accordingly.
557 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
558 (split_group, imm_store_chain_info::output_merged_store): Likewise.
559 (mem_valid_for_store_merging): Return the bitsize, bitpos,
560 bitregion_start and bitregion_end as poly_uint64s rather than
561 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
562 (handled_load): Take the bitsize, bitpos,
563 bitregion_start and bitregion_end as poly_uint64s rather than
564 unsigned HOST_WIDE_INTs.
565 (pass_store_merging::process_store): Update call to
566 mem_valid_for_store_merging.
568 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
569 Alan Hayward <alan.hayward@arm.com>
570 David Sherwood <david.sherwood@arm.com>
572 * builtins.c (get_object_alignment_2): Track polynomial offsets
573 and sizes. Update the alignment handling.
575 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
576 Alan Hayward <alan.hayward@arm.com>
577 David Sherwood <david.sherwood@arm.com>
579 * tree.h (get_inner_reference): Add a version that returns the
580 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
581 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
582 the case in which bitpos is not associated with the first byte.
584 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
585 Alan Hayward <alan.hayward@arm.com>
586 David Sherwood <david.sherwood@arm.com>
588 * tree-affine.h (get_inner_reference_aff): Return the size as a
590 * tree-affine.c (get_inner_reference_aff): Likewise.
591 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
592 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
594 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
595 Alan Hayward <alan.hayward@arm.com>
596 David Sherwood <david.sherwood@arm.com>
598 * fold-const.c (pointer_may_wrap_p): Take the offset as a
599 HOST_WIDE_INT rather than a poly_int64.
601 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
602 Alan Hayward <alan.hayward@arm.com>
603 David Sherwood <david.sherwood@arm.com>
605 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
606 HOST_WIDE_INT to poly_int64_pod.
607 (perform_symbolic_merge): Update accordingly.
608 (bswap_replace): Likewise.
610 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
611 Alan Hayward <alan.hayward@arm.com>
612 David Sherwood <david.sherwood@arm.com>
614 * tree-affine.h (aff_tree::offset): Change from widest_int
616 (wide_int_ext_for_comb): Delete.
617 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
618 constants as poly_widest_int rather than widest_int.
619 (aff_combination_constant_multiple_p): Return the multiplier
620 as a poly_widest_int.
621 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
623 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
624 version static and add an overload for poly_widest_int.
625 (aff_combination_const, aff_combination_add_cst)
626 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
627 the constants as poly_widest_int rather than widest_int.
628 (tree_to_aff_combination): Generalize INTEGER_CST case to
630 (aff_combination_to_tree): Track offsets as poly_widest_ints.
631 (aff_combination_add_product, aff_combination_mult): Handle
633 (aff_combination_constant_multiple_p): Return the multiplier
634 as a poly_widest_int.
635 * tree-predcom.c (determine_offset): Return the offset as a
637 (split_data_refs_to_components, suitable_component_p): Update
639 (valid_initializer_p): Update call to
640 aff_combination_constant_multiple_p.
641 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
642 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
643 as a poly_int64 rather than a HOST_WIDE_INT.
644 (get_address_cost): Handle polynomial offsets.
645 (iv_elimination_compare_lt): Likewise.
646 (rewrite_use_nonlinear_expr): 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_addr_base_and_unit_offset_1): Return the offset
653 as a poly_int64_pod rather than a HOST_WIDE_INT.
654 (get_addr_base_and_unit_offset): Likewise.
655 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
656 (get_addr_base_and_unit_offset): Likewise.
657 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
658 to poly_int64 in example.
659 * fold-const.c (fold_binary_loc): Update call to
660 get_addr_base_and_unit_offset.
661 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
662 (maybe_canonicalize_mem_ref_addr): Likewise.
663 (gimple_fold_stmt_to_constant_1): Likewise.
664 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
666 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
667 * match.pd: Likewise.
668 * omp-low.c (lower_omp_target): Likewise.
669 * tree-sra.c (build_ref_for_offset): Likewise.
670 (build_debug_ref_for_model): Likewise.
671 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
672 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
673 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
674 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
675 (constant_pointer_difference): Likewise.
676 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
677 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
678 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
679 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
680 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
681 (set_ssa_val_to): Likewise.
682 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
683 (maybe_diag_stxncpy_trunc): Likewise.
684 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
685 * tree.c (build_simple_mem_ref_loc): Likewise.
686 (array_at_struct_end_p): Likewise.
688 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
689 Alan Hayward <alan.hayward@arm.com>
690 David Sherwood <david.sherwood@arm.com>
692 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
693 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
694 (get_ref_base_and_extent_hwi): Declare.
695 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
696 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
697 (get_ref_base_and_extent_hwi): New function.
698 * cfgexpand.c (expand_debug_expr): Update call to
699 get_ref_base_and_extent.
700 * dwarf2out.c (add_var_loc_to_decl): Likewise.
701 * gimple-fold.c (get_base_constructor): Return the offset as a
702 poly_int64_pod rather than a HOST_WIDE_INT.
703 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
704 * ipa-polymorphic-call.c
705 (ipa_polymorphic_call_context::set_by_invariant)
706 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
707 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
708 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
709 rather than get_ref_base_and_extent.
710 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
711 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
712 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
714 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
715 call to get_ref_base_and_extent.
716 * tree-sra.c (create_access, get_access_for_expr): Likewise.
717 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
718 (stmt_kills_ref_p): Likewise.
719 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
720 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
722 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
723 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
724 when calling native_encode_expr.
725 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
726 call to get_ref_base_and_extent.
727 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
728 get_ref_base_and_extent.
729 * var-tracking.c (track_expr_p): 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 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
736 HOST_WIDE_INT to poly_int64_pod.
737 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
738 polynomail parameter offsets.
740 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
741 Alan Hayward <alan.hayward@arm.com>
742 David Sherwood <david.sherwood@arm.com>
744 * gengtype.c (main): Handle poly_int64_pod.
745 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
746 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
747 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
748 from HOST_WIDE_INT to poly_int64_pod.
749 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
750 (copy_cfa): New function.
751 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
752 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
753 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
754 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
755 (initial_return_save): Treat offsets as poly_ints.
756 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
757 offset is nonconstant.
758 (reg_save): Take the offset as a poly_int64. Fall back to
759 DW_CFA_expression for nonconstant offsets.
760 (queue_reg_save): Take the offset as a poly_int64.
761 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
763 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
764 Alan Hayward <alan.hayward@arm.com>
765 David Sherwood <david.sherwood@arm.com>
767 * rtl.h (operand_subword, operand_subword_force): Take the offset
768 as a poly_uint64 an unsigned int.
769 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
771 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
772 Alan Hayward <alan.hayward@arm.com>
773 David Sherwood <david.sherwood@arm.com>
775 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
776 'p' format code. Use INT_LIST rather than SUBREG as the example of
777 a code with an XINT and an XEXP. Remove the implication that
778 accessing an rtx field using XINT is expected to work.
779 * rtl.def (SUBREG): Change format from "ei" to "ep".
780 * rtl.h (rtunion::rt_subreg): New field.
781 (XCSUBREG): New macro.
782 (SUBREG_BYTE): Use it.
783 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
784 Update constructor accordingly.
785 (subreg_shape::operator ==): Update accordingly.
786 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
787 than an unsigned int.
788 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
789 a poly_uint64 rather than an unsigned int.
790 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
791 than an unsigned int.
792 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
793 (subreg_size_highpart_offset): Return a poly_uint64 rather than
794 an unsigned int. Take the sizes as poly_uint64s.
795 (subreg_offset_from_lsb): Return a poly_uint64 rather than
796 an unsigned int. Take the shift as a poly_uint64 rather than
798 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
799 as a poly_uint64 rather than an unsigned int.
800 (simplify_subreg_regno): Likewise.
801 (byte_lowpart_offset): Return the memory offset as a poly_int64
803 (subreg_memory_offset): Likewise. Take the subreg offset as a
804 poly_uint64 rather than an unsigned int.
805 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
806 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
807 poly_uint64 rather than an unsigned int.
808 * rtl.c (rtx_format): Describe 'p' in comment.
809 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
810 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
811 offset as a poly_uint64 rather than an unsigned int.
812 (byte_lowpart_offset): Return the memory offset as a poly_int64
814 (subreg_memory_offset): Likewise. Take the subreg offset as a
815 poly_uint64 rather than an unsigned int.
816 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
817 mode sizes as poly_uint64s rather than unsigned ints. Return a
818 poly_uint64 rather than an unsigned int.
819 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
820 (copy_insn_1): Handle 'p'.
821 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
822 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
823 an unsigned int. Return the shift in the same way.
824 (subreg_lsb): Return the shift as a poly_uint64 rather than an
826 (subreg_size_offset_from_lsb): Take the sizes and shift as
827 poly_uint64s rather than unsigned ints. Return the offset as
829 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
830 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
832 * rtlhash.c (add_rtx): Handle 'p'.
833 * genemit.c (gen_exp): Likewise.
834 * gengenrtl.c (type_from_format, gendef): Likewise.
835 * gensupport.c (subst_pattern_match, get_alternatives_number)
836 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
837 (subst_dup): Likewise.
838 * gengtype.c (adjust_field_rtx_def): Likewise.
839 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
840 (match_pattern_2): Likewise.
841 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
842 (rtx_test::subreg_field): New function.
843 (operator ==, safe_to_hoist_p, transition_parameter_type)
844 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
845 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
846 * genpeep.c (match_rtx): Likewise.
847 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
848 (rtx_writer::print_rtx_operand): Handle 'p'.
849 (print_value): Handle SUBREG.
850 * read-rtl.c (apply_int_iterator): Likewise.
851 (rtx_reader::read_rtx_operand): Handle 'p'.
852 * alias.c (rtx_equal_for_memref_p): Likewise.
853 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
854 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
856 * calls.c (expand_call): Likewise.
857 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
858 (make_extraction, gen_lowpart_for_combine): Likewise.
859 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
861 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
862 rather than an unsigned int. Treat subreg offsets as poly_ints.
863 (exp_equiv_p): Handle 'p'.
864 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
865 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
866 * dse.c (find_shift_sequence): Likewise.
867 * dwarf2out.c (rtl_for_decl_location): Likewise.
868 * expmed.c (extract_low_bits): Likewise.
869 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
870 (expand_expr_real_2): Likewise.
871 * final.c (alter_subreg): Likewise.
872 (leaf_renumber_regs_insn): Handle 'p'.
873 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
874 Treat subreg offsets as poly_ints.
875 * fwprop.c (forward_propagate_and_simplify): Likewise.
876 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
877 * ira.c (get_subreg_tracking_sizes): Likewise.
878 * ira-conflicts.c (go_through_subreg): Likewise.
879 * ira-lives.c (process_single_reg_class_operands): Likewise.
880 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
881 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
882 as a poly_uint64 rather than an unsigned int.
883 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
884 subreg offsets as poly_ints.
885 * lra-constraints.c (operands_match_p): Handle 'p'.
886 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
887 * lra-spills.c (assign_mem_slot): Likewise.
888 * postreload.c (move2add_valid_value_p): Likewise.
889 * recog.c (general_operand, indirect_operand): Likewise.
890 * regcprop.c (copy_value, maybe_mode_change): Likewise.
891 (copyprop_hardreg_forward_1): Likewise.
892 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
893 (record_subregs_of_mode): Likewise.
894 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
895 * reload.c (operands_match_p): Handle 'p'.
896 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
897 * reload1.c (alter_reg, choose_reload_regs): Likewise.
898 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
899 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
900 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
901 (simplify_const_poly_int_tests<N>::run): Likewise.
902 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
903 a poly_uint64 rather than an unsigned int.
904 * valtrack.c (debug_lowpart_subreg): Likewise.
905 * var-tracking.c (var_lowpart): Likewise.
906 (loc_cmp): Handle 'p'.
908 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
909 Alan Hayward <alan.hayward@arm.com>
910 David Sherwood <david.sherwood@arm.com>
912 * ira.c (get_subreg_tracking_sizes): New function.
913 (init_live_subregs): Take an integer size rather than a register.
914 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
915 to init_live_subregs.
917 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
918 Alan Hayward <alan.hayward@arm.com>
919 David Sherwood <david.sherwood@arm.com>
921 * expr.c (store_constructor_field): Change bitsize from a
922 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
923 HOST_WIDE_INT to a poly_int64.
924 (store_constructor): Change size from a HOST_WIDE_INT to
926 (store_field): Likewise bitsize and bitpos.
928 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
929 Alan Hayward <alan.hayward@arm.com>
930 David Sherwood <david.sherwood@arm.com>
932 * expmed.h (store_bit_field): Change bitregion_start and
933 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
934 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
935 (store_bit_field_1, store_integral_bit_field, store_bit_field)
936 (store_fixed_bit_field, store_split_bit_field): Likewise.
937 * expr.c (store_constructor_field, store_field): Likewise.
938 (optimize_bitfield_assignment_op): Likewise. Make the same change
939 to bitsize and bitpos.
940 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
941 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
942 same change in the constructor arguments.
943 (get_best_mode): Change bitregion_start and bitregion_end from
944 unsigned HOST_WIDE_INT to poly_uint64.
945 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
946 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
948 (bit_field_mode_iterator::next_mode): Update for new types
949 of m_bitregion_start and m_bitregion_end.
950 (get_best_mode): Change bitregion_start and bitregion_end from
951 unsigned HOST_WIDE_INT to poly_uint64.
953 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * rtl.h (simplify_gen_subreg): Add a temporary overload that
958 accepts poly_uint64 offsets.
959 * expmed.h (extract_bit_field): Take bitsize and bitnum as
960 poly_uint64s rather than unsigned HOST_WIDE_INTs.
961 * expmed.c (lowpart_bit_field_p): Likewise.
962 (extract_bit_field_as_subreg): New function, split out from...
963 (extract_bit_field_1): ...here. Take bitsize and bitnum as
964 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
965 extractions, check that BITSIZE matches the size of the extracted
966 value and that BITNUM is an exact multiple of that size.
967 If all else fails, try forcing the value into memory if
968 BITNUM is variable, and adjusting the address so that the
969 offset is constant. Split the part that can only handle constant
970 bitsize and bitnum out into...
971 (extract_integral_bit_field): ...this new function.
972 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
973 rather than unsigned HOST_WIDE_INTs.
975 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
976 Alan Hayward <alan.hayward@arm.com>
977 David Sherwood <david.sherwood@arm.com>
979 * expmed.h (store_bit_field): Take bitsize and bitnum as
980 poly_uint64s rather than unsigned HOST_WIDE_INTs.
981 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
982 that returns the byte size.
983 (store_bit_field_1): Take bitsize and bitnum as
984 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
985 to simple_mem_bitfield_p. Split the part that can only handle
986 constant bitsize and bitnum out into...
987 (store_integral_bit_field): ...this new function.
988 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
989 than unsigned HOST_WIDE_INTs.
990 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
992 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
993 Alan Hayward <alan.hayward@arm.com>
994 David Sherwood <david.sherwood@arm.com>
996 * lra-int.h (lra_reg): Change offset from int to poly_int64.
997 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
999 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1000 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1001 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1002 offset as a poly_int64 rather than an int.
1003 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1004 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1005 * lra-constraints.c (equiv_address_substitution): Track offsets
1007 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1008 (curr_insn_transform): Handle the new form of sp_offset.
1009 * lra-eliminations.c (lra_elim_table): Change previous_offset
1010 and offset from HOST_WIDE_INT to poly_int64.
1011 (print_elim_table, update_reg_eliminate): Update accordingly.
1012 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1013 (get_elimination): Update accordingly.
1014 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1015 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1016 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1017 poly_int64 offsets generally.
1018 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1019 (mark_not_eliminable, init_elimination): Update accordingly.
1020 (remove_reg_equal_offset_note): Return a bool and pass the new
1021 offset back by pointer as a poly_int64.
1022 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1023 rather than a HOST_WIDE_INT.
1024 (do_remat): Track offsets poly_int64s.
1025 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1027 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1028 Alan Hayward <alan.hayward@arm.com>
1029 David Sherwood <david.sherwood@arm.com>
1031 * rtl.h (mem_attrs): Add a default constructor. Change size and
1032 offset from HOST_WIDE_INT to poly_int64.
1033 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1034 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1035 (widen_memory_access): Take the sizes and offsets as poly_int64s
1036 rather than HOST_WIDE_INTs.
1037 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1038 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1040 (adjust_offset_for_component_ref): Change the offset from a
1041 HOST_WIDE_INT to a poly_int64.
1042 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1043 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1044 * dce.c (find_call_stack_args): Likewise.
1045 * dse.c (record_store): Likewise.
1046 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1047 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1048 * read-rtl-function.c (test_loading_mem): Likewise.
1049 * rtlanal.c (may_trap_p_1): Likewise.
1050 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1051 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1052 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1053 (mem_attrs::mem_attrs): New function.
1054 (set_mem_attributes_minus_bitpos): Change bitpos from a
1055 HOST_WIDE_INT to poly_int64.
1056 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1057 (clear_mem_offset, clear_mem_size, change_address)
1058 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1059 initialize mem_attrs.
1060 (set_mem_offset, set_mem_size, adjust_address_1)
1061 (adjust_automodify_address_1, offset_address, widen_memory_access):
1062 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1064 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1065 Alan Hayward <alan.hayward@arm.com>
1066 David Sherwood <david.sherwood@arm.com>
1068 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1069 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1070 rather than 0 to represent an unknown size. Assert that the size
1071 is known when the mode isn't BLKmode.
1072 (may_trap_p_1): Use -1 for unknown sizes.
1073 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1075 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1080 to poly_int64. Update commentary for positions_needed.large.
1081 (read_info_type): Change offset and width from HOST_WIDE_INT
1083 (set_usage_bits): Likewise.
1084 (canon_address): Return the offset as a poly_int64 rather than
1085 a HOST_WIDE_INT. Use strip_offset_and_add.
1086 (set_all_positions_unneeded, any_positions_needed_p): Use
1087 positions_needed.large to track stores with non-constant widths.
1088 (all_positions_needed_p): Likewise. Take the offset and width
1089 as poly_int64s rather than ints. Assert that rhs is nonnull.
1090 (record_store): Cope with non-constant offsets and widths.
1091 Nullify the rhs of an earlier store if we can't tell which bytes
1093 (find_shift_sequence): Take the access_size and shift as poly_int64s
1095 (get_stored_val): Take the read_offset and read_width as poly_int64s
1096 rather than HOST_WIDE_INTs.
1097 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1098 non-constant offsets and widths.
1100 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1101 Alan Hayward <alan.hayward@arm.com>
1102 David Sherwood <david.sherwood@arm.com>
1104 * inchash.h (inchash::hash::add_poly_int): New function.
1105 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1106 Use poly_int64 rather than HOST_WIDE_INT.
1107 (ao_ref::max_size_known_p): New function.
1108 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1109 rather than HOST_WIDE_INT.
1110 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1111 to temporaries until its interface is adjusted to match.
1112 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1113 (aliasing_component_refs_p, decl_refs_may_alias_p)
1114 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1115 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1116 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1118 * alias.c (ao_ref_from_mem): Likewise.
1119 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1120 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1121 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1122 (maybe_trim_complex_store, maybe_trim_constructor_store)
1123 (live_bytes_read, dse_classify_store): Likewise.
1124 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1125 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1126 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1127 (vn_reference_lookup_3): Likewise.
1128 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1130 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1131 Alan Hayward <alan.hayward@arm.com>
1132 David Sherwood <david.sherwood@arm.com>
1134 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1135 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1136 instead of ranges_overlap_p.
1138 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1139 Alan Hayward <alan.hayward@arm.com>
1140 David Sherwood <david.sherwood@arm.com>
1142 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1143 sizes as poly_int64s rather than HOST_WIDE_INTs.
1145 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1146 Alan Hayward <alan.hayward@arm.com>
1147 David Sherwood <david.sherwood@arm.com>
1149 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1150 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1151 * gimple-fold.c (fold_ctor_reference): Likewise.
1153 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1154 Alan Hayward <alan.hayward@arm.com>
1155 David Sherwood <david.sherwood@arm.com>
1157 * target.def (dwarf_poly_indeterminate_value): New hook.
1158 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1159 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1160 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1161 * doc/tm.texi: Regenerate.
1162 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1163 offset as a poly_int64.
1164 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1165 offset as a poly_int64.
1166 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1167 Take the offset as a poly_int64.
1168 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1169 (frame_pointer_fb_offset): Change to a poly_int64.
1170 (int_loc_descriptor): Take the offset as a poly_int64. Use
1171 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1172 (based_loc_descr): Take the offset as a poly_int64.
1173 Use strip_offset_and_add to handle (plus X (const)).
1174 Use new_reg_loc_descr instead of an open-coded version of the
1175 previous implementation.
1176 (mem_loc_descriptor): Handle CONST_POLY_INT.
1177 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1178 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1180 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1181 Alan Hayward <alan.hayward@arm.com>
1182 David Sherwood <david.sherwood@arm.com>
1184 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1185 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1186 * inchash.h (inchash::hash::add_poly_hwi): New function.
1187 * gengtype.c (main): Register poly_int64.
1188 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1189 offset as a poly_int.
1190 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1191 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1192 offset as a poly_int64.
1193 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1194 * print-rtl.c (print_poly_int): New function.
1195 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1197 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1199 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1200 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1201 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1202 Enforce track_offset_p for parts of a PARALLEL.
1203 (vt_add_function_parameter): Use const_offset for the final
1204 offset to track. Use get_tracked_reg_offset for the parts
1207 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1208 Alan Hayward <alan.hayward@arm.com>
1209 David Sherwood <david.sherwood@arm.com>
1211 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1212 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1213 * doc/tm.texi: Regenerate.
1214 * hooks.h (hook_bool_uint_uint_true): Delete.
1215 (hook_bool_puint64_puint64_true): Declare.
1216 * hooks.c (hook_bool_uint_uint_true): Delete.
1217 (hook_bool_puint64_puint64_true): New function.
1218 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1219 instead of unsigned ints.
1220 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1221 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1223 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * optabs.h (expand_operand): Add an int_value field.
1228 (create_expand_operand): Add an int_value parameter and use it
1229 to initialize the new expand_operand field.
1230 (create_integer_operand): Replace with a declaration of a function
1231 that accepts poly_int64s. Move the implementation to...
1232 * optabs.c (create_integer_operand): ...here.
1233 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1234 the mode preserves the value of int_value, instead of calling
1235 const_int_operand on the rtx. Use gen_int_mode to generate
1238 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * dumpfile.h (dump_dec): Declare.
1243 * dumpfile.c (dump_dec): New function.
1244 * pretty-print.h (pp_wide_integer): Turn into a function and
1245 declare a poly_int version.
1246 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1248 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1249 Alan Hayward <alan.hayward@arm.com>
1250 David Sherwood <david.sherwood@arm.com>
1252 * doc/generic.texi (POLY_INT_CST): Document.
1253 * tree.def (POLY_INT_CST): New tree code.
1254 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1255 * tree-core.h (tree_poly_int_cst): New struct.
1256 (tree_node): Add a poly_int_cst field.
1257 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1258 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1259 instead of a wide_int_ref.
1260 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1262 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1263 instead of an unsigned HOST_WIDE_INT.
1264 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1265 (ptrdiff_tree_p): Declare.
1266 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1267 extern inline implementations if the target doesn't use POLY_INT_CST.
1268 (poly_int_tree_p): New function.
1269 (wi::unextended_tree): New class.
1270 (wi::int_traits <unextended_tree>): New override.
1271 (wi::extended_tree): Add a default constructor.
1272 (wi::extended_tree::get_tree): New function.
1273 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1274 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1275 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1276 (wi::tree_to_poly_wide_ref): New typedefs.
1277 (wi::ints_for): Provide overloads for extended_tree and
1279 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1280 (wi::to_wide): New functions.
1281 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1282 * tree.c (poly_int_cst_hasher): New struct.
1283 (poly_int_cst_hash_table): New variable.
1284 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1285 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1287 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1288 (init_ttree): Initialize poly_int_cst_hash_table.
1289 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1290 a poly_int64 instead of a HOST_WIDE_INT.
1291 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1292 instead of an unsigned HOST_WIDE_INT.
1293 (wide_int_to_tree): Rename to...
1294 (wide_int_to_tree_1): ...this.
1295 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1296 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1297 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1298 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1299 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1300 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1302 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1303 (streamer_read_tree_body): Likewise.
1304 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1305 (streamer_write_tree_body): Likewise.
1306 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1307 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1308 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1309 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1310 * gimple-expr.h (is_gimple_constant): Likewise.
1311 * gimplify.c (maybe_with_size_expr): Likewise.
1312 * print-tree.c (print_node): Likewise.
1313 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1314 * tree-pretty-print.c (dump_generic_node): Likewise.
1315 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1316 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1317 * tree-vrp.c (compare_values_warnv): Likewise.
1318 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1319 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1320 (force_expr_to_var_cost): Likewise.
1321 * tree-ssa-loop.c (for_each_index): Likewise.
1322 * fold-const.h (build_invariant_address, size_int_kind): Take a
1323 poly_int64 instead of a HOST_WIDE_INT.
1324 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1325 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1327 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1328 (int_const_binop_2): New function, split out from...
1329 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1330 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1331 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1332 * gimple-ssa-strength-reduction.c (slsr_process_add)
1333 (slsr_process_mul): Check for INTEGER_CSTs before using them
1335 * stor-layout.c (bits_from_bytes): New function.
1336 (bit_from_pos): Use it.
1337 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1338 by BITS_PER_UNIT to get the TYPE_SIZE.
1339 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1340 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1342 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1343 Alan Hayward <alan.hayward@arm.com>
1344 David Sherwood <david.sherwood@arm.com>
1346 * doc/rtl.texi (const_poly_int): Document. Also document the
1347 rtl sharing behavior.
1348 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1349 * rtl.h (const_poly_int_def): New struct.
1350 (rtx_def::u): Add a cpi field.
1351 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1352 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1353 (wi::rtx_to_poly_wide_ref): New typedef
1354 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1355 (poly_int_rtx_p): New functions.
1356 (trunc_int_for_mode): Declare a poly_int64 version.
1357 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1358 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1360 (strip_offset): Declare.
1361 (strip_offset_and_add): New function.
1362 * rtl.def (CONST_POLY_INT): New rtx code.
1363 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1364 (shared_const_p): Use poly_int_rtx_p.
1365 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1367 (gen_int_shift_amount): Likewise.
1368 * emit-rtl.c (const_poly_int_hasher): New class.
1369 (const_poly_int_htab): New variable.
1370 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1371 (const_poly_int_hasher::hash): New function.
1372 (const_poly_int_hasher::equal): Likewise.
1373 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1374 (immed_wide_int_const): Rename to...
1375 (immed_wide_int_const_1): ...this and make static.
1376 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1377 instead of a wide_int_ref.
1378 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1379 (gen_lowpart_common): Handle CONST_POLY_INT.
1380 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1381 * cselib.c (cselib_hash_rtx): Likewise.
1382 * dwarf2out.c (const_ok_for_output_1): Likewise.
1383 * expr.c (convert_modes): Likewise.
1384 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1385 * rtlhash.c (add_rtx): Likewise.
1386 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1387 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1388 Handle existing CONST_POLY_INT rtxes.
1389 * expmed.h (expand_shift): Take a poly_int64 instead of a
1391 * expmed.c (expand_shift): Likewise.
1392 * rtlanal.c (strip_offset): New function.
1393 (commutative_operand_precedence): Give CONST_POLY_INT the same
1394 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1396 * rtl-tests.c (const_poly_int_tests): New struct.
1397 (rtl_tests_c_tests): Use it.
1398 * simplify-rtx.c (simplify_const_unary_operation): Handle
1400 (simplify_const_binary_operation): Likewise.
1401 (simplify_binary_operation_1): Fold additions of symbolic constants
1402 and CONST_POLY_INTs.
1403 (simplify_subreg): Handle extensions and truncations of
1405 (simplify_const_poly_int_tests): New struct.
1406 (simplify_rtx_c_tests): Use it.
1407 * wide-int.h (storage_ref): Add default constructor.
1408 (wide_int_ref_storage): Likewise.
1409 (trailing_wide_ints): Use GTY((user)).
1410 (trailing_wide_ints::operator[]): Add a const version.
1411 (trailing_wide_ints::get_precision): New function.
1412 (trailing_wide_ints::extra_size): Likewise.
1414 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1415 Alan Hayward <alan.hayward@arm.com>
1416 David Sherwood <david.sherwood@arm.com>
1418 * emit-rtl.h (gen_int_shift_amount): Declare.
1419 * emit-rtl.c (gen_int_shift_amount): New function.
1420 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1422 * calls.c (shift_return_value): Likewise.
1423 * cse.c (fold_rtx): Likewise.
1424 * dse.c (find_shift_sequence): Likewise.
1425 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1426 (expand_shift, expand_smod_pow2): Likewise.
1427 * lower-subreg.c (shift_cost): Likewise.
1428 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1429 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1430 (expand_vec_perm_var): Likewise.
1431 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1432 (simplify_binary_operation_1): Likewise.
1433 * combine.c (try_combine, find_split_point, force_int_to_mode)
1434 (simplify_shift_const_1, simplify_shift_const): Likewise.
1435 (change_zero_ext): Likewise. Use simplify_gen_binary.
1437 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1439 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1441 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1443 * doc/invoke.texi (ARM Options): Document accepted extension options
1444 for -march=armv8.3-a.
1446 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1449 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1450 or --with-float=softfp, set the default CPU to arm10e.
1452 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1454 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1455 * config/visium/predicates.md (const_shift_operand): Likewise.
1456 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1457 (visium_legitimize_reload_address): Likewise.
1459 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1462 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1464 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1467 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1469 2017-12-20 Julia Koval <julia.koval@intel.com>
1471 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1472 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1473 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1474 * config.gcc: Include vpclmulqdqintrin.h.
1475 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1476 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1478 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1479 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1480 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1481 * config/i386/i386.c (isa2_opts): Add -mcx16.
1482 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1483 (ix86_option_override_internal): Move mcx16 to flags2.
1484 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1485 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1486 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1487 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1488 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1489 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1490 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1491 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1492 * doc/invoke.texi: Add -mvpclmulqdq.
1494 2017-12-20 Tom de Vries <tom@codesourcery.com>
1497 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1498 * calls.c (rtx_for_static_chain): ... here. New function.
1499 * calls.h (rtx_for_static_chain): Declare.
1500 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1501 instead of targetm.calls.static_chain.
1502 * df-scan.c (df_get_entry_block_def_set): Same.
1504 2017-12-19 Tom de Vries <tom@codesourcery.com>
1506 PR tree-optimization/83493
1507 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1510 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1512 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1513 inform with hardcoded english plural handling.
1515 2017-12-18 Jeff Law <law@redhat.com>
1517 PR tree-optimization/83477
1518 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1519 a non-virtual PHI, always push a new range.
1521 2017-12-19 Martin Sebor <msebor@redhat.com>
1524 * builtins.c (compute_objsize): Handle non-constant offsets.
1526 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1528 PR tree-optimization/83444
1529 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1530 character load case, if get_stridx on MEM_REF's operand doesn't
1531 look usable, retry with get_addr_stridx.
1533 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1536 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1537 (variable_tracking_main_1): Keep markers even when VTA fails.
1540 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1541 even if there are markers before it.
1542 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1544 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1546 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1547 typical order conditions.
1548 * sel-sched.c (extract_new_fences_from): Likewise.
1549 * config/visium/constraints.md (J, K, L): Likewise.
1550 * config/visium/predicates.md (const_shift_operand): Likewise.
1551 * config/visium/visium.c (visium_legitimize_address,
1552 visium_legitimize_reload_address): Likewise.
1553 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1554 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1555 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1556 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1557 SET_DUMP_DETAIL): Likewise.
1558 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1559 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1560 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1561 avr_set_core_architecture, avr_set_current_function,
1562 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1563 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1564 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1565 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1566 avr_map_decompose, avr_fold_builtin): Likewise.
1567 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1568 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1569 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1570 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1571 * config/m32c/m32c.c (m32c_conditional_register_usage,
1572 m32c_address_cost): Likewise.
1573 * config/m32c/predicates.md (shiftcount_operand,
1574 longshiftcount_operand): Likewise.
1575 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1576 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1577 can_use_cdx_ldstw): Likewise.
1578 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1579 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1581 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1582 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1583 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1584 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1585 * config/vax/vax.c (adjacent_operands_p): Likewise.
1586 * config/ft32/constraints.md (L, b, KA): Likewise.
1587 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1589 * cfgexpand.c (expand_stack_alignment): Likewise.
1590 * gcse.c (insert_expr_in_table): Likewise.
1591 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1592 * cgraphunit.c (cgraph_node::expand): Likewise.
1593 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1594 * emit-rtl.c (add_insn): Likewise.
1595 * input.c (dump_location_info): Likewise.
1596 * passes.c (NEXT_PASS): Likewise.
1597 * read-rtl-function.c (parse_note_insn_name,
1598 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1600 * sched-rgn.c (sched_rgn_init): Likewise.
1601 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1602 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1603 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1604 * explow.c (eliminate_constant_term): Likewise.
1605 * final.c (leaf_renumber_regs_insn): Likewise.
1606 * cfgrtl.c (print_rtl_with_bb): Likewise.
1607 * genhooks.c (emit_init_macros): Likewise.
1608 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1609 * tree-data-ref.c (conflict_fn): Likewise.
1610 * selftest.c (assert_streq): Likewise.
1611 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1612 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1613 fold_binary_loc, multiple_of_p): Likewise.
1614 * reload.c (push_reload, find_equiv_reg): Likewise.
1615 * et-forest.c (et_nca, et_below): Likewise.
1616 * dbxout.c (dbxout_symbol_location): Likewise.
1617 * reorg.c (relax_delay_slots): Likewise.
1618 * dojump.c (do_compare_rtx_and_jump): Likewise.
1619 * gengtype-parse.c (type): Likewise.
1620 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1621 simplify_const_relational_operation): Likewise.
1622 * reload1.c (do_output_reload): Likewise.
1623 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1624 * gengtype.c (type_for_name): Likewise.
1625 * gimple-ssa-sprintf.c (format_directive): Likewise.
1627 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1630 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1631 accessing it. Adjust comment.
1633 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1636 * predict.c (zero_one_minusone): New function.
1637 (apply_return_prediction): Avoid return prediction for functions
1638 returning only -1, 0 and 1 values, unless they only return -1 and 0
1641 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1643 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1644 fall-through warning.
1645 (arc_compute_frame_size): Remove unused variables.
1646 (arc_print_operand): Fix fprintif format.
1647 (arc_can_follow_jump): Clean fall-through warning.
1649 2017-12-19 Marek Polacek <polacek@redhat.com>
1652 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1655 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1657 * config/arc/arc.c (overriderregs): New variable.
1658 (arc_override_options): Track fixed/call saved/call options.
1659 (arc_conditional_register_usage): Check against overrideregs
1660 variable whenever we change register properties.
1662 2017-12-19 Nathan Sidwell <nathan@acm.org>
1664 * opts.c (finish_options): Don't prefix dump_base_name if it
1665 already contains directories.
1667 2017-12-19 Martin Liska <mliska@suse.cz>
1669 PR rtl-optimization/82675
1670 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1671 more element in sbitmap.
1673 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1675 * gimplify.c (gimplify_expr): Use error_operand_p.
1677 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1680 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1683 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1685 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1686 the jump is to a label.
1688 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1690 PR tree-optimization/83444
1691 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1696 * ipa-inline.c (flatten_remove_node_hook): New function.
1697 (ipa_inline): Keep only nodes with flatten attribute at the end of
1698 the array in the order from ipa_reverse_postorder, only walk that
1699 portion of array for flattening, if there is more than one such
1700 node, temporarily register a removal hook and ignore removed nodes.
1702 PR tree-optimization/80631
1703 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1704 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1705 IFN_REDUC_MAX or IFN_REDUC_MIN.
1707 2017-12-18 Jeff Law <law@redhat.com>
1709 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1710 record symbolic equivalences from backedges in the CFG.
1713 2017-11-19 Jeff Law <law@redhat.com>
1715 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1716 of degenerates resulting from ignoring an edge.
1718 2017-12-18 Martin Sebor <msebor@redhat.com>
1721 PR tree-optimization/78450
1722 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1723 (handle_builtin_strlen): Call it.
1725 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1727 PR rtl-optimization/83424
1728 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1730 2017-12-18 Marek Polacek <polacek@redhat.com>
1733 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1734 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1737 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1739 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1740 (umaddsidi4, umaddsidi_split): Likewise.
1742 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1744 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1747 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1750 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1752 2017-12-18 Richard Biener <rguenther@suse.de>
1754 PR tree-optimization/81877
1755 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1756 (outermost_indep_loop): Adjust.
1757 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1758 (can_sm_ref_p): Adjust.
1760 2017-12-18 Richard Biener <rguenther@suse.de>
1763 * tree.c (array_at_struct_end_p): Return true if the underlying
1764 object has space for at least one element in excess of what
1765 the array domain specifies.
1767 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1769 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1770 example to avoid overfull hbox.
1771 * doc/invoke.texi (Option Summary): Add missing @gol.
1772 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1775 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1776 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1778 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1779 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1780 (AArch64 Options): Document -mverbose-cost-dump.
1781 (ARM Options): Likewise, plus -mflip-thumb.
1783 2017-12-17 Martin Sebor <msebor@redhat.com>
1786 * gimple-ssa-warn-restrict.c
1787 (builtin_memref::offset_out_of_bounds): Correct the handling of
1790 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1792 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1794 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1796 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1798 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1800 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1803 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1805 * doc/invoke.texi: Fix some typos.
1807 2017-12-16 Martin Sebor <msebor@redhat.com>
1809 PR tree-optimization/78918
1810 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1811 * builtins.c (check_sizes): Rename...
1812 (check_access): ...to this. Rename function arguments for clarity.
1813 (check_memop_sizes): Adjust names.
1814 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1815 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1816 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1817 (check_strncat_sizes, expand_builtin_strncat): Same.
1818 (expand_builtin_strncpy, expand_builtin_memset): Same.
1819 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1820 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1821 (maybe_emit_sprintf_chk_warning): Same.
1822 (expand_builtin_strcpy): Adjust.
1823 (expand_builtin_stpcpy): Same.
1824 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1825 in pointer-checking forms of memcpy, memmove, and mempcpy.
1826 (gcall_to_tree_minimal, max_object_size): Define new functions.
1827 * builtins.h (max_object_size): Declare.
1828 * calls.c (alloc_max_size): Call max_object_size instead of
1829 hardcoding ssizetype limit.
1830 (get_size_range): Handle new argument.
1831 * calls.h (get_size_range): Add a new argument.
1832 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1833 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1834 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1836 (gimple_fold_builtin_memory_chk): Same.
1837 (gimple_fold_builtin_stxcpy_chk): New function.
1838 * gimple-ssa-warn-restrict.c: New source.
1839 * gimple-ssa-warn-restrict.h: New header.
1840 * gimple.c (gimple_build_call_from_tree): Propagate location.
1841 * passes.def (pass_warn_restrict): Add new pass.
1842 * tree-pass.h (make_pass_warn_restrict): Declare.
1843 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1845 (handle_builtin_strcat): Same.
1846 (strlen_optimize_stmt): Rename...
1847 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1848 stpncpy, strncpy, and their checking forms.
1850 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1852 PR rtl-optimization/82849
1853 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1854 and get_max_loop_iterations_int.
1856 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1857 Alan Hayward <alan.hayward@arm.com>
1858 David Sherwood <david.sherwood@arm.com>
1860 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1861 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1862 size as a poly_uint64.
1863 (mode_for_vector, mode_for_int_vector): Take the number of vector
1864 elements as a poly_uint64.
1865 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1866 size as a poly_uint64.
1867 (mode_for_vector, mode_for_int_vector): Take the number of vector
1868 elements as a poly_uint64.
1870 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1871 Alan Hayward <alan.hayward@arm.com>
1872 David Sherwood <david.sherwood@arm.com>
1874 * machmode.h (MACRO_MODE): New macro.
1875 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1876 * targhooks.c (default_libcall_value, default_secondary_reload)
1877 (default_memory_move_cost, default_register_move_cost)
1878 (default_class_max_nregs): Likewise.
1880 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1881 Alan Hayward <alan.hayward@arm.com>
1882 David Sherwood <david.sherwood@arm.com>
1884 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1885 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1887 * genautomata.c (main): Likewise.
1888 * genconditions.c (write_header): Likewise.
1889 * genemit.c (main): Likewise.
1890 * genextract.c (print_header): Likewise.
1891 * genopinit.c (main): Likewise.
1892 * genoutput.c (output_prologue): Likewise.
1893 * genpeep.c (main): Likewise.
1894 * genpreds.c (write_insn_preds_c): Likewise.
1895 * genrecog.c (writer_header): Likewise.
1896 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1897 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1898 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1899 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1900 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1901 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1902 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1903 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1904 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1905 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1906 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1907 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1908 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1909 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1910 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1911 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1912 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1913 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1914 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1915 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1916 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1917 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1918 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1919 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1920 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1921 * config/darwin.c (IN_TARGET_CODE): Likewise.
1922 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1923 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1924 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1925 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1926 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1927 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1928 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1929 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1930 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1931 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1932 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1933 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1934 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1935 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1936 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1937 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1938 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1939 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1940 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1941 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1942 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1943 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1944 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1945 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1946 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1947 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1948 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1949 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1950 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1951 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1952 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1953 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1954 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1955 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1956 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1957 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1958 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1959 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1960 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1961 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1962 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1963 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1964 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1965 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1966 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1967 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1968 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1969 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1970 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1971 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1972 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1973 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1974 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1975 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1976 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1977 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1978 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1979 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1980 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1981 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1982 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1983 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1984 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1985 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1986 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1987 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1988 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1989 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1990 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1991 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1992 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1993 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1994 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1995 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1996 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1997 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1998 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1999 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2000 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2001 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2002 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2003 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2004 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2005 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2006 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2007 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2008 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2009 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2010 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2011 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2012 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2013 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2014 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2015 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2016 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2017 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2018 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2019 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2020 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2021 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2022 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2023 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2024 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2025 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2026 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2027 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2029 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2030 Alan Hayward <alan.hayward@arm.com>
2031 David Sherwood <david.sherwood@arm.com>
2033 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2036 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2037 Alan Hayward <alan.hayward@arm.com>
2038 David Sherwood <david.sherwood@arm.com>
2040 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2041 * doc/md.texi (vec_series@var{m}): Document.
2042 * tree.def (VEC_SERIES_EXPR): New tree code.
2043 * tree.h (build_vec_series): Declare.
2044 * tree.c (build_vec_series): New function.
2045 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2046 * tree-pretty-print.c (dump_generic_node): Likewise.
2047 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2048 * tree-inline.c (estimate_operator_cost): Likewise.
2049 * expr.c (expand_expr_real_2): Likewise.
2050 * optabs-tree.c (optab_for_tree_code): Likewise.
2051 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2052 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2053 * expmed.c (make_tree): Handle VEC_SERIES.
2054 * optabs.def (vec_series_optab): New optab.
2055 * optabs.h (expand_vec_series_expr): Declare.
2056 * optabs.c (expand_vec_series_expr): New function.
2057 * tree-vect-generic.c (expand_vector_operations_1): Check that
2058 the operands also have vector type.
2060 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2061 Alan Hayward <alan.hawyard@arm.com>
2062 David Sherwood <david.sherwood@arm.com>
2064 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2065 (VEC_COND_EXPR): Add missing @tindex.
2066 * doc/md.texi (vec_duplicate@var{m}): Document.
2067 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2068 * tree.c (build_vector_from_val): Add stubbed-out handling of
2069 variable-length vectors, using VEC_DUPLICATE_EXPR.
2070 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2071 * cfgexpand.c (expand_debug_expr): Likewise.
2072 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2073 * tree-inline.c (estimate_operator_cost): Likewise.
2074 * tree-pretty-print.c (dump_generic_node): Likewise.
2075 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2076 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2077 (test_vec_duplicate_folding): New function.
2078 (fold_const_c_tests): Call it.
2079 * optabs.def (vec_duplicate_optab): New optab.
2080 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2081 * optabs.h (expand_vector_broadcast): Declare.
2082 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2083 vec_duplicate_optab.
2084 * expr.c (store_constructor): Try using vec_duplicate_optab for
2086 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2088 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2091 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2092 div/mod latencies a bit.
2094 2017-12-15 Jeff Law <law@redhat.com>
2096 PR tree-optimization/36550
2097 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2098 (mark_threaded_blocks): Rewrite code to avoid block copying when
2099 optimizing for size. Don't pessimize blocks which will be
2100 copied, but all the statements will be dead.
2102 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2104 PR tree-optimization/81165
2105 * tree-ssa-threadupdate.c (uses_in_bb): New.
2106 (estimate_threading_killed_stmts): New.
2107 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2108 * tree-ssa-threadedge.c
2109 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2112 2017-12-15 Jeff Law <law@redhat.com>
2114 PR tree-optimization/83410
2115 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2116 threads when parallelizing loops.
2118 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2120 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2122 * config/alpha/alpha.c (vms_attribute_table): Swap
2123 affects_type_identity and handler fields, adjust comments.
2124 * config/mips/mips.c (mips_attribute_table): Likewise.
2125 * config/visium/visium.c (visium_attribute_table): Likewise.
2126 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2127 * config/microblaze/microblaze.c (microblaze_attribute_table):
2129 * config/spu/spu.c (spu_attribute_table): Likewise.
2130 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2131 * config/arc/arc.c (arc_attribute_table): Likewise.
2132 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2133 * config/v850/v850.c (v850_handle_interrupt_attribute,
2134 v850_handle_data_area_attribute): Formatting fixes.
2135 (v850_attribute_table): Swap affects_type_identity and handler
2136 fields, adjust comments.
2137 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2138 * config/arm/arm.c (arm_attribute_table): Likewise.
2139 * config/avr/avr.c (avr_attribute_table): Likewise.
2140 * config/s390/s390.c (s390_attribute_table): Likewise.
2141 * config/sh/sh.c (sh_attribute_table): Likewise.
2142 * config/i386/i386.c (ix86_handle_cconv_attribute,
2143 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2144 (ix86_attribute_table): Swap affects_type_identity and handler
2145 fields, adjust comments.
2146 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2147 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2148 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2149 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2150 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2151 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2152 * config/rx/rx.c (rx_attribute_table): Likewise.
2153 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2154 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2155 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2156 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2157 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2158 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2159 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2160 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2161 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2162 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2163 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2165 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2167 (register_scoped_attributes, decl_attributes): Formatting fixes.
2169 PR tree-optimization/83269
2170 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2171 subtraction in arg0's type if type is signed and arg0 is unsigned.
2175 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2176 convert? on @0 instead of convert. Check type of @1, not @0.
2177 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2178 convert? on @0 instead of convert. Check type of @1, not @0.
2179 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2180 only optimize if either both @1 and @2 types are narrower
2181 precision, or both are wider or equal precision, and in the former
2182 case only if both have undefined overflow.
2184 2017-12-15 Richard Biener <rguenther@suse.de>
2187 * internal-fn.def (IFN_NOP): Add.
2188 * internal-fn.c (expand_NOP): Do nothing.
2189 * lto-streamer-in.c (input_function): Instead of removing
2190 sanitizer calls replace them with IFN_NOP calls.
2192 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * dse.c (store_info, read_info_type): Replace begin and end with
2198 (print_range): New function.
2199 (set_all_positions_unneeded, any_positions_needed_p)
2200 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2202 (record_store): Likewise. Optimize the case in which all positions
2204 (get_stored_val): Replace read_begin and read_end with read_offset
2206 (replace_read): Update call accordingly.
2208 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2210 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2211 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2212 (loop_cand::loop_cand): Initialize above members.
2213 (loop_cand::supported_operations): Delete.
2214 (loop_cand::can_interchange_p): Inline above function.
2215 (loop_cand::classify_simple_reduction): Record number of constant
2216 initialized simple reductions.
2217 (should_interchange_loops): New parameters. Check stmt cost of loops
2219 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2220 Update call to should_interchange_loops.
2221 (should_interchange_loop_nest): Update call to
2222 should_interchange_loops.
2224 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2227 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2228 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2229 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2231 2017-12-15 Julia Koval <julia.koval@intel.com>
2233 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2234 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2235 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2236 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2237 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2239 2017-12-15 Julia Koval <julia.koval@intel.com>
2241 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2242 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2243 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2244 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2245 _mm_aesenc_epi128): New intrinsics.
2247 2017-12-15 Julia Koval <julia.koval@intel.com>
2249 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2250 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2251 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2252 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2253 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2255 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2257 * gimple-ssa-strength-reduction.c (analyze_increments):
2258 Distinguish replacement costs for constant strides from those for
2261 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2263 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2265 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2267 * doc/invoke.texi: Document -Wcast-function-type.
2268 * recog.h (stored_funcptr): Change signature.
2269 * tree-dump.c (dump_node): Avoid warning.
2270 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2272 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2275 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2278 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2280 PR tree-optimization/83312
2281 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2282 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2283 "val" param, and to cope with arbitrary basic blocks.
2284 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2285 handle NULL_TREE for "val", dropping "bb" param.
2286 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2287 drop "bb" param. Handle NULL_TREE for "val".
2288 (find_case_label_for_value): Make "switch_stmt" param const.
2289 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2291 (vrp_prop::check_all_array_refs): Reimplement as...
2292 (check_array_bounds_dom_walker::before_dom_children): ...this new
2293 vfunc. Replace linear search through BB block list, excluding
2294 those with non-executable in-edges via dominator walk.
2296 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2298 * config/arm/arm.opt (mverbose-cost-dump): New option.
2299 * config/arm/arm.c (arm_rtx_costs): Use it.
2301 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2304 * reload1.c (emit_input_reload_insns): Skip debug markers.
2306 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2308 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2312 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2313 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2314 * config/sh/sh.c: Skip debug insns besides notes.
2315 * config/sh/sh.md: Likewise.
2316 * config/sh/sh_treg_combine.cc: Likewise.
2317 * config/sh/sync.md: Likewise.
2319 2017-12-14 Tom de Vries <tom@codesourcery.com>
2321 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2322 Add item for weak_undefined.
2324 2017-12-14 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/67842
2327 * sese.h (bb_in_region): Remove #if 0'ed code.
2329 2017-12-14 Richard Biener <rguenther@suse.de>
2331 PR tree-optimization/83326
2332 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2333 may_be_zero parameter and handle it by not marking the first
2334 peeled copy as not exiting the loop.
2335 (try_peel_loop): Likewise.
2336 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2337 to handle the case of constant or zero iterations and perform
2338 loop header copying on-the-fly.
2340 2017-12-14 Richard Biener <rguenther@suse.de>
2342 PR tree-optimization/83418
2343 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2344 Instead of asserting we don't get unfolded comparisons deal with
2347 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2350 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2351 basic blocks. Assert debug bind insns don't appear outside of bbs,
2352 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2355 PR tree-optimization/83198
2356 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2357 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2358 value if arg is a REAL_CST with incompatible type.
2360 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2361 Bin Cheng <bin.cheng@arm.com>
2364 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2366 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2369 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2371 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2372 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2374 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2375 Alan Hayward <alan.hayward@arm.com>
2376 David Sherwood <david.sherwood@arm.com>
2378 * poly-int.h: New file.
2379 * poly-int-types.h: Likewise.
2380 * coretypes.h: Include them.
2381 (POLY_INT_CONVERSION): Define.
2382 * target.def (estimated_poly_value): New hook.
2383 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2384 * doc/tm.texi: Regenerate.
2385 * doc/poly-int.texi: New file.
2386 * doc/gccint.texi: Include it.
2387 * doc/rtl.texi: Describe restrictions on subreg modes.
2388 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2389 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2390 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2391 * targhooks.h (default_estimated_poly_value): Declare.
2392 * targhooks.c (default_estimated_poly_value): New function.
2393 * target.h (estimated_poly_value): Likewise.
2394 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2395 (wi::unary_traits): Delete.
2396 (wi::binary_traits::signed_shift_result_type): Define for
2397 offset_int << HOST_WIDE_INT, etc.
2398 (generic_wide_int::operator <<=): Define for all types and use
2399 wi::lshift instead of <<.
2400 (wi::hwi_with_prec): Add a default constructor.
2401 (wi::ints_for): New class.
2402 (operator <<): Define for all wide-int types.
2403 (operator /): New function.
2404 (operator %): Likewise.
2405 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2406 (ASSERT_MAYBE_NE_AT): New macros.
2408 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2409 Dominik Vogt <vogt@linux.vnet.ibm.com>
2412 * emit-rtl.c (init_emit): Remove ??? comment.
2413 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2414 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2415 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2416 alignment of 3 virtual registers to BITS_PER_WORD.
2418 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2420 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2422 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2423 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2425 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2428 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2430 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2432 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2433 with the exception of debug begin stmt markers appear before
2437 * final.c (rest_of_handle_final): Call variable_tracking_main only
2438 if !flag_var_tracking.
2440 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2441 Jakub Jelinek <jakub@redhat.com>
2445 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2446 labels debug stmts that can only appear after labels.
2448 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2450 PR rtl-optimization/82398
2451 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2452 EXPR_USEFULNESS in priority comparison.
2454 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2456 PR rtl-optimization/83393
2457 * combine.c (move_deaths): If reg_stat points to a too new insn in
2458 last_death, do not use it: find the proper insn instead.
2460 2017-12-12 Jeff Law <law@redhat.com>
2462 PR tree-optimization/83298
2463 PR tree-optimization/83362
2464 PR tree-optimization/83383
2465 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2466 push_value_range a public interface. Add new argument to
2467 record_ranges_from_stmt.
2468 * gimple-ssa-evrp-analyze.c
2469 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2470 Update comments. Handle recording temporary equivalences.
2471 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2472 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2473 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2474 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2475 gimple-ssa-evrp-analyze.h.
2476 (record_temporary_equivalences_from_phis): Add new argument. When
2477 the PHI arg is an SSA_NAME, set the result's range to the range
2479 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2480 from statements too.
2481 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2482 Pass it down to children as needed.
2483 (thread_outgoing_edges): Likewise.
2484 (thread_across_edge): Likewise. Push/pop range state as needed.
2485 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2487 2017-12-12 Julia Koval <julia.koval@intel.com>
2489 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2490 (PTA_CANNONLAKE): Remove PTA_CLWB.
2492 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2495 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2496 (vcondv2div2df): Likewise.
2497 (vconduv2dfv2di): Likewise.
2499 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2501 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2502 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2503 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2504 ATTR_MATHFN_FPROUNDING_ERRNO.
2506 2017-12-12 Richard Biener <rguenther@suse.de>
2508 PR tree-optimization/83385
2509 * graphite-scop-detection.c (get_order, order): Remove.
2510 (bb_to_rpo): New global.
2512 (build_scops): Sort pbbs in RPO order.
2514 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2516 * combine.c (simplify_set): Do not transform subregs to zero_extends
2517 if the destination mode is a vector mode.
2519 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2521 PR tree-optimization/83359
2522 * tree-cfg.h (fold_loop_internal_call): Declare.
2523 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2524 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2525 (find_loop_dist_alias): New function.
2526 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2527 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2530 PR tree-optimization/80631
2531 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2532 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2533 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2534 hardcoding zero as the value if COND_EXPR is never true. For
2535 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2536 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2537 hardcoding MAX_EXPR as the reduction operation.
2538 (is_nonwrapping_integer_induction): Allow negative step.
2539 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2540 vect_create_epilog_for_reduction, if no value is suitable, don't
2541 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2543 2017-12-12 Richard Biener <rguenther@suse.de>
2545 PR tree-optimization/81889
2546 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2547 range info from the non-wrapping IV instead of just the range
2550 2017-12-12 Julia Koval <julia.koval@intel.com>
2552 * config.gcc: Add vaesintrin.h.
2553 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2555 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2556 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2558 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2559 * config/i386/immintrin.h: Include vaesintrin.h.
2560 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2561 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2562 _mm_aesdec_epi128): New intrinsics.
2564 2017-12-12 Julia Koval <julia.koval@intel.com>
2566 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2567 OPTION_MASK_ISA_VAES_UNSET): New.
2568 (ix86_handle_option): Handle -mvaes.
2569 * config/i386/cpuid.h: Define bit_VAES.
2570 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2571 * config/i386/i386-c.c (__VAES__): New.
2572 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2573 (ix86_valid_target_attribute_inner_p): Ditto.
2574 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2575 * config/i386/i386.opt: Add -mvaes.
2576 * doc/invoke.texi: Ditto.
2578 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2580 * debug.h (gcc_debug_hooks): Add inline_entry.
2581 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2582 * debug.c (do_nothing_debug_hooks): Likewise.
2583 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2584 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2585 (dwarf2_lineno_debug_hooks): Likewise.
2587 * common.opt (gstatement-frontiers): New, setting
2588 debug_nonbind_markers_p.
2589 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2590 * toplev.c (process_options): Autodetect value for debug statement
2592 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2593 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2595 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2596 markers. Integrate source bind into debug stmt expand loop.
2597 (pass_expand::execute): Check debug marker limit. Avoid deep
2598 TER and expand debug locations for debug bind insns only.
2599 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2601 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2602 * final.c (reemit_insn_block_notes): Take current block from
2603 nonbind markers. Declare note where it's first set.
2604 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2605 begin stmt markers if enabled.
2606 (notice_source_line): Handle nonbind markers. Fail if their
2607 location is unknown or that of builtins.
2608 (rest_of_handle_final): Convert begin stmt markers to notes if
2609 var-tracking didn't run.
2610 (rest_of_clean_state): Skip begin stmt markers.
2611 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2613 * function.c (allocate_struct_function): Set begin_stmt_markers.
2614 * function.h (struct function): Add debug_marker_count counter
2615 and debug_nonbind_markers flag.
2616 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2617 * gimple-low.c (lower_function_body): Adjust
2618 debug_nonbind_markers.
2619 (lower_stmt): Drop or skip gimple debug stmts.
2620 (lower_try_catch): Skip debug stmts.
2621 * gimple.c (gimple_build_debug_begin_stmt): New.
2622 (gimple_copy): Increment debug_marker_count if copying one.
2623 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2624 * gimplify.c (rexpr_location): New.
2625 (rexpr_has_location): New.
2626 (warn_switch_unreachable_r): Handle gimple debug stmts.
2627 (shortcut_cond_r): Call expr_location.
2629 (find_goto_label): New.
2630 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2632 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2634 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2635 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2636 (LANG_HOOKS_INITIALIZER): ... this.
2637 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2638 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2640 (update_ebb_live_info): Skip debug insn markers.
2641 * lra.c (debug_insn_static_data): Rename to...
2642 (debug_bind_static_data): ... this.
2643 (debug_marker_static_data): New.
2644 (lra_set_insn_recog_data): Select one of the above depending
2646 (lra_update_isn_regno_info): Don't assume debug insns have
2648 (push_insns): Skip debug insns.
2649 * lto-streamer-in.c (input_function): Drop debug stmts
2650 depending on active options. Adjust debug_nonbind_markers.
2651 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2652 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2653 begin stmt marker notes.
2654 (print_insn): Likewise.
2655 * recog.c (extract_insn): Recognize rtl for debug markers.
2656 * rtl.def (DEBUG_MARKER): New.
2657 * tree-inline.c: Include params.h.
2658 (remap_gimple_stmt): Handle nonbind markers.
2659 (maybe_move_debug_stmts_to_successors): Likewise.
2660 (copy_debug_stmt): Likewise.
2661 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2662 markers regardless of no side effects.
2663 (tsi_link_before): Don't update container's side effects when adding
2664 a begin stmt marker.
2665 (tsi_link_after): Likewise.
2666 (expr_first): Skip begin stmt markers.
2667 (expr_last): Likewise.
2668 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2669 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2670 Disregard nonbind markers.
2671 * tree.c (make_node_stat): Don't set side effects for begin stmt
2673 (build1_stat): Likewise.
2674 * tree.def (DEBUG_BEGIN_STMT): New.
2675 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2676 * var-tracking.c (delete_debug_insns): Renamed to...
2677 (delete_vta_debug_insns): ... this.
2678 (reemit_marker_as_note): New.
2679 (vt_initialize): Reemit markers.
2680 (delete_vta_debug_insns): Likewise.
2681 (vt_debug_insns_local): Reemit or delete markers.
2682 (variable_tracking_main_1): Likewise.
2683 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2684 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2685 (gimple_debug_nonbind_marker_p): New.
2686 (gimple_build_debug_bind): Adjust.
2687 (gimple_build_debug_begin_stmt): New.
2688 * doc/invoke.texi (max-debug-marker-count): New param.
2689 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2690 (debug_parameter_ref, debug_marker): New.
2691 (NOTE_INSN_BEGIN_STMT): New.
2692 (DEBUG_INSN): Describe begin stmt markers.
2694 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2695 without debug insns, we wouldn't, but clean up debug insns
2696 after a control flow insn nevertheless.
2698 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2699 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2700 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2701 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2702 (rtl_tidy_fallthru_edge): Likewise.
2703 (rtl_verify_fallthru): Likewise.
2704 (rtl_verify_bb_layout): Likewise.
2705 (skip_insns_after_block): Likewise.
2706 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2707 * dwarf2out.c: Include print-rtl.h.
2708 (dwarf2out_next_real_insn): New.
2709 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2710 Dump debug binds in asm comments.
2711 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2712 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2713 callers to use gsi_start_nondebug_bb instead.
2714 (gsi_after_labels): Skip gimple debug stmts.
2715 (gsi_start_nondebug): New.
2716 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2717 (proper_loop_form_for_interchange): Adjust.
2718 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2719 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2720 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2721 (collect_fallthrough_labels): Likewise.
2722 (should_warn_for_implicit_fallthrough): Likewise.
2723 (warn_implicit_fallthrough_r): Likewise.
2724 (expand_FALLTHROUGH_r): Likewise.
2725 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2726 (graphite_copy_stmts_from_block): Skip nonbind markers.
2727 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2728 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2729 * jump.c (clean_barriers): Skip debug insns.
2730 * omp-expand.c (expand_parallel_call): Skip debug insns.
2731 (expand_task_call): Likewise.
2732 (remove_exit_barrier): Likewise.
2733 (expand_omp_taskreg): Likewise.
2734 (expand_omp_for_init_counts): Likewise.
2735 (expand_omp_for_generic): Likewise.
2736 (expand_omp_for_static_nochunk): Likewise.
2737 (expand_omp_for_static_chunk): Likewise.
2738 (expand_omp_simd): Likewise.
2739 (expand_omp_taskloop_for_outer): Likewise.
2740 (expand_omp_taskloop_for_inner): Likewise.
2741 (expand_oacc_for): Likewise.
2742 (expand_omp_sections): Likewise.
2743 (expand_omp_single): Likewise.
2744 (expand_omp_synch): Likewise.
2745 (expand_omp_atomic_load): Likewise.
2746 (expand_omp_atomic_store): Likewise.
2747 (expand_omp_atomic_fetch_op): Likewise.
2748 (expand_omp_atomic_pipeline): Likewise.
2749 (expand_omp_atomic_mutex): Likewise.
2750 (expand_omp_target): Likewise.
2751 (grid_expand_omp_for_loop): Likewise.
2752 (grid_expand_target_grid_body): Likewise.
2753 (build_omp_regions_1): Likewise.
2754 * omp-low.c (check_combined_parallel): Skip debug stmts.
2755 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2756 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2757 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2759 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2760 (make_edges): Likewise.
2761 (cleanup_dead_labels): Likewise.
2762 (gimple_can_merge_blocks_p): Likewise.
2763 (stmt_starts_bb_p): Likewise.
2764 (gimple_block_label): Likewise.
2765 (gimple_redirect_edge_and_branch): Likewise.
2766 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2768 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2770 * tree-pretty-print (print_declaration): Omit initializer in slim
2772 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2774 (eliminate_unnecessary_stmts): Stabilize block removal order.
2775 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2776 * var-tracking.c (get_first_insn): New.
2777 (vt_emit_notes): Call it.
2778 (vt_initialize): Walk any insns before the first BB.
2779 (delete_debug_insns): Likewise.
2781 * gimple.h (enum gimple_debug_subcode): Add
2782 GIMPLE_DEBUG_BEGIN_STMT.
2783 (gimple_debug_begin_stmt_p): New.
2784 (gimple_debug_nonbind_marker_p): New.
2785 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2786 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2787 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2788 * insn-notes.def (BEGIN_STMT): New.
2789 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2790 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2791 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2792 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2793 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2794 (INSN_DEBUG_MARKER_KIND): New.
2795 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2796 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2797 (INSN_VAR_LOCATION_PTR): New.
2798 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2800 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2801 deps only if debug bind insns are enabled.
2802 (pass_expand::execute): Avoid deep TER and expand
2803 debug locations for debug bind insns only.
2804 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2805 debug stmts special handling down to debug bind stmts.
2806 * combine.c (try_combine): Narrow debug insns special handling
2807 down to debug bind insns.
2808 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2809 Narrow debug insns preexisting special handling down to debug
2811 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2812 handling down to debug bind insns.
2813 * function.c (instantiate_virtual_regs): Skip debug markers,
2814 adjust handling of debug binds.
2815 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2816 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2817 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2818 insns to debug bind insns.
2819 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2820 special handling of debug stmts to debug bind stmts.
2821 * ipa-split.c (split_function): Likewise.
2822 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2823 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2825 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2826 * regrename.c (build_def_use): Likewise.
2827 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2828 (pass_cprop_hardreg): Narrow special casing of debug insns to
2830 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2831 * reload1.c (reload): Likewise.
2832 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2833 casing of debug stmts to debug bind stmts.
2834 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2835 * ssa-iterators.h (num_imm_uses): Likewise.
2836 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2837 debug stmts to debug bind stmts.
2838 * tree-inline.c (tree_function_versioning): Narrow special casing
2839 of debug stmts to debug bind stmts.
2840 * tree-loop-distribution.c (generate_loops_for_partition):
2841 Narrow special casing of debug stmts to debug bind stmts.
2842 * tree-sra.c (analyze_access_subtree): Narrow special casing
2843 of debug stmts to debug bind stmts.
2844 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2845 stmts to debug bind stmts.
2846 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2847 casing of debug stmts to debug bind stmts.
2848 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2849 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2850 casing of debug stmts to debug bind stmts.
2851 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2853 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2854 debug stmts to debug bind stmts.
2855 (gimple_replace_ssa_lhs): Likewise.
2856 (insert_debug_temp_for_var_def): Likewise.
2857 (insert_debug_temps_for_defs): Likewise.
2858 (reset_debug_uses): Likewise.
2859 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2860 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2861 (adjust_debug_stmts): Likewise.
2862 (adjust_phi_and_debug_stmts): Likewise.
2863 (vect_do_peeling): Likewise.
2864 * tree-vect-loop.c (vect_transform_loop): Likewise.
2865 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2866 * var-tracking.c (adjust_mems): Narrow special casing of debug
2867 insns to debug bind insns.
2868 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2869 (compute_bb_dataflow, vt_find_locations): Likewise.
2870 (vt_expand_loc, emit_notes_for_changes): Likewise.
2871 (vt_init_cfa_base): Likewise.
2872 (vt_emit_notes): Likewise.
2873 (vt_initialize): Likewise.
2874 (vt_finalize): Likewise.
2876 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2877 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2878 (next_nonnote_nondebug_insn_bb): New.
2879 (prev_nonnote_nondebug_insn_bb): New.
2880 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2881 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2882 (prev_nonnote_nondebug_insn_bb): Declare.
2883 (next_nonnote_nondebug_insn_bb): Declare.
2884 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2885 * cfgrtl.c (get_last_bb_insn): Likewise.
2886 * lra.c (push_insns): Likewise.
2888 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2891 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2893 * selftest.c (selftest::test_runner::test_runner): New ctor.
2894 (selftest::test_runner::~test_runner): New dtor.
2895 * selftest.h (class selftest::test_runner): New class.
2897 2017-12-11 Carl Love <cel@us.ibm.com>
2899 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2900 vec_extract_fp32_from_shortl]): Add #defines.
2901 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2902 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2903 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2904 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2905 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2906 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2907 * doc/extend.texi: Add documentation for the added builtins.
2909 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2911 PR rtl-optimization/80693
2912 PR rtl-optimization/81019
2913 PR rtl-optimization/81020
2914 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2915 are not mentioned in i3. Place the REG_UNUSED note on i2,
2916 possibly modified to REG_DEAD, if it did not originate in i3.
2918 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2920 * recog.c (store_data_bypass_p_1): New function.
2921 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2922 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2924 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2926 PR rtl-optimization/83361
2927 * ifcvt.c (if_convert): Call fixup_partitions.
2929 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2931 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2932 early folding of splat_u{8,16,32}.
2934 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2936 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2937 output_operand_lossage first argument with capital letter.
2938 (aarch64_override_options): Don't start error and sorry first argument
2939 with capital letter.
2941 2017-12-11 Andi Kleen <ak@linux.intel.com>
2943 PR gcov-profile/83355
2944 * auto-profile.c (string_table::get_index_by_decl): Don't
2945 recurse when abstract origin points to itself.
2947 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2949 PR tree-optimization/83320
2950 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2951 (prune_datarefs_not_in_loop): Ditto.
2953 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2955 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2957 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2959 PR tree-optimization/83337
2960 * gimple-loop-interchange.cc (compute_access_stride): Handle
2961 bitfield DRs properly.
2963 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2965 PR tree-optimization/83338
2966 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2967 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2968 vector of unsigned integers to vector of signed integers.
2970 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2972 PR rtl-optimization/83317
2973 * lra-constraints.c (process_address_1): Add insn code check.
2975 2017-12-08 Michael Matz <matz@suse.de>
2977 Fix PR tree-optimization/83323
2978 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2979 head-controlled loops and loop BBs.
2980 * common.opt (funroll-and-jam): Remove, instead ...
2981 (floop-unroll-and-jam): ... reuse this option.
2982 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2983 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2984 (-floop-unroll-and-jam): ... this option.
2986 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2988 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2989 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2990 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2992 2017-12-08 Richard Biener <rguenther@suse.de>
2995 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2996 handle accesses outside of zero-sized vars.
2998 2017-12-08 Martin Jambor <mjambor@suse.cz>
3000 PR tree-optimization/83141
3001 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3002 test for MEM_REFs implicitely changing types with padding. Remove
3004 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3006 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3008 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3010 * config/avr/avr.c (avr_attribute_table): Likewise.
3011 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3012 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3013 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3014 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3015 of struct attribute_spec.
3016 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3018 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3021 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3023 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3026 * profile-count.c (profile_count::from_gcov_type): Move from
3027 profile-count.h; handle overflow.
3028 * profile-count.h (profile_count::from_gcov_type): Move offline.
3030 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3032 PR rtl-optimization/83304
3033 * combine.c (move_deaths): If we do not know where a register died,
3036 2017-12-08 Richard Biener <rguenther@suse.de>
3038 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3039 Provide -fopt-info-loop feedback when we interchange in a nest.
3041 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3043 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3044 for armv6 ARM CPU IDs.
3046 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3048 * common/config/arm/arm-common.c: Include <algorithm>.
3049 (INCLUDE_VECTOR): Define.
3050 (compare_opt_names): New function.
3051 (arm_rewrite_selected_arch): Only strip out extensions that can be
3052 expressed through -mfpu. Sort the remaining extensions
3055 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3057 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3058 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3059 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3060 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3062 2017-06-08 Tristan Gingold <gindold@adacore.com>
3065 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3066 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3068 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3070 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3071 of struct attribute_spec.
3073 2017-12-08 Julia Koval <julia.koval@intel.com>
3075 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3076 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3077 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3078 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3079 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3080 _mm_maskz_dpwssds_epi32): Ditto.
3082 2017-12-08 Richard Biener <rguenther@suse.de>
3084 PR tree-optimization/81303
3085 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3086 conditions try to create a comparison vector type matching
3087 the data vector type.
3088 (vectorizable_condition): Adjust.
3089 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3090 Leave invariant conditions alone in case we can vectorize those.
3092 2017-12-08 Julia Koval <julia.koval@intel.com>
3094 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3095 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3096 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3097 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3098 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3100 2017-12-08 Julia Koval <julia.koval@intel.com>
3102 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3103 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3104 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3105 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3106 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3107 _mm_maskz_dpbusds_epi32): New intrinsics.
3109 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3111 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3113 (ld<bh>io_signed): Likewise.
3114 (st<bhw>io): Likewise.
3115 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3116 SMALL_INT12 constant integer operand.
3118 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3120 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3122 (nios2_large_constant_allowed): ...this. Adjust uses.
3123 (nios2_plus_symbolic_constant_p): Rename to...
3124 (nios2_plus_large_constant_p): ...this. Adjust uses.
3125 (nios2_legitimate_address_p): Correct CONST_INT handling.
3126 (nios2_symbolic_memory_operand_p): Rename to...
3127 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3128 (nios2_large_constant_p): Check for large constant integers too.
3129 (nios2_split_large_constant): Handle constant integers.
3130 (nios2_split_symbolic_memory_operand): Rename to...
3131 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3132 (nios2_legitimize_constant_address): Handle constant integers.
3133 (r0rel_constant_p): Handle small constant integers.
3134 (nios2_print_operand_address): Handle r0-relative integer addresses.
3135 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3136 * config/nios2/nios2.md: Adjust for renamed functions.
3138 2017-12-07 Andrew Waterman <andrew@sifive.com>
3140 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3141 (TARGET_HAVE_SRODATA_SECTION): New define.
3142 (riscv_select_section): New function.
3144 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3145 Alexander Monakov <amonakov@ispras.ru>
3146 Jakub Jelinek <jakub@redhat.com>
3149 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3151 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3153 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3154 a bool success value. Don't call output_operand_lossage here.
3155 (aarch64_print_ldpstp_address): Return a bool success value.
3156 (aarch64_print_operand_address): Call output_addr_const if
3157 aarch64_print_address_internal fails.
3158 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3159 'y'; call output_operand_lossage instead. Call output_operand_lossage
3160 if aarch64_print_ldpstp_address fails.
3162 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3164 * tree-vector-builder.h
3165 (tree_vector_builder::binary_encoded_nelts): Declare.
3166 * tree-vector-builder.c
3167 (tree_vector_builder::binary_encoded_nelts): New function.
3168 * fold-const.c (negate_expr_p): Likewise.
3169 (operand_equal_p, fold_checksum_tree): Likewise.
3170 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3171 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3172 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3173 (uniform_vector_p): Likewise.
3174 * varasm.c (const_hash_1, compare_constant): Likewise.
3175 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3176 (valid_lattice_transition): Operate directly on the VECTOR_CST
3178 * ipa-icf.c: Include tree-vector-builder.h.
3179 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3180 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3182 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3184 * tree.c (build_vector): Delete.
3185 * tree.h (build_vector): Make static and move into the self-testing
3188 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3190 * vector-builder.h (vector_builder::derived): New const overload.
3191 (vector_builder::elt): New function.
3192 * tree-vector-builder.h (tree_vector_builder::type): New function.
3193 (tree_vector_builder::apply_step): Declare.
3194 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3196 * gimple-fold.h (tree_vector_builder): Declare.
3197 (gimple_build_vector): Take a tree_vector_builder instead of a
3198 type and vector of elements.
3199 * gimple-fold.c (gimple_build_vector): Likewise.
3200 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3202 (get_initial_defs_for_reduction): Likewise.
3203 (vectorizable_induction): Likewise.
3205 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3207 * tree-vector-builder.h
3208 (tree_vector_builder::new_binary_operation): Declare.
3209 * tree-vector-builder.c
3210 (tree_vector_builder::new_binary_operation): New function.
3211 * fold-const.c (fold_relational_const): Use it.
3212 (const_binop): Likewise. Check that both input vectors have
3213 the same number of elements, thus excluding things like WIDEN_SUM.
3214 Check whether it is possible to operate directly on the encodings
3217 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3219 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3220 new_unary_operation, operating only on the encoded elements.
3221 (const_unop): Likewise.
3222 (exact_inverse): Likewise.
3223 (distributes_over_addition_p): New function.
3224 (const_binop): Use tree_vector_builder and new_unary_operation
3225 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3226 on the encoded elements unless the encoding is strided and the
3227 operation does not distribute over addition.
3228 (fold_convert_const): Use tree_vector_builder and
3229 new_unary_operation. Operate only on the encoded elements
3230 for truncating integer conversions, or for non-stepped encodings.
3232 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3234 * config/sparc/sparc.c: Include tree-vector-builder.h.
3235 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3236 * expmed.c: Include tree-vector-builder.h.
3237 (make_tree): Use tree_vector_builder instead of build_vector.
3238 * fold-const.c: Include tree-vector-builder.h.
3239 (const_binop): Use tree_vector_builder instead of build_vector.
3240 (const_unop): Likewise.
3241 (native_interpret_vector): Likewise.
3242 (fold_vec_perm): Likewise.
3243 (fold_ternary_loc): Likewise.
3244 * gimple-fold.c: Include tree-vector-builder.h.
3245 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3247 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3248 (simplify_vector_constructor): Use tree_vector_builder instead
3250 * tree-vect-generic.c: Include tree-vector-builder.h.
3251 (add_rshift): Use tree_vector_builder instead of build_vector.
3252 (expand_vector_divmod): Likewise.
3253 (optimize_vector_constructor): Likewise.
3254 * tree-vect-loop.c: Include tree-vector-builder.h.
3255 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3256 of build_vector. Explicitly use a stepped encoding for
3258 * tree-vect-slp.c: Include tree-vector-builder.h.
3259 (vect_get_constant_vectors): Use tree_vector_builder instead
3261 (vect_transform_slp_perm_load): Likewise.
3262 (vect_schedule_slp_instance): Likewise.
3263 * tree-vect-stmts.c: Include tree-vector-builder.h.
3264 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3265 (vect_gen_perm_mask_any): Likewise.
3266 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3267 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3269 (build_vector_from_val): Likewise. Explicitly use a duplicate
3272 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3274 * doc/generic.texi (VECTOR_CST): Describe new representation of
3276 * vector-builder.h: New file.
3277 * tree-vector-builder.h: Likewise.
3278 * tree-vector-builder.c: Likewise.
3279 * Makefile.in (OBJS): Add tree-vector-builder.o.
3280 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3281 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3282 (tree_vector): Change the number of elements to
3283 vector_cst_encoded_nelts.
3284 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3285 (VECTOR_CST_ELTS): Delete.
3286 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3287 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3288 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3289 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3290 (VECTOR_CST_ENCODED_ELT): Likewise.
3291 (vector_cst_encoded_nelts): New function.
3292 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3293 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3294 (vector_cst_int_elt, vector_cst_elt): Declare.
3295 * tree.c: Include tree-vector-builder.h.
3296 (tree_code_size): Abort if passed VECTOR_CST.
3297 (tree_size): Update for new VECTOR_CST layout.
3298 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3299 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3300 (build_vector): Use tree_vector_builder.
3301 (vector_cst_int_elt, vector_cst_elt): New functions.
3302 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3303 encoded elements and then create the vector in the canonical form.
3304 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3305 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3306 (tree_c_tests): Call test_vector_cst_patterns.
3307 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3309 (hash_tree): Likewise.
3310 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3311 (streamer_write_tree_header): Likewise.
3312 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3313 (streamer_alloc_tree): Likewise. Update call to make_vector.
3314 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3316 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3318 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3319 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3320 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3322 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3323 Richard Biener <rguenther@suse.de>
3325 PR tree-optimization/81303
3326 * Makefile.in (gimple-loop-interchange.o): New object file.
3327 * common.opt (floop-interchange): Reuse the option from graphite.
3328 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3329 -floop-interchange and mention it for -O3.
3330 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3331 * gimple-loop-interchange.cc: New file.
3332 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3333 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3334 * passes.def (pass_linterchange): New pass.
3335 * timevar.def (TV_LINTERCHANGE): New time var.
3336 * tree-pass.h (make_pass_linterchange): New declaration.
3337 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3338 interchange. Record IV before/after increment in new parameters.
3339 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3340 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3342 (check_reduction_path): ...New function here.
3343 * tree-vectorizer.h (check_reduction_path): New declaration.
3345 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3348 PR rtl-optimization/80818
3349 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3350 always early clobbered.
3351 * lra-lives.c (process_bb_lives): Check input hard regs for early
3352 clobbered non-operand hard reg.
3354 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3357 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3358 types_compatible_p, just that TYPE_MODE is the same.
3360 2017-12-07 Martin Sebor <msebor@redhat.com>
3363 * attribs.c (empty_attribute_table): Initialize new member of
3364 struct attribute_spec.
3365 (decl_attributes): Add argument. Handle mutually exclusive
3366 combinations of attributes.
3367 (selftests::test_attribute_exclusions): New function.
3368 (selftests::attribute_c_tests): Ditto.
3369 * attribs.h (decl_attributes): Add default argument.
3370 * selftest.h (attribute_c_tests): Declare.
3371 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3372 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3374 * doc/extend.texi (Common Function Attributes): Update const and pure.
3375 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3376 of struct attribute_spec.
3377 * config/arc/arc.c (arc_attribute_table): Same.
3378 * config/arm/arm.c (arm_attribute_table): Same.
3379 * config/avr/avr.c ( avr_attribute_table): Same.
3380 * config/bfin/bfin.c (bfin_attribute_table): Same.
3381 * config/cr16/cr16.c (cr16_attribute_table): Same.
3382 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3383 * config/h8300/h8300.c (h8300_attribute_table): Same.
3384 * config/i386/i386.c (ix86_attribute_table): Same.
3385 * config/ia64/ia64.c (ia64_attribute_table): Same.
3386 * config/m32c/m32c.c (m32c_attribute_table): Same.
3387 * config/m32r/m32r.c (m32r_attribute_table): Same.
3388 * config/m68k/m68k.c (m68k_attribute_table): Same.
3389 * config/mcore/mcore.c (mcore_attribute_table): Same.
3390 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3391 * config/mips/mips.c (mips_attribute_table): Same.
3392 * config/msp430/msp430.c (msp430_attribute_table): Same.
3393 * config/nds32/nds32.c (nds32_attribute_table): Same.
3394 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3395 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3396 * config/rl78/rl78.c (rl78__attribute_table): Same.
3397 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3398 * onfig/rx/rx.c (rx_attribute_table): Same.
3399 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3400 * config/sh/sh.c (sh_attribute_table): Same.
3401 * config/sparc/sparc.c (sparc_attribute_table): Same.
3402 * config/spu/spu.c (spu_attribute_table): Same.
3403 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3404 * config/v850/v850.c (v850_attribute_table): Same.
3405 * config/visium/visium.c (visium_attribute_table): Same.
3407 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3410 * config/arm/arm.c (INCLUDE_STRING): Define.
3411 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3412 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3415 2017-12-07 Michael Matz <matz@suse.de>
3417 Add unroll and jam pass
3419 * gimple-loop-jam.c: New file.
3420 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3421 * common.opt (funroll-and-jam): New option.
3422 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3423 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3424 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3425 * passes.def: Add pass_loop_jam.
3426 * timevar.def (TV_LOOP_JAM): Add.
3427 * tree-pass.h (make_pass_loop_jam): Declare.
3428 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3429 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3430 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3431 to flow_loop_tree_node_add.
3432 (duplicate_subloops, copy_loops_to): Append to sibling list.
3433 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3434 * doc/invoke.texi (-funroll-and-jam): Document new option.
3435 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3437 2017-12-07 Richard Biener <rguenther@suse.de>
3439 PR tree-optimization/83296
3440 PR tree-optimization/67769
3441 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3442 flow sensitive info in an unrelated BB.
3443 (value_replacement): Use reset_flow_sensitive_info.
3444 (minmax_replacement): Reset flow sensitive info on the def
3445 we move. Do not reset flow sensitive info in the whole BB
3446 we move the stmt to.
3447 (abs_replacement): Likewise.
3449 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3453 rs6000_cpu to the given -mcpu=, or to the default processor.
3455 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3457 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3458 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3459 instead of rs6000_cpu_attr.
3460 (rs6000_adjust_cost): Ditto.
3461 (is_microcoded_insn): Ditto.
3462 (rs6000_adjust_priority): Ditto.
3463 (rs6000_issue_rate): Ditto.
3464 (rs6000_use_sched_lookahead): Ditto.
3465 (rs6000_use_sched_lookahead_guard): Ditto.
3466 (rs6000_sched_reorder): Ditto.
3467 (force_new_group): Ditto.
3468 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3469 (group_ending_nop): Ditto.
3471 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3473 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3474 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3475 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3476 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3478 (rs6000_reassociation_width): Ditto.
3479 (rs6000_emit_epilogue): Ditto.
3480 (rs6000_adjust_cost): Ditto.
3481 (is_microcoded_insn): Ditto.
3482 (is_cracked_insn): Ditto.
3483 (rs6000_adjust_priority): Ditto.
3484 (rs6000_sched_reorder): Ditto.
3485 (rs6000_sched_reorder2): Ditto.
3486 (insn_must_be_first_in_group): Ditto.
3487 (insn_must_be_last_in_group): Ditto.
3488 (rs6000_register_move_cost): Ditto.
3489 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3492 2017-12-07 Julia Koval <julia.koval@intel.com>
3494 * config.gcc: Add vaesintrin.h.
3495 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3497 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3498 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3500 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3501 * config/i386/immintrin.h: Include vaesintrin.h.
3502 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3503 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3504 _mm_aesdec_epi128): New intrinsics.
3506 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3508 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3509 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3510 c-family/c-spellcheck.cc.
3511 (best_macro_match::best_macro_match): Likewise.
3513 (struct edit_distance_traits<cpp_hashnode *>): Move to
3514 c-family/c-spellcheck.h.
3515 (class best_macro_match): Likewise.
3517 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3519 PR tree-optimization/83293
3520 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3521 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3522 might insert into empty bb.
3525 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3526 simplify for plus with :c added, and pointer_plus without that.
3527 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3528 with undefined overflow and the conversion is not widening,
3529 perform negation in utype and only convert to type afterwards.
3530 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3531 simplify for plus with :c added, and pointer_plus without that.
3532 If type is integral with undefined overflow and the conversion is
3533 not widening, perform minus in utype and only convert to type
3534 afterwards. Move the last pointer_diff_expr simplify into the
3537 2017-12-06 Martin Sebor <msebor@redhat.com>
3539 PR tree-optimization/82646
3540 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3541 strncpy, not maxlen.
3543 2017-12-06 Martin Sebor <msebor@redhat.com>
3545 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3548 PR tree-optimization/83075
3549 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3550 strncat/strncpy don't change length of source string.
3552 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3555 2017-11-29 Martin Aberg <maberg@gaisler.com>
3557 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3558 to prevent b2bst errata sequence.
3559 (sqrtdf2_fix): Likewise.
3561 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3563 PR tree-optimization/81945
3564 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3565 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3566 to dest_cfun has orig_loop_num set, either remap it to the new
3567 loop number if the loop got moved too, or clear it.
3569 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3571 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3573 (thunderx2t99_nothing): New insn reservation.
3574 (thunderx2t99_mrs): New insn reservation.
3575 (thunderx2t99_multiple): New insn reservation.
3576 (thunderx2t99_alu_basi): Add bfx to reservation.
3577 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3579 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3582 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3584 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3586 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3587 tree-ssa-pre.c::remove_dead_inserted_code.
3588 * tree-ssa-dce.h: New file.
3589 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3590 (remove_dead_inserted_code): Move and rename to function
3591 tree-ssa-dce.c::simple_dce_from_worklist.
3592 (pass_pre::execute): Update use.
3594 2017-12-05 Richard Biener <rguenther@suse.de>
3596 PR tree-optimization/83277
3597 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3598 to code-gen liveout vars.
3600 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3602 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3603 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3604 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3607 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3609 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3611 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3612 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3615 2017-12-05 Richard Biener <rguenther@suse.de>
3617 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3618 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3619 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3620 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3622 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3624 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3626 2017-12-05 Martin Liska <mliska@suse.cz>
3627 Jakub Jelinek <jakub@redhat.com>
3629 * doc/invoke.texi: Document the options.
3630 * flag-types.h (enum sanitize_code): Add
3631 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3632 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3633 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3634 * opts.c: Define new sanitizer options.
3635 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3636 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3638 2017-12-05 Julia Koval <julia.koval@intel.com>
3640 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3641 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3642 (ix86_handle_option): Handle -mavx512vnni.
3643 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3644 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3645 * config/i386/i386-c (__AVX512VNNI__): New.
3646 * config/i386/i386.c (ix86_target_string): Handle new option.
3647 (ix86_valid_target_attribute_inner_p): Handle new option.
3648 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3649 * config/i386/i386.opt (mavx512vnni): New option.
3651 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3654 * athlon.md: Disable for generic.
3655 * haswell.md: Enable for generic.
3656 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3657 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3659 (ix86_adjust_cost): Move generic to haswell path.
3661 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3663 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3664 instead of MEM_P in a couple more places. Fix formatting issues.
3666 2017-12-04 Jim Wilson <jimw@sifive.com>
3668 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3669 instead of GP_REG_LAST-1.
3670 (riscv_adjust_libcall_cfi_prologue): Likewise.
3671 (riscv_adjust_libcall_cri_epilogue): Likewise.
3672 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3675 2017-12-04 Luis Machado <luis.machado@linaro.org>
3677 * ipa-pure-const.c (check_decl): Add missing newline.
3678 (state_from_flags): Likewise.
3680 2017-12-04 Jeff Law <law@redhat.com>
3682 PR tree-optimizatin/78496
3683 * gimple-ssa-evrp-analyze.h
3684 (evrp_range_analyzer::get_vr_values): Simplify.
3685 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3686 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3687 and gimple-ssa-evrp-analyze.h.
3688 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3689 (simplify_stmt_for_jump_threading): Copy a blob of code from
3690 tree-vrp.c to use ranges to simplify statements.
3691 (dom_opt_dom_walker::before_dom_children): Call
3692 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3693 (dom_opt_dom_walker::after_dom_children): Similarly for
3694 evrp_range_analyzer::leave.
3695 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3698 * gimple-ssa-evrp-analyze.c
3699 (evrp_range_analyzer::extract_range_from_stmt): Always use
3700 vr_values::update_value_range so preexisting range info is
3701 medged with new range info, even if the new range is VR_VARYING.
3703 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3705 * combine.c: Adjust comment.
3706 (use_crosses_set_p): Delete.
3707 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3708 (try_combine): Ditto.
3710 2017-12-04 Richard Biener <rguenther@suse.de>
3712 PR tree-optimization/83255
3713 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3714 Re-add zero-iteration check.
3716 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3718 PR rtl-optimization/83245
3719 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3720 hard registers as earlyclobber, also if not in an asm.
3722 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3726 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3730 rs6000_cpu based on cpu_index, not tune_index.
3732 2017-12-04 Richard Biener <rguenther@suse.de>
3734 PR tree-optimization/83238
3735 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3736 code match comment, rejecting invalid SESE regions.
3738 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3740 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3741 require base operand is a REG_POINTER prior to reload on targets
3742 with non-equivalent space registers.
3744 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3746 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3747 (update_specialized_profile): Fix updating of counts.
3748 (perhaps_add_new_callers): Likewise.
3750 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3753 * x86-tune.def: Remove obsolette FIXMEs.
3754 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3755 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3756 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3758 (X86_TUNE_PAD_RETURNS): Disable for generic.
3760 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3762 PR tree-optimization/83170
3763 PR tree-optimization/83241
3764 * gimple-ssa-store-merging.c
3765 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3766 gimple_vuse (ins_stmt) in case it has changed.
3767 (imm_store_chain_info::output_merged_store): Likewise.
3769 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3773 * tree-cfg.c (pass_warn_function_return::execute): Handle
3774 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3775 with BUILTINS_LOCATION.
3779 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3780 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3781 instead of DECL_MODE.
3783 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3784 last argument to rtx pointer.
3785 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3786 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3787 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3788 depending on the chosen ISAs.
3789 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3790 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3791 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3793 * config/i386/sse.md (mov<mode>_internal): Likewise.
3794 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3796 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3798 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3800 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3802 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3803 parameters from prototype.
3804 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3805 parameters. Don't print a comment.
3806 (emit_fusion_gpr_load): Adjust.
3807 (emit_fusion_load_store): Adjust.
3808 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3809 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3810 comment on the second line.
3812 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3816 rs6000_cpu based on cpu_index, not tune_index.
3818 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3820 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3821 which_alternative instead of which_alternative + 1.
3822 (output_asm_insn): Print an extra tab if the template is short.
3824 2017-12-01 Jim Wilson <jimw@sifive.com>
3826 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3828 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3829 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3831 * doc/tm.texi: Regenerate.
3833 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3836 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3837 whether we can allocate pseudos before trying to fix an address.
3838 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3839 memory address is indexed or indirect.
3840 (floatuns_<mode>si2_hw2): Likewise.
3842 2017-12-01 Jason Merrill <jason@redhat.com>
3844 * Makefile.in (TAGS): Add c-family/*.cc.
3846 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3848 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3849 (call_value_insn): Likewise.
3850 (sibcall_insn): Likewise.
3851 (sibcall_value_insn): Likewise.
3852 (movsi_aarch64): Likewise.
3853 (movdi_aarch64): Likewise.
3854 (add_losym_): Likewise.
3855 (ldr_got_small_): Likewise.
3856 (ldr_got_small_sidi): Likewise.
3857 (ldr_got_small_28k_): Likewise.
3858 (ldr_got_small_28k_sidi): Likewise.
3859 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3860 Move output_addr_const to symbolic case. Add error check.
3862 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3864 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3865 * config/s390/s390.md: Disable MVC merging peephole if it would
3866 disable operand forwarding.
3867 (new peephole2): Split MVCs if it would turn them into up to 2
3870 2017-12-01 Richard Biener <rguenther@suse.de>
3872 PR tree-optimization/83232
3873 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3874 detection of same access. Instead of breaking the group here
3875 do not consider the duplicate. Add comment explaining real fix.
3877 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3879 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3881 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3883 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3885 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3887 * function.h (struct function): Remove cilk_frame_decl,
3888 is_cilk_function and calls_cilk_spawn fields.
3889 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3891 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3892 cilk_elemental field.
3893 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3894 * target.def: Adjust comment.
3895 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3896 Don't test cilk_elemental.
3898 PR tree-optimization/83233
3899 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3900 bswap_stat name for the struct.
3903 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3904 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3905 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3906 SWITCH_BREAK_LABEL_P set on the label.
3907 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3908 added for default case if it was missing and not all cases covered.
3909 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3910 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3912 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3913 to avoid -Wimplicit-fallthrough warning.
3914 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3915 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3917 PR tree-optimization/83221
3918 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3920 (init_reassoc): Formatting fix.
3923 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3924 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3928 * tree-cfg.c: Include asan.h.
3929 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3931 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3933 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3935 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3937 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3938 vpcmpeqd instruction.
3940 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3942 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3943 registers type for 128bit mode.
3945 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3947 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3949 * gengtype-state.c (read_state_pair): Likewise.
3950 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3951 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3952 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3953 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3954 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3955 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3956 * builtins.c (fold_builtin_3): Likewise.
3957 * graphite-scop-detection.c
3958 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3959 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3962 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3965 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3966 parameter. Return true if ignore_decl_rtl_set_p is true and other
3967 conditions are satisfied.
3968 * asan.h (asan_protect_global): Add new parameter.
3969 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3970 to asan_protect_global calls.
3972 2017-11-30 Jim Wilson <jimw@sifive.com>
3974 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3975 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3976 -mstrict-align, add info on default value. Delete redundant lines for
3977 -mabi. Add missing -mexplicit-relocs docs.
3979 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3981 * config/arc/arc.md (trap): New pattern.
3983 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3985 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3986 instruction to end into a delay slot.
3987 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3988 can be placed into a delay slot against reg_note.
3990 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3992 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3993 labels number of usages.
3995 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3997 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3999 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4001 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4004 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4005 multiplication by power of 2 constant into two shifts + comparison.
4007 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4010 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4012 2017-11-30 Richard Biener <rguenther@suse.de>
4014 PR tree-optimization/83202
4015 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4016 allow_peel argument and guard peeling.
4017 (canonicalize_loop_induction_variables): Likewise.
4018 (canonicalize_induction_variables): Pass false.
4019 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4020 peeling from cunrolli.
4022 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4024 * combine.c (try_combine): Print a message to dump file whenever
4025 I0, I1, or I2 cannot be combined into I3.
4027 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4029 PR rtl-optimization/83156
4030 PR rtl-optimization/82621
4031 * combine.c (try_combine): Don't split an I2 if one of the dests is
4032 set again before I3. Allow unused dests.
4034 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4036 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4038 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4040 PR rtl-optimization/80818
4041 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4042 recursively. Use insn code for clobber.
4043 (lra_set_insn_recog_data): Pass the new arg to
4044 collect_non_operand_hard_regs.
4045 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4047 (lra_update_insn_regno_info): Pass insn to
4048 add_regs_to_insn_regno_info.
4050 2017-11-29 Jim Wilson <jimw@sifive.com>
4051 Andrew Waterman <andrew@sifive.com>
4053 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4054 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4055 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4056 use SINGLE_SHIFT_COST.
4057 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4058 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4060 2017-11-29 Julia Koval <julia.koval@intel.com>
4062 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4063 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4064 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4065 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4066 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4067 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4068 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4069 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4070 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4071 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4072 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4073 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4074 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4075 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4076 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4077 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4078 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4079 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4080 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4081 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4082 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4083 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4084 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4085 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4086 __builtin_ia32_vpshldv_v2di_mask,
4087 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4088 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4089 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4091 2017-11-29 Julia Koval <julia.koval@intel.com>
4093 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4094 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4095 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4096 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4097 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4098 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4099 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4100 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4101 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4102 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4103 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4104 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4105 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4106 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4107 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4108 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4109 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4110 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4111 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4112 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4113 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4115 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4117 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4118 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4121 2017-11-29 Martin Aberg <maberg@gaisler.com>
4123 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4124 to prevent b2bst errata sequence.
4125 (sqrtdf2_fix): Likewise.
4127 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4129 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4130 (div_sqrt_insn_p): New function.
4131 (sparc_do_work_around_errata): Insert NOP instructions to
4132 prevent sequences that could trigger the TN-0013 errata for
4133 certain LEON3 processors.
4134 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4135 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4136 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4137 (in_branch_delay): Prevent div and sqrt in delay slot if
4139 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4141 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4143 * config/sparc/sparc.c (atomic_insn_p): New function.
4144 (sparc_do_work_around_errata): Insert NOP instructions to
4145 prevent sequences that could trigger the TN-0010 errata for
4147 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4148 instruction referable in atomic_insns_p.
4150 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4152 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4153 (atomic_compare_and_swap_leon3_1): Likewise.
4156 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4158 * config/sparc/sparc.c (fpop_insn_p): New function.
4159 (sparc_do_work_around_errata): Insert NOP instructions to
4160 prevent sequences that could trigger the TN-0012 errata for
4162 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4163 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4164 (in_branch_annul_delay): Prevent floating-point instructions
4165 in delay slot of annulled integer branch.
4167 2017-11-29 Richard Biener <rguenther@suse.de>
4169 PR tree-optimization/83202
4170 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4172 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4173 nodes vectorized to the same stmts multiple times.
4174 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4175 (vect_analyze_slp_instance): Adjust.
4176 (scalar_stmts_to_slp_tree_map_t): New typedef.
4177 (vect_schedule_slp_instance): Add a map recording the SLP node
4178 representing the vectorized stmts for a set of scalar stmts.
4179 Avoid code-generating redundancies.
4180 (vect_schedule_slp): Allocate map and pass it down.
4182 2017-11-29 Nathan Sidwell <nathan@acm.org>
4185 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4188 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4190 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4191 cases for printing LDP/STP memory addresses.
4192 (aarch64_print_address_internal): Renamed from
4193 aarch64_print_operand_address, added parameter, add Pmode check.
4194 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4195 (aarch64_print_operand_address): Indirect to
4196 aarch64_print_address_internal.
4197 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4198 'y' operand output specifier.
4200 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4203 * tree.c (build_simple_mem_ref_loc): Handle
4204 get_addr_base_and_unit_offset returning a MEM_REF.
4207 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4210 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4213 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4215 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4217 (layout::move_to_column): Likewise.
4219 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4222 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4223 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4225 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4226 Martin Jambor <mjambor@suse.cz>
4229 * tree.h (expr_type_first_operand_type_p): Declare
4230 * tree.c (expr_type_first_operand_type_p): New function.
4231 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4232 (ipa_value_from_jfunc): Adjust declaration.
4233 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4234 Use it as result type for arithmetics, unless it is NULL in which case
4235 be more conservative.
4236 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4237 ipa_get_jf_pass_through_result.
4238 (propagate_vals_across_pass_through): Likewise.
4239 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4240 is to propagate_vals_across_pass_through.
4241 (propagate_constants_across_call): Pass PARM_TYPE to
4242 propagate_scalar_across_jump_function.
4243 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4244 ipa_value_from_jfunc.
4245 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4246 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4247 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4248 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4249 target_type, pass it to ipa_value_from_jfunc.
4250 (update_indirect_edges_after_inlining): Pass parameter type to
4251 try_make_edge_direct_simple_call.
4253 2017-11-28 Jeff Law <law@redhat.com>
4255 * gimple-ssa-evrp-analyze.c
4256 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4257 refine ranges if scev_initialized_p returns true.
4258 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4260 2017-11-28 Julia Koval <julia.koval@intel.com>
4262 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4263 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4264 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4265 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4266 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4267 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4268 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4269 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4270 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4271 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4272 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4273 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4274 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4275 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4276 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4277 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4278 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4279 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4280 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4281 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4282 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4283 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4285 2017-11-28 Julia Koval <julia.koval@intel.com>
4287 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4288 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4289 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4290 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4291 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4292 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4293 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4294 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4295 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4296 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4297 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4298 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4299 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4300 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4301 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4302 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4303 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4304 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4305 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4306 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4307 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4308 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4309 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4310 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4311 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4312 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4313 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4314 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4315 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4316 __builtin_ia32_vpshld_v2di_mask): New builtins.
4317 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4318 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4320 2017-11-28 Richard Biener <rguenther@suse.de>
4322 PR tree-optimization/80776
4323 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4325 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4327 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4328 If the incoming edge is an effective fallthru because the other
4329 edge only reaches a __builtin_unreachable () then record ranges
4330 derived from the controlling condition in SSA info.
4331 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4332 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4334 2017-11-28 Olivier Hainque <hainque@adacore.com>
4336 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4337 on mingw build hosts.
4339 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4342 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4344 2017-11-28 Richard Biener <rguenther@suse.de>
4347 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4348 copies generated from memcpy use a character array as reference
4351 2017-11-28 Julia Koval <julia.koval@intel.com>
4352 Sebastian Peryt <sebastian.peryt@intel.com>
4354 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4355 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4356 cilk-common.o, cilk.h, cilk-common.c): Remove.
4358 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4359 * builtins.c (is_builtin_name): Remove cilkplus condition.
4360 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4361 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4362 cilk-builtins.def, cilkplus.def): Remove.
4363 * cif-code.def (CILK_SPAWN): Remove.
4364 * cilk-builtins.def: Delete.
4365 * cilk-common.c: Ditto.
4367 * cilkplus.def: Ditto.
4368 * config/darwin.h (fcilkplus): Delete.
4369 * cppbuiltin.c: Ditto.
4370 * doc/extend.texi: Remove cilkplus doc.
4371 * doc/generic.texi: Ditto.
4372 * doc/invoke.texi: Ditto.
4373 * doc/passes.texi: Ditto.
4374 * gcc.c (fcilkplus): Remove.
4375 * gengtype.c (cilk.h): Remove.
4376 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4378 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4380 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4382 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4383 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4384 cilkplus conditions.
4385 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4386 inline_read_section): Ditto.
4387 * ipa-inline-analysis.c (cilk.h): Remove.
4388 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4389 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4390 append_linker_options): Remove condition for fcilkplus.
4391 * lto/lto-lang.c (cilk.h): Remove.
4392 (lto_init): Remove condition for fcilkplus.
4393 * omp-expand.c (expand_cilk_for_call): Delete.
4394 (expand_omp_taskreg, expand_omp_for_static_chunk,
4395 expand_omp_for): Remove cilkplus
4397 (expand_cilk_for): Delete.
4398 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4399 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4400 execute_lower_omp, diagnose_sb_0): Ditto.
4401 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4402 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4403 * tree-nested.c: Ditto.
4404 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4405 (dump_generic_node): Ditto.
4406 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4407 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4409 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4411 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4413 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4414 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4415 for vector mode and !TARGET_SIMD.
4417 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4419 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4421 * tree.h (SWITCH_LABELS): Remove.
4422 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4423 assert SWITCH_BODY is non-NULL.
4424 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4426 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4428 PR tree-optimization/80788
4429 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4430 has TREE_OVERFLOW set, call drop_tree_overflow.
4432 2017-11-28 Richard Biener <rguenther@suse.de>
4434 PR tree-optimization/83158
4435 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4437 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4440 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4441 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4443 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4445 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4446 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4448 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4451 * varasm.c (compare_constant): Take the mode of the constants into
4452 account when comparing floating point constants.
4454 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4456 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4457 from explicit instantiation of debug_helper.
4458 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4460 2017-11-27 Richard Biener <rguenther@suse.de>
4462 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4465 2017-11-27 Richard Biener <rguenther@suse.de>
4467 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4468 (type_hash_canon): Also clear int_cst_hash_table entry for
4470 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4472 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4474 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4475 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4476 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4478 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4480 * hash-map.h (gt_cleare_cache): Avoid UB.
4482 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4484 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4485 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4486 (decide_unroll_runtime_iterations): Likewise.
4487 (decide_unroll_stupid): Likewise.
4489 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4492 * config/i386/i386.md: Add a loop with incssp.
4494 2017-11-27 Martin Jambor <mjambor@suse.cz>
4496 PR tree-optimization/81248
4497 * tree-sra.c (splice_param_accesses): Remove size check.
4498 (decide_one_param_reduction): Fix size check.
4499 * gimple-pretty-print.c (dump_profile): Silence warning.
4500 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4502 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4505 * dbxout.c (lastlineno): New variable.
4506 (dbx_debug_hooks): Use dbxout_switch_text_section as
4507 switch_text_section debug hook.
4508 (dbxout_function_end): Switch to current_function_section
4509 rather than function_section. If crtl->has_bb_partition,
4510 output just one N_FUN, depending on in_cold_section_p.
4511 (dbxout_source_line): Remember last lineno in lastlineno.
4512 (dbxout_switch_text_section): New function.
4513 (dbxout_function_decl): Adjust dbxout_block caller.
4514 (dbx_block_with_cold_children): New function.
4515 (dbxout_block): Return true if any LBRAC/RBRAC have been
4516 emitted. Use dbx_block_with_cold_children at depth == 0
4517 in second partition. Add PARENT_BLOCKNUM argument, pass
4518 it optionally adjusted to children. Output LBRAC/RBRAC
4519 around recursive call only if the block is in the current
4520 partition, if not and anything was output, emit empty
4522 * final.c (final_scan_insn): Compute cold_function_name
4523 before calling switch_text_section debug hook. Call
4524 that hook even if dwarf2out_do_frame if not emitting
4528 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4529 TREE_READONLY decls.
4531 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4533 PR rtl-optimization/82488
4534 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4536 2017-11-26 Julia Koval <julia.koval@intel.com>
4538 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4540 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4543 2017-11-26 Julia Koval <julia.koval@intel.com>
4545 * config/i386/driver-i386.c (host_detect_local_cpu):
4546 Detect skylake-avx512.
4548 2017-11-26 Julia Koval <julia.koval@intel.com>
4550 * config.gcc: Add -march=cannonlake.
4551 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4552 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4553 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4554 (PTA_CANNONLAKE): New.
4555 (processor_target_table): Add cannonlake.
4556 (ix86_option_override_internal): Ditto.
4557 (fold_builtin_cpu): Ditto.
4558 (get_builtin_code_for_version): Handle cannonlake.
4559 (M_INTEL_COREI7_CANNONLAKE): New.
4560 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4561 * doc/invoke.texi: Add -march=cannonlake.
4563 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4565 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4566 (try_init_one_plugin): Alternative implementation for MinGW.
4567 * Makefile.in (plugin_implib): New.
4568 (gengtype-lex.c): Fix broken AIX workaround.
4569 * configure: Regenerate.
4570 * doc/plugins.texi: Document support for MinGW.
4572 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4574 PR rtl-optimization/81553
4575 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4576 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4577 is a shift where C1 has different mode than the whole shift, use C1's
4578 mode for MULT rather than the shift's mode.
4581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4582 builtins not enabled in the currently selected ISA.
4584 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4586 PR tree-optimization/71026
4587 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4588 (insert_reciprocals): Change to insert reciprocals before a division
4589 by a square and to insert the square of a reciprocal.
4590 (execute_cse_reciprocals_1): Change to consider division by a square.
4591 (register_division_in): Add importance parameter.
4593 2017-11-24 Richard Biener <rguenther@suse.de>
4595 PR tree-optimization/82402
4596 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4597 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4599 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4601 * match.pd (0-ptr): New transformation.
4603 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4606 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4607 when skipping checking.
4609 2017-11-24 Richard Biener <rguenther@suse.de>
4611 PR tree-optimization/83128
4612 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4613 (vn_reference_lookup_3): Likewise.
4615 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4618 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4619 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4622 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4623 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4624 check for the latter with gimple_call_builtin_p. Do not handle
4625 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4627 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4629 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4632 2017-11-23 Julia Koval <julia.koval@intel.com>
4634 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4635 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4636 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4637 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4638 _mm512_maskz_expandloadu_epi16): New intrinsics.
4639 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4640 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4641 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4642 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4643 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4644 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4645 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4646 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4647 _mm256_maskz_expandloadu_epi8): New intrinsics.
4648 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4649 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4650 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4651 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4652 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4653 * config/i386/sse.md (VI248_VLBW): New iterator.
4654 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4656 2017-11-23 Julia Koval <julia.koval@intel.com>
4658 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4659 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4660 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4661 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4662 _mm512_mask_compressstoreu_epi16): New.
4663 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4664 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4665 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4666 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4667 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4668 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4669 _mm256_mask_compressstoreu_epi8): New.
4670 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4671 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4672 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4673 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4674 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4675 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4676 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4677 __builtin_ia32_compresshi128_mask,
4678 __builtin_ia32_compressstoreuqi512_mask,
4679 __builtin_ia32_compressstoreuhi512_mask,
4680 __builtin_ia32_compressstoreuqi256_mask,
4681 __builtin_ia32_compressstoreuqi128_mask,
4682 __builtin_ia32_compressstoreuhi256_mask,
4683 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4684 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4686 (ix86_expand_special_args_builtin): Handle new types.
4687 (s4fma_expand): Handle new builtin array.
4688 * config/i386/immintrin.h: Include new headers.
4689 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4690 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4692 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4694 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4695 fixes. Declare temp and g variables at the top in order to avoid
4696 {} in most of the cases.
4698 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4700 * match.pd (ptr-0): New transformation.
4702 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4704 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4705 (struct addr_mode_cost_table): New.
4706 (struct tune_params): Add field addr_mode_costs.
4707 * config/arm/arm.c (generic_addr_mode_costs): New.
4708 (arm_slowmul_tune): Initialise addr_mode_costs field.
4709 (arm_fastmul_tune): Likewise.
4710 (arm_strongarm_tune): Likewise.
4711 (arm_xscale_tune): Likewise.
4712 (arm_9e_tune): Likewise.
4713 (arm_marvell_pj4_tune): Likewise.
4714 (arm_v6t2_tune): Likewise.
4715 (arm_cortex_tune): Likewise.
4716 (arm_cortex_a8_tune): Likewise.
4717 (arm_cortex_a7_tune): Likewise.
4718 (arm_cortex_a15_tune): Likewise.
4719 (arm_cortex_a35_tune): Likewise.
4720 (arm_cortex_a53_tune): Likewise.
4721 (arm_cortex_a57_tune): Likewise.
4722 (arm_exynosm1_tune): Likewise.
4723 (arm_xgene1_tune): Likewise.
4724 (arm_cortex_a5_tune): Likewise.
4725 (arm_cortex_a9_tune): Likewise.
4726 (arm_cortex_a12_tune): Likewise.
4727 (arm_cortex_a73_tune): Likewise.
4728 (arm_v7m_tune): Likewise.
4729 (arm_cortex_m7_tune): Likewise.
4730 (arm_v6m_tune): Likewise.
4731 (arm_fa726te_tune): Likewise.
4732 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4734 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4736 * config/arm/arm.c (arm_mem_costs): New function.
4737 (arm_rtx_costs_internal): Use arm_mem_costs.
4739 2017-11-23 Mark Wielaard <mark@klomp.org>
4741 * dwarf2out.c (init_sections_and_labels): Use generation to create
4742 unique ranges_section_label and ranges_base_label. Return generation.
4743 (output_rnglists): Add generation argument. Use generation to create
4744 unique ranges labels.
4745 (dwarf2out_finish): Get generation from init_sections_and_labels
4746 and pass generation to output_rnglists.
4748 2017-11-23 Mike Stump <mikestump@comcast.net>
4749 Eric Botcazou <ebotcazou@adacore.com>
4751 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4752 * cfgloop.h (struct loop): Add unroll field.
4753 * function.h (struct function): Add has_unroll bitfield.
4754 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4755 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4756 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4757 (pass_rtl_unroll_loops::gate): Likewise.
4758 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4759 for which loop->unroll==1.
4760 (decide_unroll_constant_iterations): Use note for consistency and
4761 take loop->unroll into account. Return early if loop->unroll is set.
4762 Fix thinko in existing test.
4763 (decide_unroll_runtime_iterations): Use note for consistency and
4764 take loop->unroll into account.
4765 (decide_unroll_stupid): Likewise.
4766 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4767 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4768 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4770 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4771 (print_loop): Print loop->unroll if set.
4772 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4773 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4774 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4776 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4777 loop->unroll is set and smaller than the trip count. Otherwise bypass
4778 entirely the heuristics if loop->unroll is set. Remove dead note.
4779 Fix off-by-one bug in other note.
4780 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4781 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4783 (tree_unroll_loops_completely): Make static.
4784 (pass_complete_unroll::execute): Use correct type for variable.
4785 (pass_complete_unrolli::execute): Fix formatting.
4786 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4788 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4790 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4791 enable when TARGET_PREFER_AVX128 is set.
4793 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4795 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4797 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4798 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4800 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4802 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4803 frequencies. Fix estimation of aggregate parameters.
4805 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4807 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4810 2017-11-23 Tom de Vries <tom@codesourcery.com>
4812 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4814 2017-11-23 Tom de Vries <tom@codesourcery.com>
4816 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4819 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4822 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4823 sibcall_value_pcrel_fdpic): Use local variable instead of
4825 (calli_tbr_rel): Add missing operand 2.
4826 (call_valuei_tbr_rel): Add missing operand 3.
4828 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4831 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4832 bitpos/bitsize covering the whole destination, use store_expr only if
4833 the complex mode is the same. Otherwise, use expand_normal and if
4834 it returns CONCAT, subreg each part separately instead of trying to
4835 subreg the whole result.
4837 2017-11-23 Richard Biener <rguenther@suse.de>
4839 PR tree-optimization/23094
4840 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4841 come here from walking over backedges in the first iteration.
4842 (vn_reference_lookup_3): Skip clobbers that store the same value.
4844 2017-11-23 Richard Biener <rguenther@suse.de>
4846 PR tree-optimization/81403
4847 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4848 a block we need a leader relative to.
4849 (phi_translate_1): For nary processing require a leader from
4850 get_representative_for given we run expression simplification
4851 using match-and-simplify. Remove previous fix.
4853 2017-11-22 Jeff Law <law@redhat.com>
4855 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4856 Use new method allocate_value_range rather than accessing the
4857 vrp_value_range_pool data member directly.
4858 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4859 to use extract_range_from_stmt method to avoid need for
4860 extract_range_from_assignment method.
4861 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4862 method rather than setting values_propgated data member directly.
4863 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4864 and values propagated data members and extract_range_from_assignment
4865 method. Reorder private data members to conform to standards.
4866 Add new methods set_lattice_propagation_complete and
4867 allocate_value_range.
4869 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4871 PR rtl-optimization/83030
4872 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4873 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4874 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4875 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4876 CROSSING_JUMP_P flag is already set before setting it.
4877 * cfgrtl.c (fixup_partition_crossing): Likewise.
4878 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4881 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4883 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4884 Handle the case where both arguments are using gen_const_vec_series.
4886 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4889 * pretty-print.c (pp_format): Move quoting implementation to
4890 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4891 to pass address of "quote" local.
4892 (pp_begin_quote): New function.
4893 (pp_end_quote): New function.
4894 * pretty-print.h (printer_fn): Convert penultimate param from bool
4896 (pp_begin_quote): New decl.
4897 (pp_end_quote): New decl.
4898 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4899 param from bool to bool *.
4900 * tree-diagnostic.h (default_tree_printer): Likewise.
4902 2017-11-22 Jeff Law <law@redhat.com>
4904 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4905 Initialize vr_values.
4906 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4907 vr_values via vr_values class instance rather than delegators.
4908 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4909 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4910 (evrp_range_analyzer::push_value_range): Likewise.
4911 (evrp_range_analyzer::pop_value_range): Likewise.
4912 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4913 most delegators. Those remaining are exposed as public interfaces.
4914 Make vr_values a pointer and private.
4915 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4917 (evrp_range_analyzer::get_vr_value): New method.
4918 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4919 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4920 (class evrp_dom_walker): Attach evrp_folder class, initialize
4921 it in the ctor. Remove temporary delegators.
4922 (evrp_dom_walker::before_dom_children): Call methods in attached
4923 evrp_range_analyzer class via class instance pointer. Use
4924 free value_range_constant_singleton to remove need for
4925 op_with_constant_singleton_value delegator method. Do not
4926 create a vrp_prop class instance for every call! Narrow
4927 scope of a couple variables.
4928 (evrp_dom_walker::cleanup): Call methods in attached
4929 evrp_range_analyzer class via class instance pointer.
4930 * vr-values.h (class vr_values): Privatize many methods and
4933 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4935 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4936 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4937 * expr.c (expand_expr_real_2): Likewise.
4938 * fold-const.c (const_unop): Likewise.
4939 * optabs-tree.c (optab_for_tree_code): Likewise.
4940 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4941 * tree-inline.c (estimate_operator_cost): Likewise.
4942 * tree-pretty-print.c (dump_generic_node): Likewise.
4943 (op_code_prio): Likewise.
4944 (op_symbol_code): Likewise.
4945 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4946 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4947 * internal-fn.c (direct_internal_fn_optab): New function.
4948 (direct_internal_fn_array, direct_internal_fn_supported_p
4949 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4950 * fold-const-call.c (fold_const_reduction): New function.
4951 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4953 * tree-vect-loop.c: Include internal-fn.h.
4954 (reduction_code_for_scalar_code): Rename to...
4955 (reduction_fn_for_scalar_code): ...this and return an internal
4957 (vect_model_reduction_cost): Take an internal_fn rather than
4959 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4961 (vectorizable_reduction): Use internal functions rather than tree
4962 codes for the reduction operation. Update calls to the functions
4964 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4965 Use calls to internal functions rather than REDUC tree codes.
4966 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4968 2017-11-22 Olivier Hainque <hainque@adacore.com>
4970 * config/vxworks.c (vxworks_override_options): Pick default
4971 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4972 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4973 DWARF_GNAT_ENCODINGS_DEFAULT.
4974 * config/vxworksae.h: Likewise.
4976 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4978 PR tree-optimization/83104
4979 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4982 2017-11-22 Marek Polacek <polacek@redhat.com>
4983 H.J. Lu <hongjiu.lu@intel.com>
4984 Jason Merrill <jason@redhat.com>
4989 * calls.c (initialize_argument_information): Call
4990 warn_parameter_passing_abi target hook.
4991 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4992 argument onto stack.
4993 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4994 * common.opt: Update -fabi-version description.
4995 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4996 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4998 (ix86_is_empty_record): New function.
4999 (ix86_warn_parameter_passing_abi): New function.
5000 (TARGET_EMPTY_RECORD_P): Redefine.
5001 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5002 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5003 * doc/tm.texi: Regenerated.
5004 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5005 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5006 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5007 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5008 instead of int_size_in_bytes.
5009 * expr.c (copy_blkmode_to_reg): Likewise.
5010 * function.c (aggregate_value_p): Return 0 for empty types.
5011 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5013 (locate_and_pad_parm): Call arg size_in_bytes instead
5015 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5016 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5017 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5019 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5020 (std_gimplify_va_arg_expr): Skip empty records. Call
5021 arg_size_in_bytes instead size_in_bytes.
5022 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5023 * tree-core.h (tree_type_common): Add empty_flag.
5024 (tree_decl_common): Update comments.
5025 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5027 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5028 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5030 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5031 * tree.c (default_is_empty_type): New function.
5032 (default_is_empty_record): New function.
5033 (arg_int_size_in_bytes): New function.
5034 (arg_size_in_bytes): New function.
5035 (get_ultimate_context): New function.
5036 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5037 TRANSLATION_UNIT_WARN_EMPTY_P.
5038 (default_is_empty_record, arg_int_size_in_bytes,
5039 arg_size_in_bytes, get_ultimate_context): Declare.
5041 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5043 * config/arm/arm.c (cmse_clear_registers): New function.
5044 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5045 code by call to cmse_clear_registers.
5046 (cmse_nonsecure_entry_clear_before_return): Likewise.
5048 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5050 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5051 (vdot_s32, vdotq_s32): New.
5052 (vdot_lane_u32, vdotq_lane_u32): New.
5053 (vdot_lane_s32, vdotq_lane_s32): New.
5056 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5059 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5060 for unsigned values with fewer HWIs than the precision.
5061 (test_overflow): New function.
5062 (wide_int_cc_tests): Call it.
5064 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5066 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5067 has the same mode class as Pmode.
5069 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5071 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5072 padding_bits_to_clear_ptr.
5073 (cmse_nonsecure_entry_clear_before_return): Likewise.
5075 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5077 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5078 auto_sbitap instead of integer bitfield to control register needing
5081 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5083 PR tree-optimization/83044
5084 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5085 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5086 that need the upper bound. Subtract offset from
5087 get_addr_base_and_unit_offset only if positive and subtract it
5088 before division by eltsize rather than after it.
5091 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5092 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5093 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5097 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5098 expand_binop with *mul_widen_optab, make sure at least one of the
5099 operands doesn't have VOIDmode.
5102 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5104 PR rtl-optimization/82044
5105 PR tree-optimization/82042
5106 * dse.c (record_store): Check for overflow.
5107 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5108 add_wild_read instead of clear_rhs_from_active_local_stores on
5109 overflow and log it into dump_file.
5111 2017-11-22 Richard Biener <rguenther@suse.de>
5113 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5114 fake edges to exit when looking for a place to insert.
5115 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5117 (insert_into_preds_of_block): Commit edge insertion immediately,
5118 assert that doesn't require new BBs.
5119 (fini_pre): Release expressions.
5120 (pass_pre::execute): Shuffle things around a bit, if the fn
5121 is too large do not compute AVAIL either as this is really the
5124 2017-11-22 Richard Biener <rguenther@suse.de>
5126 PR tree-optimization/83089
5127 * tree-if-conv.c (pass_if_conversion::execute): If anything
5128 changed reset SCEV and free the number of iteration estimates.
5130 2017-11-21 Martin Sebor <msebor@redhat.com>
5132 PR tree-optimization/82945
5133 * calls.h (warn_nonstring_bound): Remove unused function.
5135 2017-11-21 Martin Sebor <msebor@redhat.com>
5137 PR tree-optimization/82945
5138 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5139 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5140 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5142 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5143 * calls.h (get_attr_nonstring_decl): Declare new function.
5144 * doc/extend.texi (attribute nonstring): Update.
5145 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5146 get_attr_nonstring_decl and handle it.
5147 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5148 detection of nul-termination.
5149 (strlen_to_stridx): Change to a pointer.
5150 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5151 (pass_strlen::execute): Same.
5153 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5155 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5156 for the new option -mprefer-vector-width=[none|128|256|512].
5157 * config/i386/i386.c (ix86_target_string): remove old style options
5158 -mprefer-avx256 and make -mprefer-avx128 as alias.
5159 (ix86_option_override_internal): Apply defaults for the
5160 -mprefer-vector-width=[128|256] option.
5161 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5162 Implement macros to work with -mprefer-vector-width=.
5163 * config/i386/i386.opt: Implemented option
5164 -mprefer-vector-width=[none|128|256|512].
5165 * doc/invoke.texi: Documentation for
5166 -mprefer-vector-width=[none|128|256|512].
5168 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5170 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5172 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5174 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5176 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5178 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5180 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5181 * expr.c (expand_expr_real_2): Likewise.
5182 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5183 fold_binary_loc): Likewise.
5184 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5185 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5186 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5187 MINUS_EXPR transformations.
5188 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5189 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5190 * tree-inline.c (estimate_operator_cost): Likewise.
5191 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5192 op_symbol_code): Likewise.
5193 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5194 * vr-values.c (extract_range_from_binary_expr): Likewise.
5195 * varasm.c (initializer_constant_valid_p_1): Likewise.
5196 * tree.def: New tree code POINTER_DIFF_EXPR.
5198 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5200 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5201 integer suffix to movbe mnemonic.
5202 (*bswaphi2_movbe): Ditto.
5203 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5205 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5208 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5209 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5210 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5212 2017-11-21 Martin Liska <mliska@suse.cz>
5214 * tree-inline.c (expand_call_inline): Remove not needed
5217 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5218 Jakub Jelinek <jakub@redhat.com>
5221 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5222 Remove static keyword from f variable.
5224 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5226 PR tree-optimization/83086
5227 * gimple-ssa-store-merging.c
5228 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5229 rather than n.base_addr.
5231 2017-11-21 Martin Liska <mliska@suse.cz>
5233 PR rtl-optimization/82044
5234 PR tree-optimization/82042
5235 * dse.c (check_mem_read_rtx): Check for overflow.
5237 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5239 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5242 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5244 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5245 and its corresponding call arg location note.
5247 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5248 Andrew Burgess <andrew.burgess@embecosm.com>
5250 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5252 (arc_return_slot_offset): Likewise.
5253 (arc_eh_return_address_location): New declaration.
5254 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5255 (MUST_SAVE_REGISTER): Add exception handler case.
5256 (MUST_SAVE_RETURN_ADDR): Likewise.
5257 (arc_frame_pointer_required): Likewise.
5258 (arc_frame_pointer_needed): New function.
5259 (arc_compute_frame_size): Changed.
5260 (arc_expand_prologue): Likewise.
5261 (arc_expand_epilogue): Likewise.
5262 (arc_initial_elimination_offset): Likewise.
5263 (arc_return_slot_offset): Delete.
5264 (arc_eh_return_address_location): New function.
5265 (arc_builtin_setjmp_frame_value): Likewise.
5266 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5267 (EH_RETURN_STACKADJ_RTX): Define.
5268 (EH_RETURN_HANDLER_RTX): Likewise.
5269 * config/arc/arc.md (eh_return): Delete.
5271 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5273 * print-tree.h (debug_vec_tree): Remove prototype.
5274 * gdbinit.in (pvt): Remove macro.
5276 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5278 PR tree-optimization/83047
5279 * gimple-ssa-store-merging.c
5280 (imm_store_chain_info::output_merged_store): If the loads with the
5281 same vuse are in different basic blocks, for load_gsi pick a load
5282 location that is dominated by the other loads.
5285 * config/i386/i386.c (ix86_memmodel_check): Start
5286 -Winvalid-memory-model diagnostics with lowercase letter.
5289 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5290 set in_cold_section_p to the partition containing loc_list->first.
5291 When seeing loc_list->last_before_switch node, update secname and
5292 perform range_across_switch second partition handling only after that.
5295 * run-rtl-passes.c: Include debug.h.
5296 (run_rtl_passes): Call debug_hooks->assembly_start.
5297 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5301 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5302 OPTAB_DIRECT in calls to expand_simple_binop.
5304 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5307 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5309 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5312 * spellcheck.h (best_match::blithely_get_best_candidate): New
5315 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5317 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5318 with lower case letter.
5320 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5322 * config/i386/i386.md (bswaphi2): New expander.
5323 (*bswaphi2_movbe): New insn pattern.
5324 (bswaphi -> rorhi pepehole2): New peephole pattern.
5326 2017-11-20 Jeff Law <law@redhat.com>
5328 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5329 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5330 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5331 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5332 gimple-ssa-evrp-analyze.h.
5334 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5335 set BB_VISITED here.
5336 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5338 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5340 * tree-predcom.c: Add general comment on Store-Store chains.
5341 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5343 (get_chain_last_ref_at): Rename into...
5344 (get_chain_last_write_at): ...this.
5345 (get_chain_last_write_before_load): New function.
5346 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5347 CT_STORE_STORE when write reference is added.
5348 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5349 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5350 (initialize_root_vars_store_elim_1): Ditto.
5351 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5352 definition is created.
5353 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5354 chain by replacing it with dominant stored value.
5356 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5358 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5360 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5362 * vr-values.c (extract_range_from_binary_expr): Use a full range
5365 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5367 * config/arm/arm.md (R4_REGNUM): Define constant.
5368 (nonsecure_call_internal): Remove r4 clobber.
5369 (nonsecure_call_value_internal): Likewise.
5370 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5371 clobber and resequence match_operands.
5372 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5373 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5374 (nonsecure_call_value_reg_thumb2): Likewise.
5376 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5378 PR tree-optimization/78821
5379 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5380 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5381 to the address of the base rather than the base itself.
5382 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5383 (find_bswap_or_nop_finalize): New function.
5384 (find_bswap_or_nop): Use it.
5385 (bswap_replace): Return a tree rather than bool, change first
5386 argument from gimple * to gimple_stmt_iterator, allow inserting
5387 into an empty sequence, allow ins_stmt to be NULL - then emit
5388 all stmts into gsi. Fix up MEM_REF address gimplification.
5389 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5390 (struct store_immediate_info): Add N and INS_STMT non-static
5392 (store_immediate_info::store_immediate_info): Initialize them
5393 from newly added ctor args.
5394 (merged_store_group::apply_stores): Formatting fixes. Sort by
5396 (stmts_may_clobber_ref_p): For stores call also
5397 refs_anti_dependent_p.
5398 (gather_bswap_load_refs): New function.
5399 (imm_store_chain_info::try_coalesce_bswap): New method.
5400 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5401 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5402 (imm_store_chain_info::output_merged_store): Fail if number of
5403 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5404 and NOP_EXPR rhs_code.
5405 (pass_store_merging::process_store): Compute n and ins_stmt, if
5406 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5407 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5409 (pass_store_merging::execute): Calculate dominators.
5411 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5412 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5413 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5414 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5415 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5416 class pass_optimize_bswap, bswap_replace,
5417 pass_optimize_bswap::execute): Moved to ...
5418 * gimple-ssa-store-merging.c: ... this file.
5419 Include optabs-tree.h.
5420 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5421 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5422 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5423 anonymous namespace, remove static keywords.
5424 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5425 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5427 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5430 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5432 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5434 * vec.h (debug_helper): New function.
5435 (DEFINE_DEBUG_VEC): New macro.
5436 * hash-set.h (debug_helper): New function.
5437 (DEFINE_DEBUG_HASH_SET): New macro.
5438 * cfg.c (debug_slim (edge)): New function.
5439 Call DEFINE_DEBUG_VEC for edges.
5440 Call DEFINE_DEBUG_HASH_SET for edges.
5441 * cfghooks.c (debug_slim (basic_block)): New function.
5442 Call DEFINE_DEBUG_VEC for basic blocks.
5443 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5444 * print-tree.c (debug_slim): New function to handle trees.
5445 Call DEFINE_DEBUG_VEC for trees.
5446 Call DEFINE_DEBUG_HASH_SET for trees.
5447 (debug (vec<tree, va_gc>) &): Remove.
5448 (debug (<vec<tree, va_gc>) *): Remove.
5449 * print-rtl.c (debug_slim): New function to handle const_rtx.
5450 Call DEFINE_DEBUG_VEC for rtx_def.
5451 Call DEFINE_DEBUG_VEC for rtx_insn.
5452 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5453 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5454 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5455 (debug (vec<rtx_insn *> *ptr): Remove.
5456 (debug_insn_vector): Remove.
5457 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5459 2017-11-20 Tom de Vries <tom@codesourcery.com>
5461 PR rtl-optimization/82020
5462 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5463 IF_THEN_ELSE condition.
5465 2017-11-19 Jeff Law <law@redhat.com>
5467 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5468 of degenerates resulting from ignoring an edge.
5470 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5473 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5475 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5478 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5479 for uninitialied counts.
5481 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5484 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5486 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5489 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5491 2017-11-19 Tom de Vries <tom@codesourcery.com>
5493 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5496 2017-11-19 Tom de Vries <tom@codesourcery.com>
5498 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5501 2017-11-19 Tom de Vries <tom@codesourcery.com>
5503 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5506 2017-11-19 Tom de Vries <tom@codesourcery.com>
5508 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5510 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5512 2017-11-19 Tom de Vries <tom@codesourcery.com>
5514 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5516 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5517 semicolon after MCORE_EXPORT_NAME call.
5519 2017-11-19 Tom de Vries <tom@codesourcery.com>
5522 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5523 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5525 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5527 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5528 (cgraph_edge::clone): Cleanup updating of profile.
5529 * ipa-cp.c (update_profiling_info): Likewise.
5530 * ipa-inline-transform.c (inline_transform): Likewise.
5531 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5532 * ipa-split.c (execute_split_functions): Do not split when function
5534 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5535 * profile-count.c (profile_count::dump): Add global0.
5536 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5538 (profile_count::to_sreal_scale): Likewise.
5539 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5540 (profile_count::combine_with_ipa_count): New function.
5541 * profile-count.h (profile_guessed_global0adjusted): New.
5542 (profile_count::adjusted_zero): New.
5543 (profile_count::global0adjusted): New.
5544 (profile_count::combine_with_ipa_count): New.
5545 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5546 correct profile of return block of split functions.
5547 (copy_cfg_body): Remove unused profile_count.
5548 (copy_body): Likewise.
5549 (expand_call_inline): Update.
5550 (tree_function_versioning): Update.
5552 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5554 * hash-set.h (hash_set::empty): New.
5555 * tree-ssa-threadbackward.h: Delete.
5556 * tree-ssa-threadbackward.c (class thread_jumps): New.
5557 Move max_threaded_paths into class.
5558 (fsm_find_thread_path): Remove arguments that are now in class.
5559 (profitable_jump_thread_path): Rename to...
5560 (thread_jumps::profitable_jump_thread_path): ...this.
5561 (convert_and_register_jump_thread_path): Rename to...
5562 (thread_jumps::convert_and_register_current_path): ...this.
5563 (check_subpath_and_update_thread_path): Rename to...
5564 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5565 (register_jump_thread_path_if_profitable): Rename to...
5566 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5567 (handle_phi): Rename to...
5568 (thread_jumps::handle_phi): ...this.
5569 (handle_assignment): Rename to...
5570 (thread_jumps::handle_assignment): ...this.
5571 (fsm_find_control_statement_thread_paths): Rename to...
5572 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5573 (find_jump_threads_backwards): Rename to...
5574 (thread_jumps::find_jump_threads_backwards): ...this.
5575 Initialize path local data.
5576 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5577 from within thread_jumps class.
5578 (pass_early_thread_jumps::execute): Same.
5580 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5582 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5584 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5586 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5587 to canonical location.
5589 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5592 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5594 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5596 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5598 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5599 rcrt1.o%s/grcrt1.o%s for -static-pie.
5601 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5603 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5604 ix86_shift_rotate_cost): Break out from ...
5605 (ix86_rtx_costs): ... here.
5606 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5609 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5611 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5612 when entry block was promoted unlikely.
5613 (estimate_bb_frequencies): Increase frequency scale.
5614 * profile-count.h (profile_count): Export precision info.
5616 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5618 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5619 disturbin profile of entry block.
5621 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5623 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5626 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5628 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5629 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5631 2017-11-17 Jeff Law <law@redhat.com>
5633 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5634 from evrp_dom_walker class. Various methods moved into new class.
5635 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5636 (evrp_range_analyzer::enter): New method.
5637 (evrp_range_analyzer::leave): New method.
5638 (evrp_dom_walker): Remove delegators no longer needed by this class.
5639 Replace vr_values data member with evrp_range_analyzer
5641 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5642 method extracted from evrp_dom_walker::before_dom_children.
5643 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5644 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5646 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5647 Add private copy constructor and move assignment operators.
5648 Privatize methods and class data where trivially possible.
5649 (evrp_dom_walker::cleanup): New function, extracted from
5650 execute_early_vrp. Simplify access to class data.
5652 * vr-values.h (get_output_for_vrp): Prototype.
5653 * vr-values.c (get_output_for_vrp): New function extracted from
5654 vrp_visit_assignment_or_call and extract_range_from_stmt.
5655 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5657 2017-11-17 Luis Machado <luis.machado@linaro.org>
5659 * config/aarch64/aarch64.c
5660 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5661 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5662 tune_params::AUTOPREFETCHER_WEAK.
5664 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5667 * config/arm/arm.c (arm_valid_target_attribute_rec):
5668 Parse "arch=" and "+<ext>".
5669 (arm_valid_target_attribute_tree): Re-init global options.
5670 (arm_option_override): Make non-static.
5671 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5672 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5673 (__ARM_FEATURE_CRC32): Support undef.
5674 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5675 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5677 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5679 * gdbinit.in (break-on-diagnostic): New command.
5681 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5683 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5684 return to indirect jump for EH return if control-flow
5685 protection is enabled. Change explicit 'false' argument in
5686 pro_epilogue_adjust_stack with a value of
5688 * config/i386/i386.md (simple_return_indirect_internal):
5689 Remove SImode restriction to support 64-bit.
5691 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5693 * combine.c (added_notes_insn): New.
5694 (try_combine): Handle added_notes_insn like added_links_insn.
5695 Rewrite return value code.
5696 (distribute_notes): Set added_notes_insn to the earliest insn we added
5699 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5701 PR rtl-optimization/82621
5702 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5703 dest of one of those SETs is unused.
5705 2017-11-17 Richard Biener <rguenther@suse.de>
5708 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5709 * tree-pretty-print.c (dump_generic_node): Handle
5710 annot_expr_parallel_kind.
5711 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5712 * gimplify.c (gimple_boolify): Likewise.
5714 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5716 * config.gcc (extra_headers): Add cet.h for x86 targets.
5717 * config/i386/cet.h: New file.
5718 * doc/install.texi: Add --enable-cet/--disable-cet.
5720 2017-11-17 Richard Biener <rguenther@suse.de>
5722 PR tree-optimization/83017
5723 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5724 (gen_parallel_loop): Properly count iterations.
5725 (parallelize_loops): Handle loop->can_be_parallel independent
5726 of flag_loop_parallelize_all. Make static profitability test match
5728 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5729 * invoke.texi (parloops-min-per-thread): Document.
5731 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5733 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5734 upstreaming review comments.
5736 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5738 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5739 with fast unaligned access.
5740 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5742 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5744 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5745 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5747 2017-11-17 Richard Biener <rguenther@suse.de>
5749 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5750 folding of references.
5752 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5755 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5756 of replacing call to strncmp with corresponding call to strcmp when
5759 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5761 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5762 option prefer-avx256 for skylake-avx512 configuration.
5763 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5764 (get_builtin_code_for_version): Ditto.
5766 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5767 Monk Chiang <sh.chiang04@gmail.com>
5769 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5770 (FIXED_REGISTERS): Reserve more register numbers.
5771 (CALL_USED_REGISTERS): Likewise.
5772 (REG_ALLOC_ORDER): Likewise.
5773 (REG_CLASS_CONTENTS): Likewise.
5774 (REGISTER_NAMES): Likewise.
5776 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5777 Kito Cheng <kito.cheng@gmail.com>
5779 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5781 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5783 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5785 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5786 parameter list for vec_splats.
5788 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5790 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5792 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5794 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5797 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5798 with the required version of automake 1.11.6.
5800 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5802 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5804 (DU_C2_3_power9): Correct reservation combinations.
5805 (FP_DIV_power9, VEC_DIV_power9): New.
5806 (power9-alu): Split out rotate/shift...
5807 (power9-rot): ...to here, correct dispatch resource.
5808 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5810 (power9-fp): Correct latency.
5811 (power9-sdiv): Add div/sqrt resource.
5812 (power9-ddiv): Correct latency, add div/sqrt resource.
5813 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5814 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5816 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5818 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5820 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5821 switch statement mapping KF built-ins to TF built-ins if we don't
5822 have the proper ISA 3.0 assembler support.
5824 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5826 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5827 (lower_emutls_function_body): Do not compute it.
5829 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5831 * ipa-split.c (split_bb_info): Turn time to sreal.
5832 (split_point): Likewise.
5833 (dump_split_point): Likewise.
5834 (fine_split_points): Likewise.
5835 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5837 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5839 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5842 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5844 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5846 (expensive_function_p): Remove useless assert.
5847 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5849 2017-11-16 Martin Sebor <msebor@redhat.com>
5851 PR tree-optimization/82588
5852 PR tree-optimization/82583
5853 * tree-vrp.c (check_array_ref): Handle flexible array members,
5854 string literals, and inner indices.
5855 (search_for_addr_array): Add detail to diagnostics.
5857 2017-11-16 Nathan Sidwell <nathan@acm.org>
5861 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5862 (SET_DECL_ASSEMBLER_NAME): Forward to
5863 overwrite_decl_assembler_name.
5864 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5865 (overwrite_decl_assembler_name): Declare.
5866 * tree.c (overwrite_decl_assembler_name): New.
5867 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5868 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5869 (LANG_HOOKS_INITIALIZER): Add it.
5870 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5871 * langhooks.c (lhd_set_decl_assembler_name): Use
5872 SET_DECL_ASSEMBLER_NAME.
5873 (lhd_overwrite_decl_assembler_name): Default implementation.
5875 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5876 Jackson Woodruff <jackson.woodruff@arm.com>
5878 PR tree-optimization/71026
5879 * match.pd: Canonicalize constant multiplies in division.
5881 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5883 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5884 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5885 Remove OPT_fomit_frame_pointer entry.
5886 * common/config/arc/arc-common.c: Likewise.
5887 * common/config/arm/arm-common.c: Likewise.
5888 * common/config/avr/avr-common.c: Likewise.
5889 * common/config/c6x/c6x-common.c: Likewise.
5890 * common/config/cr16/cr16-common.c: Likewise.
5891 * common/config/cris/cris-common.c: Likewise.
5892 * common/config/epiphany/epiphany-common.c: Likewise.
5893 * common/config/fr30/fr30-common.c: Likewise.
5894 * common/config/frv/frv-common.c: Likewise.
5895 * common/config/ia64/ia64-common.c: Likewise.
5896 * common/config/iq2000/iq2000-common.c: Likewise.
5897 * common/config/lm32/lm32-common.c: Likewise.
5898 * common/config/m32r/m32r-common.c: Likewise.
5899 * common/config/mcore/mcore-common.c: Likewise.
5900 * common/config/microblaze/microblaze-common.c: Likewise.
5901 * common/config/mips/mips-common.c: Likewise.
5902 * common/config/mmix/mmix-common.c: Likewise.
5903 * common/config/mn10300/mn10300-common.c: Likewise.
5904 * common/config/nios2/nios2-common.c: Likewise.
5905 * common/config/pa/pa-common.c: Likewise.
5906 * common/config/pdp11/pdp11-common.c: Likewise.
5907 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5908 * common/config/riscv/riscv-common.c: Likewise.
5909 * common/config/rs6000/rs6000-common.c: Likewise.
5910 * common/config/rx/rx-common.c: Likewise.
5911 * common/config/s390/s390-common.c: Likewise.
5912 * common/config/sh/sh-common.c: Likewise.
5913 * common/config/sparc/sparc-common.c: Likewise.
5914 * common/config/tilegx/tilegx-common.c: Likewise.
5915 * common/config/tilepro/tilepro-common.c: Likewise.
5916 * common/config/v850/v850-common.c: Likewise.
5917 * common/config/visium/visium-common.c: Likewise.
5918 * common/config/xstormy16/xstormy16-common.c: Likewise.
5919 * common/config/xtensa/xtensa-common.c: Likewise.
5920 * invoke.texi (-fomit-frame-pointer): Update documentation.
5922 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5924 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5926 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5928 * cfg.c (scale_bbs_frequencies_int,
5929 cale_bbs_frequencies_gcov_type): Remove.
5930 * cfg.h (scale_bbs_frequencies_int,
5931 cale_bbs_frequencies_gcov_type): Remove.
5933 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5935 * tree-ssa-loop-manip.c
5936 (scale_dominated_blocks_in_loop): Update to profile counts.
5937 (tree_transform_and_unroll_loop): Likewise.
5939 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5941 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5942 scale_bbs_frequencies_int.
5944 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5946 * final.c (compute_alignments): Use counts rather than frequencies.
5948 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5950 * cfgloopanal.c: Include sreal.h
5951 (average_num_loop_insns): Use counts and sreal for accounting.
5953 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5955 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5958 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5959 Kito Cheng <kito.cheng@gmail.com>
5961 * config/nds32/constraints.md: Provide more constraints.
5962 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5963 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5964 support constraints usage.
5966 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5968 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5970 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5971 Kito Cheng <kito.cheng@gmail.com>
5973 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5974 * config/nds32/nds32.opt: Refine the layout.
5975 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5976 TARGET_EXT_STRING): Support new options.
5977 * config/nds32/nds32.h: Likewise.
5978 * config/nds32/nds32.md: Likewise.
5979 * config/nds32/nds32-predicates.c: Likewise.
5980 * config/nds32/constraints.md: Likewise.
5981 * common/config/nds32/nds32-common.c: Likewise.
5983 2017-11-16 Julia Koval <julia.koval@intel.com>
5986 * config/i386/gfniintrin.h: Add sse check.
5987 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5989 2017-11-16 Julia Koval <julia.koval@intel.com>
5991 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5992 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5993 (ix86_handle_option): Handle -mavx512vbmi2.
5994 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5995 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5996 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5997 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5998 (ix86_valid_target_attribute_inner_p): Ditto.
5999 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6000 * config/i386/i386.opt (mavx512vbmi2): New option.
6001 * doc/invoke.texi: Add new option.
6003 2017-11-16 Julia Koval <julia.koval@intel.com>
6005 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6006 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6007 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6008 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6009 _mm512_gf2p8mul_epi8): New intrinsics.
6010 * config/i386/i386-builtin-types.def
6011 (V64QI_FTYPE_V64QI_V64QI): New type.
6012 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6013 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6014 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6015 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6016 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6017 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6019 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6021 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6022 explicitly as a stream of bytes.
6024 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6026 * config/rs6000/altivec.h (vec_xst_be): New #define.
6027 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6028 and externalize from *altivec_vperm_<mode>_internal.
6029 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6031 (XL_BE_V8HI): Likewise.
6032 (XL_BE_V4SI): Likewise.
6033 (XL_BE_V4SI): Likewise.
6034 (XL_BE_V2DI): Likewise.
6035 (XL_BE_V4SF): Likewise.
6036 (XL_BE_V2DF): Likewise.
6037 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6038 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6039 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6040 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6041 VSX_BUILTIN_VEC_XST_BE.
6042 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6043 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6045 (altivec_init_builtins): Replace conditional calls to def_builtin
6046 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6047 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6048 with unconditional calls. Remove calls to def_builtin for
6049 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6050 __builtin_vec_xst_be.
6051 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6052 to define_expand, and add alternate RTL generation for P8.
6053 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6054 vsx_ld_elemrev_v8hi.
6055 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6056 add alternate RTL generation for P8.
6057 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6058 vsx_ld_elemrev_v16qi.
6059 (vsx_st_elemrev_v8hi): Convert define_insn
6060 to define_expand, and add alternate RTL generation for P8.
6061 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6062 vsx_st_elemrev_v8hi.
6063 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6064 add alternate RTL generation for P8.
6065 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6066 vsx_st_elemrev_v16qi.
6068 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6071 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6072 TARGET_AVX512ER check.
6073 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6074 neither -mzeroupper nor -mno-zeroupper is used and
6075 TARGET_EMIT_VZEROUPPER is set.
6076 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6077 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6079 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6082 folding of vector compares.
6083 (fold_build_vec_cmp): New helper function.
6084 (fold_compare_helper): New helper function.
6085 (builtin_function_type): Add compare builtins to the list of functions
6086 having unsigned arguments. Cosmetic updates to comment indentation.
6087 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6088 the not+eq combination.
6090 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6092 PR tree-optimization/82726
6093 PR tree-optimization/70754
6094 * tree-predcom.c (order_drefs_by_pos): New function.
6095 (combine_chains): Move code setting has_max_use_after to...
6096 (try_combine_chains): ...here. New parameter. Sort combined chains
6097 according to position information.
6098 (tree_predictive_commoning_loop): Update call to above function.
6099 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6101 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6103 PR tree-optimization/82726
6105 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6107 PR tree-optimization/70754
6108 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6109 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6110 combined stmt before it if not NULL.
6111 (combine_chains): Process refs reversely and compute dominance point
6115 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6117 PR tree-optimization/79663
6118 * tree-predcom.c (combine_chains): Process refs in reverse order
6119 only for ZERO length chains, and add explaining comment.
6121 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6123 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6124 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6125 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6127 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6129 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6131 2017-11-15 Martin Liska <mliska@suse.cz>
6133 * tree-cfg.c (pass_warn_function_return::execute):
6134 Compare warn_return_type for greater than zero.
6136 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6140 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6141 to return true on Xeon and not on Xeon Phi.
6142 (ix86_check_avx256_register): Changed to ...
6143 (ix86_check_avx_upper_register): ... this. Add extra check for
6144 VALID_AVX512F_REG_OR_XI_MODE.
6145 (ix86_avx_u128_mode_needed): Changed
6146 ix86_check_avx256_register to ix86_check_avx_upper_register.
6147 (ix86_check_avx256_stores): Changed to ...
6148 (ix86_check_avx_upper_stores): ... this. Changed
6149 ix86_check_avx256_register to ix86_check_avx_upper_register.
6150 (ix86_avx_u128_mode_after): Changed
6151 avx_reg256_found to avx_upper_reg_found. Changed
6152 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6153 (ix86_avx_u128_mode_entry): Changed
6154 ix86_check_avx256_register to ix86_check_avx_upper_register.
6155 (ix86_avx_u128_mode_exit): Ditto.
6156 * config/i386/i386.h: (host_detect_local_cpu): New define.
6158 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6160 * config/arm/xgene1.md (xgene1): Split into automatons
6161 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6162 (xgene1_f_load): Adjust reservations and/or types.
6163 (xgene1_f_store): Likewise.
6164 (xgene1_load_pair): Likewise.
6165 (xgene1_store_pair): Likewise.
6166 (xgene1_fp_load1): Likewise.
6167 (xgene1_load1): Likewise.
6168 (xgene1_store1): Likewise.
6169 (xgene1_move): Likewise.
6170 (xgene1_alu): Likewise.
6171 (xgene1_simd): Likewise.
6172 (xgene1_bfm): Likewise.
6173 (xgene1_neon_load1): Likewise.
6174 (xgene1_neon_store1): Likewise.
6175 (xgene1_neon_logic): Likewise.
6176 (xgene1_neon_st1): Likewise.
6177 (xgene1_neon_ld1r): Likewise.
6178 (xgene1_alu_cond): Added.
6179 (xgene1_shift_reg): Likwise.
6180 (xgene1_bfx): Likewise.
6181 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6183 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6186 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6188 (can_widen_mult_without_libcall): New function.
6189 (expand_mul_overflow): If only checking unsigned mul overflow,
6190 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6191 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6192 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6193 (expand_DIVMOD): Formatting fix.
6194 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6195 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6196 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6198 PR tree-optimization/82977
6199 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6200 constructed temporary to strlen_to_stridx.put.
6202 2017-11-15 Martin Liska <mliska@suse.cz>
6204 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6205 * configure: Regenerate.
6207 2017-11-15 Martin Liska <mliska@suse.cz>
6210 * config/sh/sh-mem.cc: Use proper probability for
6213 2017-11-14 Jeff Law <law@redhat.com>
6215 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6216 the red zone for stack_clash_protection_final_dynamic_probe targets
6217 when the total dynamic stack size is zero bytes.
6219 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6220 blocks is post order.
6222 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6224 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6225 * final.c (rest_of_clean_state): Set it for the
6226 -fcompare-debug dump.
6227 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6228 class when TDF_COMPARE_DEBUG is set.
6230 * dwarf2out.c (gen_producer_string): Discard
6233 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6236 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6237 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6238 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6239 (__TGMATH_REAL_2_3): Remove macros.
6240 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6241 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6243 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6245 (remquo): Define using __TGMATH_REAL_3.
6247 2017-11-14 Jeff Law <law@redhat.com>
6249 * vr-values.c: New file with contents extracted from tree-vrp.c.
6250 * Makefile.in (OBJS): Add vr-values.o
6251 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6252 (set_value_range, set_and_canonicalize_value_range): Likewise.
6253 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6254 (value_range_constant_singleton, symbolic_range_p): Likewise.
6255 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6256 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6257 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6258 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6259 (find_case_label_range, find_case_label_index): Likewise.
6260 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6261 (range_int_cst_singleton_p, value_inside_range): Likewise.
6262 (get_single_symbol): Likewise.
6263 (switch_update): Move structure definition here.
6264 (to_remove_edges, to_update_switch_stmts): Provide externs.
6265 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6266 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6267 (vrp_val_is_min, set_value_range): Likewise.
6268 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6269 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6270 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6271 (range_is_nonnull, range_int_cst_p): Likewwise.
6272 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6273 (get_single_symbol, operand_less_p): Likewise
6274 (compare_values_warnv, compare_values): Likewise.
6275 (value_inside_range, value_range_constant_singleton): Likewise.
6276 (zero_nonzero_bitgs_from_vr): Likewise.
6277 (extract_range_from_binary_expr_1): Likewise.
6278 (overflow_comparison_p): Likewise.
6279 (to_remove_edges, to_update_switch_stmts): Likewise.
6280 (find_case_label-index, find_case_label_range): Likewise.
6281 (switch_update, set_value_range_to_nonnegative): Remove.
6282 (set_value_range_to_truthvalue): Likewise.
6283 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6284 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6285 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6286 (find_case_label_ranges, test_for_singularity): Likewise.
6287 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6288 (x_vr_values): Move to its remaining use site.
6290 2017-11-10 Jeff Law <law@redhat.com>
6292 * vr-values.h (VR_INITIALIZER): Move #define here.
6293 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6294 * Makefile.in (OBJS): Add tree-evrp.o
6295 * tree-vrp.h (assert_info): Move structure definition here.
6296 (set_value_range_to_varying): Prototype.
6297 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6298 (infer_value_range, register_edge_assert_for): Likewise.
6299 (stmt_interesting_for_vrp): Likewise.
6300 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6301 (set_value_range_to_varying): No longer static.
6302 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6303 (infer_value_range, register_edge_assert_for): Likewise.
6305 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6307 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6308 generate the XXBRD instruction.
6310 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6311 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6312 as being compatible if -mabi=ieeelongdouble.
6313 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6314 to setup float128 built-ins with hardware support.
6315 (BU_FLOAT128_HW_2): Likewise.
6316 (BU_FLOAT128_HW_3): Likewise.
6317 (BU_FLOAT128_HW_VSX_1): Likewise.
6318 (BU_FLOAT128_HW_VSX_2): Likewise.
6319 (scalar_extract_expq): Change float128 built-in functions to
6320 accommodate having both KFmode and TFmode functions. Use the
6321 KFmode variant as the default.
6322 (scalar_extract_sigq): Likewise.
6323 (scalar_test_neg_qp): Likewise.
6324 (scalar_insert_exp_q): Likewise.
6325 (scalar_insert_exp_qp): Likewise.
6326 (scalar_test_data_class_qp): Likewise.
6327 (sqrtf128_round_to_odd): Delete processing the round to odd
6328 built-in functions as special built-in functions, and define them
6329 as float128 built-ins. Use the KFmode variant as the default.
6330 (truncf128_round_to_odd): Likewise.
6331 (addf128_round_to_odd): Likewise.
6332 (subf128_round_to_odd): Likewise.
6333 (mulf128_round_to_odd): Likewise.
6334 (divf128_round_to_odd): Likewise.
6335 (fmaf128_round_to_odd): Likewise.
6336 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6337 support for KFmode and TFmode xststdcqp calls.
6338 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6339 point, switch the built-in handlers for the get/set float128
6340 exponent, get float128 mantissa, float128 test built-ins, and the
6341 float128 round to odd built-in functions. Eliminate creating the
6342 float128 round to odd built-in functions as special built-ins.
6343 (rs6000_init_builtins): Eliminate special creation of the float128
6344 round to odd built-in functions.
6345 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6346 function insns to support both TFmode and KFmode variants.
6347 (xsxsigqp_<mode>): Likewise.
6348 (xsiexpqpf_<mode>): Likewise.
6349 (xsiexpqp_<mode>): Likewise.
6350 (xststdcqp_<mode>): Likewise.
6351 (xststdcnegqp_<mode>): Likewise.
6352 (xststdcqp_<mode>): Likewise.
6354 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6356 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6357 unused path_in_freq_ptr parameter.
6358 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6360 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6362 * ipa-inline.c (edge_badness): Dump sreal frequency.
6363 (compute_inlined_call_time): Match natural implementaiton ...
6364 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6365 forgotten division by CGRAPH_FREQ_BASE.
6367 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6369 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6370 Solaris 11. Update comment.
6371 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6373 * config/sol2.h (STARTFILE_SPEC): Likewise.
6374 * configure: Regenerate.
6376 2017-11-14 Carl Love <cel@us.ibm.com>
6378 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6379 le_ and be_ prefixes to swap* variables. Remove
6380 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6383 2017-11-14 Jason Merrill <jason@redhat.com>
6385 Support GTY((cache)) on hash_map.
6386 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6387 (ggc_cache_remove): Override it instead of ggc_mx.
6388 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6389 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6390 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6391 (simple_cache_map_traits): Override maybe_mx.
6392 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6393 (hash_map): Friend gt_cleare_cache.
6394 (gt_cleare_cache): New.
6395 * tree.h (tree_cache_traits): New hash_map traits class.
6396 (tree_cache_map): New typedef.
6398 2017-11-14 Richard Biener <rguenther@suse.de>
6400 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6401 paramter and handling.
6402 (cleanup_control_flow_bb): Likewise.
6403 (cleanup_control_flow_pre): New helper performing a DFS walk
6404 to call cleanup_control_flow_bb in PRE order.
6405 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6406 via cleanup_control_flow_pre.
6408 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6410 * config/aarch64/aarch64-simd.md
6411 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6412 (*aarch64_simd_bsl<mode>_alt): Likewise.
6413 (aarch64_simd_bsldi_internal): New.
6414 (aarch64_simd_bsldi_alt): Likewise.
6416 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6418 * tracer.c (better_p): Do not compare frequencies.
6419 * reg-stack.c (better_edge): Likewise.
6420 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6423 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6425 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6426 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6427 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6428 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6429 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6430 * predict.c (maybe_hot_frequency_p): Inline to ...
6431 (maybe_hot_count_p): ... here; rewrite to counts.
6432 (counts_to_freqs): Rename to ...
6433 (update_max_bb_count): ... this one.
6434 (expensive_function_p): Use counts.
6435 (estimate_bb_frequencies): Update.
6436 (rebuild_frequencies): Update.
6437 * predict.h (counts_to_freqs): Rename to ...
6438 (update_max_bb_count): ... this one.
6439 * profile.c (compute_branch_probabilities): Add debug info
6440 * tree-inline.c (expand_call_inline): Update debug info.
6441 (optimize_inline_calls): Use update_max_bb_count..
6442 (tree_function_versioning): Use update_max_bb_count..
6443 * value-prof.c (gimple_value_profile_transformations):
6444 Do not use update_max_bb_count.
6446 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6448 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6449 always use frequencies.
6451 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6453 * bb-reorder.c: Remove frequencies from comments.
6454 (better_edge_p): Use profile counts.
6455 (find_traces): Dump profile counts.
6456 (rotate_loop): Use profile counts.
6457 (find_traces_1_round): Likewise.
6458 (connect_better_edge_p): Use counts instead of probabilities for
6460 (copy_bb_p): Drop early check for non-0 frequency.
6461 (sanitize_hot_paths): Update comments.
6463 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6465 * ipa-split.c (struct split_point): Add count.
6466 (consider_split): Do not compute incoming frequency; compute incoming
6467 count and store it to split_point.
6468 (split_function): Set count of the call to split part correctly.
6470 2017-11-13 Carl Love <cel@us.ibm.com>
6472 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6474 2017-11-13 Tom Tromey <tom@tromey.com>
6476 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6478 2017-11-13 Carl Love <cel@us.ibm.com>
6480 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6481 Add support for builtins:
6482 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6483 vector {un,}signed {char,int,short},
6484 vector {un,}signed {char,int,short}) arguments.
6485 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6486 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6487 Add BU_P9V_AV_2 expansions for the builtins.
6488 * config/rs6000/altivec.h (vec_first_match_index,
6489 vec_first_mismatch_index, vec_first_match_or_eos_index,
6490 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6491 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6492 new extern declaration.
6493 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6494 * config/rs6000/vsx.md (first_match_index_<mode>,
6495 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6496 first_mismatch_or_eos_index_<mode>): Add define expand.
6497 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6498 * doc/extend.texi: Update the built-in documenation file for the new
6501 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6503 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6504 into the min/max operations for _Float<N> and _Float<N>X types.
6506 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6509 * dwarf2out.c (do_eh_frame): New static variable.
6510 (dwarf2out_begin_prologue): Set it.
6511 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6513 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6515 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6517 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6519 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6522 * debug.h (dwarf2out_do_eh_frame): Declare.
6523 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6524 (dwarf2out_do_frame): Use it.
6525 (dwarf2out_do_cfi_asm): Likewise.
6526 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6527 (dwarf2out_assembly_start): Likewise.
6528 (dwarf2out_begin_prologue): Fix comment.
6529 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6530 if the target needs either debug or unwind DWARF2 info.
6531 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6532 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6533 -ftrapping-math, -ftrapv and -fwrapv.
6535 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6537 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6538 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6539 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6540 (estimate_edge_size_and_time): Likewise.
6541 (ipa_merge_fn_summary_after_inlining): Likewise.
6542 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6543 (compute_uninlined_call_time): Use sreal_frequency.
6544 (compute_inlined_call_time): Likewise.
6545 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6546 * profile-count.c: Include sreal.h.
6547 (profile_count::to_sreal_scale): New.
6548 * profile-count.h: Forward declare sreal.
6549 (profile_count::to_sreal_scale): Declare.
6551 2017-11-13 Nathan Sidwell <nathan@acm.org>
6553 * diagnostic.c (maybe_line_and_column): New.
6554 (diagnostic_get_location_text): Use it.
6555 (diagnostic_report_current_module): Likewise.
6556 (test_diagnostic_get_location_text): Add tests.
6558 2017-11-13 Luis Machado <luis.machado@linaro.org>
6560 * doc/md.texi (Specifying processor pipeline description): Fix
6561 incorrect latency for the div instruction example.
6563 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6565 PR tree-optimization/78821
6566 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6567 that bit_not_p is the same.
6568 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6569 (split_group): Count precisely bit_not_p bits in each statement.
6570 (invert_op): New function.
6571 (imm_store_chain_info::output_merged_store): Use invert_op to
6572 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6573 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6575 2017-11-13 Martin Liska <mliska@suse.cz>
6577 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6578 (struct source_info): Likewise.
6579 (add_branch_counts): Likewise.
6580 (add_line_counts): Likewise.
6581 (function_summary): Likewise.
6582 (output_intermediate_line): Likewise.
6583 (generate_results): Likewise.
6585 2017-11-13 Martin Liska <mliska@suse.cz>
6587 * gcov.c (struct block_info): Remove typedef for block_t.
6588 (struct line_info): Likewise.
6589 (line_info::has_block): Likewise.
6590 (EXIT_BLOCK): Likewise.
6591 (unblock): Likewise.
6592 (circuit): Likewise.
6593 (get_cycles_count): Likewise.
6594 (process_file): Likewise.
6595 (read_graph_file): Likewise.
6596 (solve_flow_graph): Likewise.
6597 (find_exception_blocks): Likewise.
6598 (add_line_counts): Likewise.
6599 (accumulate_line_info): Likewise.
6600 (output_line_details): Likewise.
6602 2017-11-13 Martin Liska <mliska@suse.cz>
6604 * gcov.c (struct arc_info): Remove typedef for arc_t.
6605 (struct line_info): Likewise.
6606 (add_branch_counts): Likewise.
6607 (output_branch_count): Likewise.
6608 (function_info::~function_info): Likewise.
6609 (circuit): Likewise.
6610 (output_intermediate_line): Likewise.
6611 (read_graph_file): Likewise.
6612 (solve_flow_graph): Likewise.
6613 (find_exception_blocks): Likewise.
6614 (add_line_counts): Likewise.
6615 (accumulate_line_info): Likewise.
6616 (output_line_details): Likewise.
6617 (output_function_details): Likewise.
6619 2017-11-13 Martin Liska <mliska@suse.cz>
6621 * gcov.c (struct function_info): Remove typedef for function_t.
6622 (struct source_info): Likewise.
6623 (source_info::get_functions_at_location): Likewise.
6624 (solve_flow_graph): Likewise.
6625 (find_exception_blocks): Likewise.
6626 (add_line_counts): Likewise.
6627 (output_intermediate_file): Likewise.
6628 (process_file): Likewise.
6629 (generate_results): Likewise.
6630 (release_structures): Likewise.
6631 (read_graph_file): Likewise.
6632 (read_count_file): Likewise.
6633 (accumulate_line_counts): Likewise.
6634 (output_lines): Likewise.
6636 2017-11-13 Martin Liska <mliska@suse.cz>
6638 * gcov.c (function_info::function_info): Remove num_counts
6639 and add vector<gcov_type>.
6640 (function_info::~function_info): Use the vector.
6641 (process_file): Likewise.
6642 (read_graph_file): Likewise.
6643 (read_count_file): Likewise.
6644 (solve_flow_graph): Likewise.
6646 2017-11-13 Martin Liska <mliska@suse.cz>
6648 * gcov.c (function_info::is_artificial): New function.
6649 (process_file): Erase all artificial early.
6650 (generate_results): Skip as all artificial are already
6653 2017-11-13 Martin Liska <mliska@suse.cz>
6655 * gcov.c (read_graph_file): Store to global vector of functions.
6656 (read_count_file): Iterate the vector.
6657 (process_file): Likewise.
6658 (generate_results): Likewise.
6659 (release_structures): Likewise.
6661 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6663 PR tree-optimization/82954
6664 * gimple-ssa-store-merging.c
6665 (imm_store_chain_info::coalesce_immediate_stores): If
6666 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6668 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6670 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6671 Upddate call to ENDIAN_LANE_N.
6672 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6673 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6674 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6675 and use aarch64_endian_lane_rtx.
6676 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6678 2017-11-12 Tom de Vries <tom@codesourcery.com>
6680 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6682 2017-11-12 Tom de Vries <tom@codesourcery.com>
6684 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6686 2017-11-12 Tom de Vries <tom@codesourcery.com>
6688 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6690 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6691 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6692 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6694 2017-11-11 Martin Sebor <msebor@redhat.com>
6697 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6700 * prefic.c (translate_name): Replace strncpy with memcpy to
6701 avoid -Wstringop-truncation.
6703 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6705 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6707 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6709 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6711 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6713 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6716 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6718 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6719 enable generating XXBRH if the value is in a vector register.
6720 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6721 value is in a vector register.
6722 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6723 register to register bswap64's instead of doing the GPR sequence
6724 used on previous machines.
6725 (bswapdi2_xxbrd): New insn.
6726 (bswapdi2_reg): Disallow on ISA 3.0.
6727 (register to register bswap64 splitter): Do not split the insn on
6728 ISA 3.0 systems that use XXBRD.
6730 2017-11-10 Martin Sebor <msebor@redhat.com>
6733 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6735 (find_subframework_file): Same.
6737 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6739 * auto-profile.c (afdo_indirect_call): Drop frequency.
6740 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6741 (cgraph_node::create_edge): Drop frequency argument.
6742 (cgraph_node::create_indirect_edge): Drop frequency argument.
6743 (cgraph_edge::make_speculative): Drop frequency arguments.
6744 (cgraph_edge::resolve_speculation): Do not update frequencies
6745 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6746 (cgraph_node::dump): Check consistency in IPA mode.
6747 (cgraph_edge::maybe_hot_p): Use IPA counter.
6748 (cgraph_edge::verify_count_and_frequency): Rename to ...
6749 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6750 (cgraph_node::verify_node): Update.
6751 * cgraph.h (struct cgraph_edge): Drop frequency.
6752 (cgraph_edge::frequency): New function.
6753 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6755 (cgraph_edge::rebuild_edges): Likewise.
6756 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6757 (duplicate_thunk_for_node): Do not pass frequency.
6758 (cgraph_node::create_clone): Scale only counts.
6759 (cgraph_node::create_virtual_clone): Do not pass frequency.
6760 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6761 (cgraph_node::create_version_clone): Do not pass frequency.
6762 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6763 (cgraph_node::expand_thunk): Do not pass frequency.
6764 (cgraph_node::create_wrapper): Do not pass frequency.
6765 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6767 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6768 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6769 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6770 (gather_caller_stats): Use frequency function.
6771 (ipcp_cloning_candidate_p): Use frequency function.
6772 (ipcp_propagate_stage): Use frequency function.
6773 (get_info_about_necessary_edges): Use frequency function.
6774 (update_profiling_info): Update only IPA profile.
6775 (update_specialized_profile): Use frequency functoin.
6776 (perhaps_add_new_callers): Update only IPA profile.
6777 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6778 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6779 (dump_ipa_call_summary): Use frequency function.
6780 (estimate_edge_size_and_time): Use frequency function.
6781 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6782 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6783 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6784 (update_noncloned_counts): ... ths one; scale counts only.
6785 (clone_inlined_nodes): Do not scale frequency.
6786 (inline_call): Do not pass frequency.
6787 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6788 (compute_inlined_call_time): Use IPA profile.
6789 (want_inline_small_function_p): Use IPA profile.
6790 (want_inline_self_recursive_call_p): Use IPA profile.
6791 (edge_badness): Use IPA profile.
6792 (lookup_recursive_calls): Use IPA profile.
6793 (recursive_inlining): Do not pass frequency.
6794 (resolve_noninline_speculation): Do not update frequency.
6795 (inline_small_functions): Collect max of IPA profile.
6796 (dump_overall_stats): Dump IPA porfile.
6797 (dump_inline_stats): Dump IPA porfile.
6798 (ipa_inline): Collect IPA stats.
6799 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6800 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6801 (ipa_propagate_frequency): Use frequency function.
6802 (ipa_profile): Cleanup.
6803 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6804 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6805 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6806 (input_node): Do not stream frequency.
6807 (input_edge): Do not stream frequency.
6808 (merge_profile_summaries): Scale only IPA profiles.
6809 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6810 * predict.c (drop_profile): Do not recompute frequency.
6811 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6812 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6813 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6814 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6815 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6816 * tree-inline.c (copy_bb): Do not scale frequency.
6817 (expand_call_inline): Do not scale frequency.
6818 (tree_function_versioning): Do not scale frequency.
6819 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6821 2017-11-10 Julia Koval <julia.koval@intel.com>
6823 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6824 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6825 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6826 (_mm256_maskz_gf2p8affine_epi64_epi8)
6827 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6828 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6829 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6830 (__builtin_ia32_vgf2p8affineqb_v32qi)
6831 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6832 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6834 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6838 (arm_option_override): Refactor.
6839 (arm_option_reconfigure_globals): New.
6840 (arm_options_perform_arch_sanity_checks): New.
6841 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6843 (arm_options_perform_arch_sanity_checks): Likewise
6845 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6847 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6848 (power9-qpmul): New.
6849 * rs6000/rs6000.md ("type" attr): Add qmul.
6850 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6851 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6852 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6854 2017-11-10 Martin Sebor <msebor@redhat.com>
6857 * builtins.c (compute_objsize): Handle arrays that
6858 compute_builtin_object_size likes to fail for. Make extern.
6859 * builtins.h (compute_objsize): Declare.
6860 (check_strncpy_sizes): New function.
6861 (expand_builtin_strncpy): Call check_strncpy_sizes.
6862 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6863 -Wstringop-truncation.
6864 (gimple_fold_builtin_strncat): Same.
6865 * gimple.c (gimple_build_call_from_tree): Set call location.
6866 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6867 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6868 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6869 (handle_builtin_strlen): Use strlen_to_stridx.
6870 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6872 Use strlen_to_stridx.
6873 (pass_strlen::execute): Release strlen_to_stridx.
6874 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6875 (-Wstringop-truncation): Document new option.
6877 2017-11-10 Martin Liska <mliska@suse.cz>
6879 PR gcov-profile/82702
6880 * gcov.c (main): Handle intermediate files in a different
6882 (get_gcov_intermediate_filename): New function.
6883 (output_gcov_file): Remove support of intermediate files.
6884 (generate_results): Allocate intermediate file.
6885 (release_structures): Clean-up properly fn_end.
6886 (output_intermediate_file): Start iterating with line 1.
6888 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6890 PR tree-optimization/82929
6891 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6892 ops_swapped_p non-static data member.
6893 (store_immediate_info::store_immediate_info): Clear it.
6894 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6895 ops set ops_swapped_p.
6896 (count_multiple_uses): Handle ops_swapped_p.
6898 2017-11-10 Martin Liska <mliska@suse.cz>
6900 * coverage.c (coverage_init): Stream information about
6901 support of has_unexecuted_blocks.
6902 * doc/gcov.texi: Document that.
6903 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6904 * gcov.c (read_graph_file): Likewise.
6905 (output_line_beginning): Fix a small issue with
6908 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6910 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6911 reference of trivial component.
6913 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6916 * gimple-ssa-store-merging.c
6917 (pass_store_merging::terminate_all_aliasing_chains): For
6918 gimple_store_p stmts also call refs_output_dependent_p.
6920 PR rtl-optimization/82913
6921 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6924 2017-11-09 Jeff Law <law@redhat.com>
6926 * vr-values.h: New file with vr_values class.
6927 * tree-vrp.c: Include vr-values.h
6928 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6929 data objects into the vr_values class.
6930 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6931 (get_value_range): Make it a member function within vr_values class.
6932 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6933 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6934 (op_with_constant_singleton_value_range): Likewise.
6935 (extract_range_for_var_from_comparison_expr): Likewise.
6936 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6937 (extract_range_from_binary_expr): Likewise.
6938 (extract_range_from_unary_expr): Likewise.
6939 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6940 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6941 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6942 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6943 (compare_name_with_value, compare_names): Likewise.
6944 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6945 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6946 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6947 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6948 (extract_range_from_phi_node): Likewise.
6949 (simplify_truth_ops_using_ranges): Likewise.
6950 (simplify_div_or_mod_using_ranges): Likewise.
6951 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6952 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6953 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6954 (simplify_float_conversion_using_ranges): Likewise.
6955 (simplify_internal_call_using_ranges): Likewise.
6956 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6957 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6958 poitner via x_vr_values for calls into gimple folder.
6959 (vrp_initialize_lattice): Make this the vr_values ctor.
6960 (vrp_free_lattice): Make this the vr_values dtor.
6961 (set_vr_value): New function.
6962 (class vrp_prop): Add vr_values data member. Add various member
6963 functions as well as member functions that delegate to vr_values.
6964 (check_array_ref): Make a member function within vrp_prop class.
6965 (search_for_addr_array, vrp_initialize): Likewise.
6966 (vrp_finalize): Likewise. Revamp to avoid direct access to
6967 vr_value, values_propagated, etc.
6968 (check_array_bounds): Extract vrp_prop class instance pointer from
6969 walk info structure. Use it to call member functions.
6970 (check_all_array_refs): Make a member function within vrp_prop class.
6971 Smuggle class instance pointer via walk info structure.
6972 (x_vr_values): New local static.
6973 (vrp_valueize): Use x_vr_values to get class instance.
6974 (vr_valueize_1): Likewise.
6975 (class vrp_folder): Add vr_values data member. Add various member
6976 functions as well as member functions that delegate to vr_values.
6977 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6978 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6979 class instance from vr_values. Use it to call member functions.
6980 (vrp_dom_walker): Add vr_values data member.
6981 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6982 instance via x_vr_values.
6983 (identify_jump_threads): Accept vr_values as argument. Store
6984 it into the walker structure.
6985 (evrp_dom_walker): Add vr_values class data member. Add various
6987 (evrp_dom_walker::try_find_new_range): Use vr_values data
6988 member to access the memory allocator.
6989 (evrp_dom_walker::before_dom_children): Store vr_values class
6990 instance into the vrp_folder class.
6991 (evrp_dom_walker::push_value_range): Rework to avoid direct
6992 access to num_vr_values and vr_value.
6993 (evrp_dom_walker::pop_value_range): Likewise.
6994 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6995 Use vr_values to get to dump_all_value_ranges member function.
6996 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6997 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6998 Pass vr_values class instance down to identify_jump_threads.
6999 Remove call to vrp_free_lattice.
7000 (debug_all_value_ranges): Remove.
7002 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7003 (vrp_folder): Likewise.
7005 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7006 Get it from the existing bitmap instead.
7007 (vrp_intersect_ranges_1): Likewise.
7009 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7011 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7013 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7014 set bit_not_p to it.
7015 (imm_store_chain_info::coalesce_immediate_stores): Break group
7016 if bit_not_p is different.
7017 (count_multiple_uses, split_group,
7018 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7019 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7020 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7021 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7022 to store_immediate_info ctor.
7024 2017-11-09 Jim Wilson <jimw@sifive.com>
7026 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7027 (scan_prog_file): Likewise.
7029 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7031 * bb-reorder.c (max_entry_frequency): Remove.
7032 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7033 connect_traces, push_to_next_round_p): Remove prototypes.
7034 (find_traces_1_round): Use counts only.
7035 (push_to_next_round_p): Likewise.
7036 (find_traces): Likewise.
7037 (rotate_loop): Likewise.
7038 (find_traces_1_round): Likewise.
7039 (connect_traces): Likewise.
7040 (edge_order): Likewise.
7042 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7044 * config/arm/arm.c (output_return_instruction): Add comments to
7045 indicate requirement for cmse_nonsecure_entry return to account
7046 for the size of clearing instruction output here.
7047 (thumb_exit): Likewise.
7048 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7049 return in hardfloat mode.
7051 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7053 * config/rs6000/rs6000.c (machine_function): Add a bool,
7054 "toc_is_wrapped_separately".
7055 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7056 if it wasn't explicitly set or unset, we are optimizing for speed, and
7057 doing separate shrink-wrapping.
7058 (rs6000_get_separate_components): Enable the TOC component if
7059 saving the TOC register in the prologue.
7060 (rs6000_components_for_bb): Handle the TOC component.
7061 (rs6000_emit_prologue_components): Store the TOC register where needed.
7062 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7063 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7065 2017-11-09 Martin Jambor <mjambor@suse.cz>
7067 * ipa-param-manipulation.c: New file.
7068 * ipa-param-manipulation.h: Likewise.
7069 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7070 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7071 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7072 (ipa_parm_adjustment): Likewise.
7073 (ipa_parm_adjustment_vec): Likewise.
7074 (ipa_get_vector_of_formal_parms): Moved declaration to
7075 ipa-param-manipulation.h.
7076 (ipa_get_vector_of_formal_parm_types): Likewise.
7077 (ipa_modify_formal_parameters): Likewise.
7078 (ipa_modify_call_arguments): Likewise.
7079 (ipa_combine_adjustments): Likewise.
7080 (ipa_dump_param_adjustments): Likewise.
7081 (ipa_modify_expr): Likewise.
7082 (ipa_get_adjustment_candidate): Likewise.
7083 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7084 ipa-param-manipulation.c.
7085 (ipa_get_vector_of_formal_parm_types): Likewise.
7086 (ipa_modify_formal_parameters): Likewise.
7087 (ipa_modify_call_arguments): Likewise.
7088 (ipa_modify_expr): Likewise.
7089 (get_ssa_base_param): Likewise.
7090 (ipa_get_adjustment_candidate): Likewise.
7091 (index_in_adjustments_multiple_times_p): Likewise.
7092 (ipa_combine_adjustments): Likewise.
7093 (ipa_dump_param_adjustments): Likewise.
7094 * tree-sra.c: Also include ipa-param-manipulation.h
7095 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7098 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7099 Alan Hayward <alan.hayward@arm.com>
7100 David Sherwood <david.sherwood@arm.com>
7102 * doc/sourcebuild.texi (vect_masked_store): Document.
7104 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7105 Alan Hayward <alan.hayward@arm.com>
7106 David Sherwood <david.sherwood@arm.com>
7108 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7110 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7111 Alan Hayward <alan.hayward@arm.com>
7112 David Sherwood <david.sherwood@arm.com>
7114 * doc/sourcebuild.texi (vect_variable_length): Document.
7116 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7117 Alan Hayward <alan.hayward@arm.com>
7118 David Sherwood <david.sherwood@arm.com>
7120 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7122 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7123 Alan Hayward <alan.hayward@arm.com>
7124 David Sherwood <david.sherwood@arm.com>
7126 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7128 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7129 Alan Hayward <alan.hayward@arm.com>
7130 David Sherwood <david.sherwood@arm.com>
7132 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7133 previously undocumented selectors.
7134 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7136 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7138 * doc/rtl.texi (const_vector): Say that elements can be
7139 const_wide_ints too.
7140 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7141 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7142 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7143 * optabs.c (expand_vector_broadcast): Likewise.
7145 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7146 Alan Hayward <alan.hayward@arm.com>
7147 David Sherwood <david.sherwood@arm.com>
7149 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7150 scaled index even if the unscaled address was invalid.
7151 Don't increase the complexity of using a scale in that case.
7153 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7154 Alan Hayward <alan.hayward@arm.com>
7155 David Sherwood <david.sherwood@arm.com>
7157 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7158 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7159 UNITS_PER_WORD bytes.
7160 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7161 based on REGMODE_NATURAL_SIZE of the inner mode.
7162 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7163 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7164 * expmed.c (lowpart_bit_field_p): Divide the value up into
7165 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7166 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7167 whether something is likely to occupy more than one register.
7169 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7172 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7173 profile_count::adjust_for_ipa_scaling.
7174 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7175 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7177 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7179 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7181 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7182 (split_group): Add total_orig and total_new arguments, estimate the
7183 number of statements related to the store group without store merging
7184 and with store merging.
7185 (imm_store_chain_info::output_merged_store): Adjust split_group
7186 callers, punt if estimated number of statements with store merging
7187 is not smaller than estimated number of statements without it.
7189 (handled_load): Remove has_single_use checks.
7190 (pass_store_merging::process_store): Likewise.
7192 2017-11-09 Richard Biener <rguenther@suse.de>
7194 PR tree-optimization/82902
7195 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7197 2017-11-09 Martin Liska <mliska@suse.cz>
7200 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7203 2017-11-09 Martin Liska <mliska@suse.cz>
7205 PR tree-optimization/82669
7206 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7208 2017-11-09 Martin Liska <mliska@suse.cz>
7210 PR gcov-profile/48463
7211 * coverage.c (coverage_begin_function): Output also end locus
7212 of a function and information whether the function is
7214 * gcov-dump.c (tag_function): Parse and print the information.
7215 * gcov.c (INCLUDE_MAP): Add include.
7216 (INCLUDE_SET): Likewise.
7217 (struct line_info): Move earlier in the source file because
7218 of vector<line_info> in function_info structure.
7219 (line_info::line_info): Likewise.
7220 (line_info::has_block): Likewise.
7221 (struct source_info): Add new member index.
7222 (source_info::get_functions_at_location): New function.
7223 (function_info::group_line_p): New function.
7224 (output_intermediate_line): New function.
7225 (output_intermediate_file): Use the mentioned function.
7226 (struct function_start): New.
7227 (struct function_start_pair_hash): Likewise.
7228 (process_file): Add code that identifies group functions.
7229 Assign lines either to global or function scope.
7230 (generate_results): Skip artificial functions.
7231 (find_source): Assign index for each source file.
7232 (read_graph_file): Read new flag artificial and end_line.
7233 (add_line_counts): Assign it either to global of function scope.
7234 (accumulate_line_counts): Isolate core of the function to
7235 accumulate_line_info and call it for both function and global
7237 (accumulate_line_info): New function.
7238 (output_line_beginning): Fix GNU coding style.
7239 (print_source_line): New function.
7240 (output_line_details): Likewise.
7241 (output_function_details): Likewise.
7242 (output_lines): Iterate both source (global) scope and function
7244 (struct function_line_start_cmp): New class.
7245 * doc/gcov.texi: Reflect changes in documentation.
7247 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7250 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7251 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7252 and similarly for not instead of neg.
7254 2017-11-08 Andi Kleen <ak@linux.intel.com>
7256 * config/i386/i386.opt: Add -mforce-indirect-call.
7257 * config/i386/predicates.md: Check for flag_force_indirect_call.
7258 * doc/invoke.texi: Document -mforce-indirect-call
7260 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7262 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7264 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7265 config/riscv/riscv.c (predict.h): New include.
7266 (riscv_slow_unaligned_access_p): No longer static.
7267 (riscv_block_move_straight): Add require.
7268 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7270 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7273 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7274 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7275 nonimmediate_operand predicate for operand 1 instead of
7278 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7280 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7282 * config/aarch64/constraints.md (Uml): New constraint.
7283 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7286 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7288 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7289 of two vec_duplicates into a vec_concat.
7291 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7293 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7294 Simplify vec_merge of vec_duplicate and vec_concat.
7295 * config/aarch64/constraints.md (Utq): New constraint.
7296 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7299 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7301 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7302 Simplify vec_merge of vec_duplicate and const_vector.
7303 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7305 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7306 mode iterator. Update predicate on operand 1 to
7307 handle non-const_vec constants. Delete constraints.
7308 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7310 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7312 PR tree-optimization/78821
7313 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7315 (store_operand_info::store_operand_info): Initialize it to false.
7316 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7317 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7318 store in the group, and if chain_info is non-NULL, to ignore altogether
7320 (compatible_load_p): Fail if bit_not_p does not match.
7321 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7322 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7323 (pass_store_merging::process_store): Adjust
7324 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7325 call terminate_all_aliasing_chains newly when adding a store into
7326 a chain with non-NULL chain_info.
7328 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7330 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7332 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7334 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7336 (aarch64_layout_frame): Initialise emit_frame_chain.
7337 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7338 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7340 2017-11-08 Martin Liska <mliska@suse.cz>
7342 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7343 of gimple_call_internal_p.
7345 2017-11-07 Tom de Vries <tom@codesourcery.com>
7347 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7349 2017-11-07 Tom de Vries <tom@codesourcery.com>
7351 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7354 2017-11-08 Martin Liska <mliska@suse.cz>
7357 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7359 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7361 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7363 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7366 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7367 (*cmp<mode>_ccz_1): New insn with $k alternative.
7370 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7371 RTX_COMM_COMPARE as commutative as well.
7372 (ix86_binary_operator_ok): Formatting fix.
7373 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7374 *<code><mode>3<mask_name><round_saeonly_name>,
7375 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7376 *<s>mul<mode>3_highpart<mask_name>,
7377 *vec_widen_umult_even_v16si<mask_name>,
7378 *vec_widen_umult_even_v8si<mask_name>,
7379 *vec_widen_umult_even_v4si<mask_name>,
7380 *vec_widen_smult_even_v16si<mask_name>,
7381 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7382 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7383 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7384 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7385 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7386 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7387 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7388 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7389 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7390 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7391 ix86_binary_operator_ok. Formatting fixes.
7392 (*<plusminus_insn><mode>3<mask_name><round_name>,
7393 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7396 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7398 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7399 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7400 eq and ne if TARGET_ISEL.
7401 (cmp): New code_iterator.
7402 (UNS, UNSU_, UNSIK): New code_attrs.
7403 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7404 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7406 ("eq<mode>3"): ... this.
7407 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7409 ("ne<mode>3"): ... this.
7411 2017-11-07 Julia Koval <julia.koval@intel.com>
7414 * common/config/i386/i386-common.c
7415 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7416 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7417 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7418 * config/i386/i386.opt: Ditto.
7419 * config/i386/i386.c (ix86_target_string): Ditto.
7420 (ix86_option_override_internal): Ditto.
7421 (ix86_init_mpx_builtins): Move MPX to args2.
7422 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7423 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7424 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7425 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7426 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7427 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7428 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7430 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7433 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7434 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7435 (zero-extendsidi peephole2): Remove peephole.
7437 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7440 * stor-layout.c: Include attribs.h.
7441 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7442 explicit lookup of "aligned" attribute.
7444 2017-11-07 Andrew Waterman <andrew@sifive.com>
7446 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7447 (riscv_expand_block_move): Likewise.
7448 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7450 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7451 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7452 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7453 (riscv_adjust_block_mem): Likewise.
7454 (riscv_block_move_loop): Likewise.
7455 (riscv_expand_block_move): Likewise.
7456 * config/riscv/riscv.md (movmemsi): New pattern.
7458 2017-11-07 Michael Clark <michaeljclark@mac.com>
7460 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7461 (MUSL_DYNAMIC_LINKER): Likewise.
7463 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7465 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7466 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7469 2017-11-07 Richard Biener <rguenther@suse.de>
7471 * match.pd: Fix build.
7473 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7474 Jackson Woodruff <jackson.woodruff@arm.com>
7476 PR tree-optimization/71026
7477 * match.pd: Canonicalize negate in division.
7479 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7482 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7484 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7486 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7487 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7489 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7491 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7492 non-scalar integral types.
7493 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7494 (-(A-B), -(~A)): New transformations.
7496 2017-11-07 Tom de Vries <tom@codesourcery.com>
7498 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7499 semicolon after "do {} while (0)".
7500 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7501 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7502 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7503 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7504 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7506 2017-11-07 Tom de Vries <tom@codesourcery.com>
7508 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7509 after "do {} while (0)".
7510 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7511 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7512 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7513 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7514 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7516 2017-11-07 Tom de Vries <tom@codesourcery.com>
7519 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7521 (arm_rtx_costs_internal): Add missing semicolon after
7522 HANDLE_NARROW_SHIFT_ARITH call.
7524 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7526 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7527 disable isel if it was not set explicitly.
7529 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7531 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7532 (add_type_attribute) likewise.
7534 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7536 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7537 of struct ix86_frame.
7538 (ix86_initial_elimination_offset): Likewise.
7539 (ix86_expand_split_stack_prologue): Likewise.
7541 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7543 * tree-vrp.h (enum value_range_type): Update stale comment.
7545 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7546 Alan Hayward <alan.hayward@arm.com>
7547 David Sherwood <david.sherwood@arm.com>
7549 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7550 (aarch64_expand_vec_perm_const): Take the number of units too.
7551 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7552 (aarch64_expand_vec_perm_const): Likewise.
7553 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7554 (vec_perm<mode>): Update accordingly.
7556 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7557 Alan Hayward <alan.hayward@arm.com>
7558 David Sherwood <david.sherwood@arm.com>
7560 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7561 Take the number of units too.
7562 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7563 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7564 but check for a vector mode before rather than after the call.
7565 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7566 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7567 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7568 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7569 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7570 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7571 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7572 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7573 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7574 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7575 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7576 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7577 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7578 (aarch64_sqdmull2_n<mode>): Update accordingly.
7580 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7581 Alan Hayward <alan.hayward@arm.com>
7582 David Sherwood <david.sherwood@arm.com>
7584 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7585 the number of units too.
7586 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7587 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7588 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7589 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7590 (vec_store_lanesxi<mode>): Update accordingly.
7592 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7593 Alan Hayward <alan.hayward@arm.com>
7594 David Sherwood <david.sherwood@arm.com>
7596 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7597 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7598 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7599 of units rather than the mode.
7600 * config/aarch64/iterators.md (nunits): New mode attribute.
7601 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7602 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7603 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7604 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7605 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7606 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7607 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7608 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7609 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7610 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7611 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7612 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7613 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7614 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7615 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7616 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7617 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7618 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7619 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7620 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7621 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7622 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7623 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7624 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7625 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7626 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7627 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7628 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7629 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7630 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7631 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7632 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7633 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7634 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7635 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7636 (aarch64_simd_vec_setv2di): Likewise.
7638 2017-11-06 Carl Love <cel@us.ibm.com>
7640 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7642 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7643 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7645 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7647 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7648 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7649 Add power 8 macro expansions.
7650 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7651 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7652 power 8 instructions. (VSX_XXBR): Add iterator.
7654 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7656 * config/arm/arm.md (predicable_short_it): Change default to "no",
7657 improve documentation, remove uses that are identical to the default.
7658 (enabled_for_depr_it): Rename to enabled_for_short_it.
7659 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7660 Remove default uses.
7661 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7662 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7663 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7664 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7666 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7669 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7670 float128 helper macros, which are no longer used after deleting
7671 the old 'q' built-in functions, and moving the round to odd
7672 built-in functions to being special built-in functions.
7673 (BU_FLOAT128_2): Likewise.
7674 (BU_FLOAT128_1_HW): Likewise.
7675 (BU_FLOAT128_2_HW): Likewise.
7676 (BU_FLOAT128_3_HW): Likewise.
7677 (FABSQ): Delete old 'q' built-in functions.
7678 (COPYSIGNQ): Likewise.
7679 (SQRTF128_ODD): Move round to odd built-in functions to be
7680 special built-in functions, so that we can handle
7681 -mabi=ieeelongdouble.
7682 (TRUNCF128_ODD): Likewise.
7683 (ADDF128_ODD): Likewise.
7684 (SUBF128_ODD): Likewise.
7685 (MULF128_ODD): Likewise.
7686 (DIVF128_ODD): Likewise.
7687 (FMAF128_ODD): Likewise.
7688 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7689 built-in names to 'f128'.
7690 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7691 old 'q' built-in functions, as the machine independent code for
7692 'f128' built-in functions handles this.
7693 (rs6000_expand_builtin): Add expansion for float128 round to odd
7694 functions, keying off on -mabi=ieeelongdouble of whether to use
7695 the KFmode or TFmode variant.
7696 (rs6000_init_builtins): Initialize the _Float128 round to odd
7698 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7699 _Float128 'q' built-in functions are now mapped into the new
7700 'f128' built-in functions.
7702 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7704 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7705 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7707 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7710 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7711 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7712 * toplev.c: Include "ipa-fnsummary.h".
7713 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7715 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7717 PR tree-optimization/82838
7718 * gimple-ssa-store-merging.c
7719 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7720 on a separate gimple_seq which is then appended to seq.
7722 2017-11-06 Jeff Law <law@redhat.com>
7725 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7726 (get_probe_interval): New functions.
7727 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7728 (ix86_adjust_stack_and_probe): Likewise.
7729 (output_adjust_stack_and_probe): Likewise.
7730 (ix86_emit_probe_stack_range): Likewise.
7731 (ix86_expand_prologue): Likewise.
7733 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7735 PR tree-optimization/82816
7736 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7737 if the modes of the two types are the same.
7738 (convert_plusminus_to_widen): Likewise.
7740 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7742 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7744 (usadv16qi): New define_expand.
7745 (usadv8hi): New define_expand.
7747 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7750 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7752 (inline_transform): Likewise.
7753 * predict.c (counts_to_freqs): Remove useless conditional.
7754 * profile-count.h (profile_count::apply_scale): Move sanity check.
7755 * tree-inline.c (copy_bb): Always scale.
7756 (copy_cfg_body): Likewise.
7758 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7761 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7763 (*cmp_ite0): Add enabled_for_depr_it attribute.
7764 (*cmp_ite1): Likewise.
7766 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7768 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7771 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7773 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7774 return the value by pointer.
7775 (extract_range_from_multiplicative_op_1): Update accordingly.
7776 Return as soon as an operation fails.
7778 2017-11-05 Tom de Vries <tom@codesourcery.com>
7781 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7782 (DEF_SANITIZER_BUILTIN): ... here.
7783 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7784 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7786 2017-11-05 Tom de Vries <tom@codesourcery.com>
7789 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7791 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7792 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7793 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7795 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7796 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7797 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7799 2017-11-05 Tom de Vries <tom@codesourcery.com>
7802 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7805 2017-11-04 Michael Clark <michaeljclark@mac.com>
7807 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7808 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7810 (*addsi3_extended): Likewise.
7811 (*addsi3_extended2): Likewise.
7812 (<optab>si3): Likewise.
7813 (<optab>di3): Likewise.
7814 (<optab><mode>3): Likewise.
7815 (<*optabe>si3_internal): Likewise.
7816 (zero_extendqi<SUPERQI:mode>2): Likewise.
7817 (*add<mode>hi3): Likewise.
7818 (*xor<mode>hi3): Likewise.
7819 (<optab>di3): Likewise.
7820 (*<optab>si3_extend): Likewise.
7821 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7822 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7823 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7825 2017-11-04 Andrew Waterman <andrew@sifive.com>
7827 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7828 TARGET_STRICT_ALIGN based upon -mtune argument.
7830 2017-11-04 Andrew Waterman <andrew@sifive.com>
7832 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7834 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7836 * config/i386/i386.c (choose_basereg): Use optional scratch
7837 register and add assertion.
7838 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7839 needed, and don't allocate stack.
7840 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7841 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7842 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7844 2017-11-03 Jeff Law <law@redhat.com>
7846 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7847 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7848 to probe at the start of a noreturn function.
7850 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7852 PR tree-optimization/78821
7853 * gimple-ssa-store-merging.c: Update the file comment.
7854 (MAX_STORE_ALIAS_CHECKS): Define.
7855 (struct store_operand_info): New type.
7856 (store_operand_info::store_operand_info): New constructor.
7857 (struct store_immediate_info): Add rhs_code and ops data members.
7858 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7859 and op1r arguments to the ctor, initialize corresponding data members.
7860 (struct merged_store_group): Add load_align_base and load_align
7862 (merged_store_group::merged_store_group): Initialize them.
7863 (merged_store_group::do_merge): Update them.
7864 (merged_store_group::apply_stores): Pick the constant for
7865 encode_tree_to_bitpos from one of the two operands, or skip
7866 encode_tree_to_bitpos if neither operand is a constant.
7867 (class pass_store_merging): Add process_store method decl. Remove
7868 bool argument from terminate_all_aliasing_chains method decl.
7869 (pass_store_merging::terminate_all_aliasing_chains): Remove
7870 var_offset_p argument and corresponding handling.
7871 (stmts_may_clobber_ref_p): New function.
7872 (compatible_load_p): New function.
7873 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7874 if there is overlap and rhs_code is not INTEGER_CST. For
7875 non-overlapping stores terminate group if rhs is not mergeable.
7876 (get_alias_type_for_stmts): Change first argument from
7877 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7878 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7879 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7881 (get_location_for_stmts): Change first argument from
7882 auto_vec<gimple *> & to vec<gimple *> &.
7883 (struct split_store): Remove orig_stmts data member, add orig_stores.
7884 (split_store::split_store): Create orig_stores rather than orig_stmts.
7885 (find_constituent_stmts): Renamed to ...
7886 (find_constituent_stores): ... this. Change second argument from
7887 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7888 to info structures rather than the statements.
7889 (split_group): Rename ALLOW_UNALIGNED argument to
7890 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7891 it. Adjust find_constituent_stores caller.
7892 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7893 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7894 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7895 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7896 (mem_valid_for_store_merging): New function.
7897 (handled_load): New function.
7898 (pass_store_merging::process_store): New method.
7899 (pass_store_merging::execute): Use process_store method. Adjust
7900 terminate_all_aliasing_chains caller.
7902 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7904 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7905 Return true for more constants, symbols and label references.
7906 (aarch64_valid_floating_const): Remove unused function.
7908 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7911 * config/aarch64/aarch64.c (aarch64_layout_frame):
7912 Undo forcing of LR at bottom of frame.
7914 2017-11-03 Jeff Law <law@redhat.com>
7917 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7918 for int_registers_saved.
7920 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7921 extracted from tree-ssa-dom.c.
7922 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7923 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7924 (record_equivalences_from_incoming_edge): Add additional argument
7925 to single_pred_edge_ignoring_loop_edges call.
7926 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7927 (uncprop_dom_walker::before_dom_children): Add additional argument
7928 to single_pred_edge_ignoring_loop_edges call.
7929 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7930 single_pred_edge_ignoring_loop_edges rather than open coding.
7931 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7933 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7935 * match.pd (-(-A)): Rewrite.
7937 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7939 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7940 (rs6000_emit_int_cmove): New declaration.
7941 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7942 (rs6000_emit_sISEL): Delete.
7943 (rs6000_emit_int_cmove): Make non-static.
7944 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7945 instead of rs6000_emit_sISEL.
7947 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7949 * asan.c (create_cond_insert_point): Maintain profile.
7950 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7952 * basic-block.h (struct basic_block_def): Remove frequency.
7953 (EDGE_FREQUENCY): Use to_frequency
7954 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7956 (find_traces): Update to use to_frequency.
7957 (find_traces_1_round): Likewise; use only IPA counts.
7958 (bb_to_key): Likewise.
7959 (connect_traces): Use IPA counts only.
7960 (copy_bb_p): Update to use to_frequency.
7961 (fix_up_crossing_landing_pad): Likewise.
7962 (sanitize_hot_paths): Likewise.
7963 * bt-load.c (basic_block_freq): Likewise.
7964 * cfg.c (init_flow): Set count_max to uninitialized.
7965 (check_bb_profile): Remove frequencies; check counts.
7966 (dump_bb_info): Do not dump frequencies.
7967 (update_bb_profile_for_threading): Update counts only.
7968 (scale_bbs_frequencies_int): Likewise.
7969 (MAX_SAFE_MULTIPLIER): Remove.
7970 (scale_bbs_frequencies_gcov_type): Update counts only.
7971 (scale_bbs_frequencies_profile_count): Update counts only.
7972 (scale_bbs_frequencies): Update counts only.
7973 * cfg.h (struct control_flow_graph): Add count-max.
7974 (update_bb_profile_for_threading): Update prototype.
7975 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7976 (find_many_sub_basic_blocks): Likewise.
7977 * cfgcleanup.c (try_forward_edges): Likewise.
7978 (try_crossjump_to_edge): Likewise.
7979 * cfgexpand.c (expand_gimple_cond): Likewise.
7980 (expand_gimple_tailcall): Likewise.
7981 (construct_init_block): Likewise.
7982 (construct_exit_block): Likewise.
7983 * cfghooks.c (verify_flow_info): Check consistency of counts.
7984 (dump_bb_for_graph): Do not dump frequencies.
7985 (split_block_1): Do not update frequencies.
7986 (split_edge): Do not update frequencies.
7987 (make_forwarder_block): Do not update frequencies.
7988 (duplicate_block): Do not update frequencies.
7989 (account_profile_record): Do not update frequencies.
7990 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7991 for global heuristics.
7992 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7993 (expected_loop_iterations_unbounded): Use counts only.
7994 * cfgloopmanip.c (scale_loop_profile): Simplify.
7995 (create_empty_loop_on_edge): Simplify
7997 (duplicate_loop_to_header_edge): Simplify
7998 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7999 (update_br_prob_note): Take care of removing note when profile
8001 (relink_block_chain): Do not dump frequency.
8002 (rtl_account_profile_record): Use to_frequency.
8003 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8004 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8005 (cgraph_update_edges_for_call_stmt_node): Likewise.
8006 (cgraph_edge::verify_count_and_frequency): Update.
8007 (cgraph_node::verify_node): Temporarily disable frequency verification.
8008 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8009 to_cgraph_frequency.
8010 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8011 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8013 (cgraph_node::expand_thunk): Update profile.
8014 * except.c (dw2_build_landing_pads): Do not update frequency.
8015 * final.c (compute_alignments): Use to_frequency.
8016 (dump_basic_block_info): Do not dump frequency.
8017 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8018 (dump_gimple_bb_header): Do not dump frequency.
8019 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8021 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8022 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8023 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8024 (init_before_recovery): Do not update frequency.
8025 (sched_create_recovery_edges): Do not update frequency.
8026 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8027 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8028 (ipa_cp_c_finalize): Set max_count to uninitialized.
8029 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8030 (param_change_prob): Use counts.
8031 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8033 * ipa-split.c (consider_split): Use to_frequency.
8034 (split_function): Use to_frequency.
8035 * ira-build.c (loop_compare_func): Likewise.
8036 (mark_loops_for_removal): Likewise.
8037 (mark_all_loops_for_removal): Likewise.
8038 * loop-doloop.c (doloop_modify): Do not update frequency.
8039 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8041 * lto-streamer-in.c (input_function): Update count_max.
8042 * omp-expand.c (expand_omp_taskreg): Update count_max.
8043 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8044 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8045 (maybe_hot_count_p): Use ipa counts only.
8046 (maybe_hot_bb_p): Simplify.
8047 (maybe_hot_edge_p): Simplify.
8048 (probably_never_executed): Do not take frequency argument.
8049 (probably_never_executed_bb_p): Do not pass frequency.
8050 (probably_never_executed_edge_p): Likewise.
8051 (combine_predictions_for_bb): Check that profile is nonzero.
8052 (propagate_freq): Do not set frequency.
8053 (drop_profile): Simplify.
8054 (counts_to_freqs): Simplify.
8055 (expensive_function_p): Use to_frequency.
8056 (propagate_unlikely_bbs_forward): Simplify.
8057 (determine_unlikely_bbs): Simplify.
8058 (estimate_bb_frequencies): Add hack to silence graphite issues.
8059 (compute_function_frequency): Use ipa counts.
8060 (pass_profile::execute): Update.
8061 (rebuild_frequencies): Use counts only.
8062 (force_edge_cold): Use counts only.
8063 * profile-count.c (profile_count::dump): Dump new count types.
8064 (profile_count::differs_from_p): Check compatiblity.
8065 (profile_count::to_frequency): New function.
8066 (profile_count::to_cgraph_frequency): New function.
8067 * profile-count.h (struct function): Declare.
8068 (enum profile_quality): Add profile_guessed_local and
8069 profile_guessed_global0.
8070 (class profile_proability): Decrease number of bits to 29;
8071 update from_reg_br_prob_note and to_reg_br_prob_note.
8072 (class profile_count: Update comment; decrease number of bits
8073 to 61. Check compatibility.
8074 (profile_count::compatible_p): New private member function.
8075 (profile_count::ipa_p): New member function.
8076 (profile_count::operator<): Handle global zero correctly.
8077 (profile_count::operator>): Handle global zero correctly.
8078 (profile_count::operator<=): Handle global zero correctly.
8079 (profile_count::operator>=): Handle global zero correctly.
8080 (profile_count::nonzero_p): New member function.
8081 (profile_count::force_nonzero): New member function.
8082 (profile_count::max): New member function.
8083 (profile_count::apply_scale): Handle IPA scalling.
8084 (profile_count::guessed_local): New member function.
8085 (profile_count::global0): New member function.
8086 (profile_count::ipa): New member function.
8087 (profile_count::to_frequency): Declare.
8088 (profile_count::to_cgraph_frequency): Declare.
8089 * profile.c (OVERLAP_BASE): Delete.
8090 (compute_frequency_overlap): Delete.
8091 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8092 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8093 * sched-ebb.c (rank): Use counts only.
8094 * shrink-wrap.c (handle_simple_exit): Use counts only.
8095 (try_shrink_wrapping): Use counts only.
8096 (place_prologue_for_one_component): Use counts only.
8097 * tracer.c (find_best_predecessor): Use to_frequency.
8098 (find_trace): Use to_frequency.
8099 (tail_duplicate): Use to_frequency.
8100 * trans-mem.c (expand_transaction): Do not update frequency.
8101 * tree-call-cdce.c: Do not update frequency.
8102 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8103 (gimple_merge_blocks): Likewise.
8104 (gimple_split_edge): Likewise.
8105 (gimple_duplicate_sese_region): Likewise.
8106 (gimple_duplicate_sese_tail): Likewise.
8107 (move_sese_region_to_fn): Likewise.
8108 (gimple_account_profile_record): Likewise.
8109 (insert_cond_bb): Likewise.
8110 * tree-complex.c (expand_complex_div_wide): Likewise.
8111 * tree-eh.c (lower_resx): Update profile.
8112 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8114 (initialize_cfun): Do not initialize frequencies
8115 (freqs_to_counts): Delete.
8116 (copy_cfg_body): Ignore count parameter.
8117 (copy_body): Update.
8118 (expand_call_inline): Update count_max.
8119 (optimize_inline_calls): Update count_max.
8120 (tree_function_versioning): Update count_max.
8121 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8122 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8124 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8125 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8126 (try_peel_loop): Likewise.
8127 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8129 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8130 (tree_transform_and_unroll_loop): Do not use frequencies
8131 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8132 Use reliable prediction only.
8133 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8134 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8135 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8136 probability scaling.
8137 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8138 not update frequency
8139 (any_remaining_duplicated_blocks): Likewise.
8140 (update_profile): Likewise.
8141 (estimated_freqs_path): Delete.
8142 (freqs_to_counts_path): Delete.
8143 (clear_counts_path): Delete.
8144 (ssa_fix_duplicate_block_edges): Likewise.
8145 (duplicate_thread_path): Likewise.
8146 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8147 * tree-tailcall.c (decrease_profile): Do not update frequency.
8148 (eliminate_tail_call): Likewise.
8149 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8150 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8151 (optimize_mask_stores): Likewise.
8152 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8153 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8154 (ubsan_expand_ptr_ifn): Update profile.
8155 * value-prof.c (gimple_ic): Simplify.
8156 * value-prof.h (gimple_ic): Update prototype.
8157 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8158 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8160 (want_inline_self_recursive_call_p): Likewise.
8161 (resolve_noninline_speculation): Only cummulate defined counts.
8162 (inline_small_functions): Use nonzero_p.
8163 (ipa_inline): Do not access freed node.
8165 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8167 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8168 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8170 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8172 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8173 non-legitimate address.
8175 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8177 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8178 (*lt0_<mode>di, *lt0_<mode>si): New.
8180 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8182 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8183 TARGET_PAIRED_FLOAT.
8185 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8186 Jim Wilson <jim.wilson@linaro.org>
8188 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8189 * config/aarch64/aarch64-tune.md: Regenerated.
8190 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8191 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8193 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8195 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8196 (arc_expand_prologue): Restore blink for millicode.
8197 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8199 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8202 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8203 gen_vec_duplicate after forcing the scalar into a register.
8205 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8207 * combine (try_combine): Print the insns input to try_combine to the
8210 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8213 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8214 Remove second argument from aarch64_process_target_attr call.
8215 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8217 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8219 (aarch64_handle_attr_arch): Remove second argument.
8220 (aarch64_handle_attr_cpu): Ditto.
8221 (aarch64_handle_attr_tune): Ditto.
8222 (aarch64_handle_attr_isa_flags): Ditto.
8223 (aarch64_process_one_target_attr): Ditto.
8224 (aarch64_process_target_attr): Ditto.
8225 (aarch64_option_valid_attribute_p): Remove second argument.
8226 on aarch64_process_target_attr call.
8228 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8230 * diagnostic.c: Include "selftest-diagnostic.h".
8231 (selftest::assert_location_text): New function.
8232 (selftest::test_diagnostic_get_location_text): New function.
8233 (selftest::diagnostic_c_tests): Call it.
8235 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8237 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8238 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8239 (class selftest::test_diagnostic_context): Move to...
8240 * selftest-diagnostic.c: New file.
8241 * selftest-diagnostic.h: New file.
8243 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8245 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8246 offset range for FT32B.
8247 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8248 * config/ft32/ft32.md: Add TARGET_NOPM.
8249 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8250 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8252 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8254 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8256 2017-11-02 Jeff Law <law@redhat.com>
8258 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8259 virtual keyword on FINAL OVERRIDE members.
8261 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8262 virtuals together. Add virtual destructor.
8263 (substitute_and_fold_engine): Similarly.
8265 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8267 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8269 2017-11-02 Richard Biener <rguenther@suse.de>
8271 PR tree-optimization/82795
8272 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8274 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8276 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8278 (gcc_GAS_CHECK_FEATURE): Remove.
8279 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8280 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8281 * configure: Regenerate.
8283 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8285 * config/arc/arc.c (hwloop_optimize): Account for empty
8288 2017-11-02 Richard Biener <rguenther@suse.de>
8291 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8292 Truncate ARRAY_REF index and element size.
8294 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8296 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8298 2017-11-01 Jeff Law <law@redhat.com>
8300 * tree-ssa-ccp.c (ccp_folder): New class derived from
8301 substitute_and_fold_engine.
8302 (ccp_folder::get_value): New member function.
8303 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8304 (ccp_fold_stmt): Remove prototype.
8305 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8306 * tree-ssa-copy.c (copy_folder): New class derived from
8307 substitute_and_fold_engine.
8308 (copy_folder::get_value): Renamed from get_value.
8309 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8310 * tree-vrp.c (vrp_folder): New class derived from
8311 substitute_and_fold_engine.
8312 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8313 (vrp_folder::get_value): New member function.
8314 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8315 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8316 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8317 provide a class interface to folder/substitute routines.
8318 (ssa_prop_fold_stmt_fn): Remove typedef.
8319 (ssa_prop_get_value_fn): Likewise.
8320 (subsitute_and_fold): Remove prototype.
8321 (replace_uses_in): Likewise.
8322 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8323 Renamed from replace_uses_in. Call the virtual member function
8324 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8325 (substitute_and_fold_dom_walker): Remove initialization of
8326 data member entries for calbacks. Add substitute_and_fold_engine
8327 member and initialize it.
8328 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8329 member functions for get_value, replace_phi_args_in c
8330 replace_uses_in, and fold_stmt calls.
8331 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8332 substitute_and_fold. Remove assert. Update ctor call.
8334 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8335 (ssa_prop_visit_phi_fn): Likewise.
8336 (class ssa_propagation_engine): New class to provide an interface
8338 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8340 (ssa_prop_visit_phi): Likewise.
8341 (ssa_propagation_engine::simulate_stmt): Moved into class.
8342 Call visit_phi/visit_stmt from the class rather than via
8343 file scoped static variables.
8344 (ssa_propagation_engine::simulate_block): Moved into class.
8345 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8346 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8347 set file scoped statics for the visit_stmt/visit_phi callbacks.
8348 * tree-complex.c (complex_propagate): New class derived from
8349 ssa_propagation_engine.
8350 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8351 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8352 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8354 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8355 ssa_propagation_engine.
8356 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8357 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8358 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8359 * tree-ssa-copy.c (copy_prop): New class derived from
8360 ssa_propagation_engine.
8361 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8362 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8363 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8364 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8365 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8366 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8367 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8369 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8371 PR rtl-optimization/82778
8372 PR rtl-optimization/82597
8373 * compare-elim.c (struct comparison): Add in_a_setter field.
8374 (find_comparison_dom_walker::before_dom_children): Remove killed
8375 bitmap and df_simulate_find_defs call, instead walk the defs.
8376 Compute last_setter and initialize in_a_setter. Merge definitions
8377 with first initialization for a few variables.
8378 (try_validate_parallel): Use insn_invalid_p instead of
8379 recog_memoized. Return insn rather than just the pattern.
8380 (try_merge_compare): Fix up comment. Don't uselessly test if
8381 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8383 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8386 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8387 Alan Hayward <alan.hayward@arm.com>
8388 David Sherwood <david.sherwood@arm.com>
8390 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8391 aarch64_hard_regno_nregs to get the number of registers
8394 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8395 Alan Hayward <alan.hayward@arm.com>
8396 David Sherwood <david.sherwood@arm.com>
8398 * config/aarch64/constraints.md (Upl): Rename to...
8400 * config/aarch64/aarch64.md
8401 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8404 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8405 Alan Hayward <alan.hayward@arm.com>
8406 David Sherwood <david.sherwood@arm.com>
8408 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8409 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8413 Alan Hayward <alan.hayward@arm.com>
8414 David Sherwood <david.sherwood@arm.com>
8416 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8417 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8418 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8420 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8421 of a single element.
8422 * config/aarch64/iterators.md: Add a comment above the permute
8423 unspecs to say that they are generated directly by
8424 aarch64_expand_vec_perm_const.
8425 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8427 2017-11-01 Nathan Sidwell <nathan@acm.org>
8429 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8431 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8433 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8434 and medany code models, and describe what they do.
8436 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8438 Revert accidental duplicate:
8440 * combine.c (can_change_dest_mode): Reject changes in
8441 REGMODE_NATURAL_SIZE.
8443 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8445 PR rtl-optimization/64682
8446 PR rtl-optimization/69567
8447 PR rtl-optimization/69737
8448 PR rtl-optimization/82683
8449 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8450 register mentioned in the note, drop the note, unless it came from I3,
8451 in which case it should go to I3 again.
8453 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8455 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8456 and return false if not.
8457 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8459 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8461 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8462 range is known to be empty.
8464 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8465 Alan Hayward <alan.hayward@arm.com>
8466 David Sherwood <david.sherwood@arm.com>
8468 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8469 and CONST_VECTOR_NUNITS instead of computing the number of units from
8470 the byte sizes of the vector and element.
8471 (simplify_binary_operation_1): Likewise.
8472 (simplify_const_binary_operation): Likewise.
8473 (simplify_ternary_operation): Likewise.
8475 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8476 Alan Hayward <alan.hayward@arm.com>
8477 David Sherwood <david.sherwood@arm.com>
8479 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8480 (int_mem_offset): ...this new function.
8481 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8482 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8483 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8486 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8487 Alan Hayward <alan.hayward@arm.com>
8488 David Sherwood <david.sherwood@arm.com>
8490 * lower-subreg.c (interesting_mode_p): New function.
8491 (compute_costs, find_decomposable_subregs, decompose_register)
8492 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8493 (resolve_clobber, dump_choices): Use it.
8495 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8496 Alan Hayward <alan.hayward@arm.com>
8497 David Sherwood <david.sherwood@arm.com>
8499 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8501 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8502 Alan Hayward <alan.hayward@arm.com>
8503 David Sherwood <david.sherwood@arm.com>
8505 * alias.c (find_base_value, find_base_term): Only process integer
8506 truncations. Check the precision rather than the size.
8508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8509 Alan Hayward <alan.hayward@arm.com>
8510 David Sherwood <david.sherwood@arm.com>
8512 * machmode.h (is_narrower_int_mode): New function
8513 * optabs.c (expand_float, expand_fix): Use it.
8514 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8516 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8517 Alan Hayward <alan.hayward@arm.com>
8518 David Sherwood <david.sherwood@arm.com>
8520 * rtl.h (narrower_subreg_mode): New function.
8521 * ira-color.c (update_costs_from_allocno): Use it.
8523 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8524 Alan Hayward <alan.hayward@arm.com>
8525 David Sherwood <david.sherwood@arm.com>
8527 * optabs-query.h (convert_optab_p): New function, split out from...
8528 (convert_optab_handler): ...here.
8529 (widening_optab_handler): Delete.
8530 (find_widening_optab_handler): Remove permit_non_widening parameter.
8531 (find_widening_optab_handler_and_mode): Likewise. Provide an
8532 override that operates on mode class wrappers.
8533 * optabs-query.c (widening_optab_handler): Delete.
8534 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8535 parameter. Assert that the two modes are the same class and that
8536 the "from" mode is narrower than the "to" mode. Use
8537 convert_optab_handler instead of widening_optab_handler.
8538 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8539 instead of widening_optab_handler.
8540 * expr.c (expand_expr_real_2): Update calls to
8541 find_widening_optab_handler.
8542 * optabs.c (expand_widen_pattern_expr): Likewise.
8543 (expand_binop_directly): Take the insn_code as a parameter.
8544 (expand_binop): Only call find_widening_optab_handler for
8545 conversion optabs; use optab_handler otherwise. Update calls
8546 to find_widening_optab_handler and expand_binop_directly.
8547 Use convert_optab_handler instead of widening_optab_handler.
8548 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8549 find_widening_optab_handler and use scalar_mode rather than
8551 (convert_plusminus_to_widen): Likewise.
8553 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8554 Alan Hayward <alan.hayward@arm.com>
8555 David Sherwood <david.sherwood@arm.com>
8557 * machmode.h (fixed_size_mode): New class.
8558 * rtl.h (get_pool_mode): Return fixed_size_mode.
8559 * gengtype.c (main): Add fixed_size_mode.
8560 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8561 (get_raw_arg_mode): Likewise.
8562 * doc/tm.texi: Regenerate.
8563 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8564 * targhooks.c (default_get_reg_raw_mode): Likewise.
8565 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8566 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8567 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8568 (msp430_get_raw_result_mode): Likewise.
8569 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8570 * dbxout.c (dbxout_parms): Require fixed-size modes.
8571 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8572 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8573 * omp-low.c (lower_oacc_reductions): Likewise.
8574 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8575 (simplify_subreg): Update accordingly.
8576 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8577 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8578 that aren't fixed-size.
8579 (get_pool_mode): Return a fixed_size_mode.
8580 (output_constant_pool_2): Take a fixed_size_mode.
8582 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8583 Alan Hayward <alan.hayward@arm.com>
8584 David Sherwood <david.sherwood@arm.com>
8586 * doc/rtl.texi (vec_series): Document.
8587 (const): Say that the operand can be a vec_series.
8588 * rtl.def (VEC_SERIES): New rtx code.
8589 * rtl.h (const_vec_series_p_1): Declare.
8590 (const_vec_series_p): New function.
8591 * emit-rtl.h (gen_const_vec_series): Declare.
8592 (gen_vec_series): Likewise.
8593 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8594 (gen_vec_series): Likewise.
8595 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8596 * simplify-rtx.c (simplify_unary_operation): Handle negations
8598 (simplify_binary_operation_series): New function.
8599 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8600 (test_vector_ops_series): New function.
8601 (test_vector_ops): Call it.
8602 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8603 gen_const_vec_series.
8604 (altivec_lvsr): Likewise.
8605 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8607 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8608 Alan Hayward <alan.hayward@arm.com>
8609 David Sherwood <david.sherwood@arm.com>
8611 * doc/rtl.texi (const): Update description of address constants.
8612 Say that vector constants are allowed too.
8613 * common.md (E, F): Use CONSTANT_P instead of checking for
8615 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8616 checking for CONST_VECTOR.
8617 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8619 * expr.c (expand_expr_real_2): Check for vector modes instead
8620 of checking for CONST_VECTOR.
8621 * rtl.h (const_vec_p): New function.
8622 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8623 (unwrap_const_vec_duplicate): Handle them here too.
8625 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8626 David Malcolm <dmalcolm@redhat.com>
8627 Alan Hayward <alan.hayward@arm.com>
8628 David Sherwood <david.sherwood@arm.com>
8630 * rtl.h (vec_duplicate_p): New function.
8631 * selftest-rtl.c (assert_rtx_eq_at): New function.
8632 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8633 (assert_rtx_eq_at): Declare.
8634 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8635 * selftest-run-tests.c (selftest::run_tests): Call it.
8636 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8637 (simplify_unary_operation_1): Recursively handle vector duplicates.
8638 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8640 (simplify_subreg): Handle subregs of vector duplicates.
8641 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8642 (selftest::simplify_rtx_c_tests): New functions.
8644 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8645 Alan Hayward <alan.hayward@arm.com>
8646 David Sherwood <david.sherwood@arm.com>
8648 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8649 (gen_vec_duplicate): Likewise.
8650 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8652 (gen_const_vector): ...here.
8653 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8654 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8655 whose elements are all equal.
8656 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8657 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8658 (simplify_relational_operation): Likewise.
8659 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8661 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8662 (aarch64_expand_vector_init): Likewise.
8663 * config/arm/arm.c (neon_vdup_constant): Likewise.
8664 (neon_expand_vector_init): Likewise.
8665 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8666 (arm_block_set_unaligned_vect): Likewise.
8667 (arm_block_set_aligned_vect): Likewise.
8668 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8669 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8670 (expand_vec_perm_even_odd_pack): Likewise.
8671 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8672 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8673 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8674 gen_const_vec_duplicate.
8675 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8676 * config/mips/mips.c (mips_gen_const_int_vector): Use
8677 gen_const_vec_duplicate.
8678 (mips_expand_vector_init): Use CONST0_RTX.
8679 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8680 (define_split): Use gen_const_vec_duplicate.
8681 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8682 (define_split): Use gen_const_vec_duplicate.
8683 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8684 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8685 * config/spu/spu.c (spu_const): Likewise.
8687 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8688 Alan Hayward <alan.hayward@arm.com>
8689 David Sherwood <david.sherwood@arm.com>
8691 * combine.c (can_change_dest_mode): Reject changes in
8692 REGMODE_NATURAL_SIZE.
8694 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8696 * configure.ac (--enable-libssp): New.
8697 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8698 trying to determine target-specific default. Adjust indentation.
8699 * configure: Regenerated.
8700 * doc/install.texi (Configuration): Expand --disable-libssp
8703 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8705 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8708 2017-10-31 Martin Jambor <mjambor@suse.cz>
8711 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8713 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8715 * auto-profile.c (autofdo_source_profile::read): Use
8716 UNKNOWN_LOCATION rather than 0.
8717 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8718 (warning_at): ...this overload.
8719 (warning_at_rich_loc_n): Rename to...
8720 (warning_n): ...this overload.
8721 (error_at_rich_loc): Rename to...
8722 (error_at): ...this overload.
8723 (pedwarn_at_rich_loc): Rename to...
8724 (pedwarn): ...this overload.
8725 (permerror_at_rich_loc): Rename to...
8726 (permerror): ...this overload.
8727 (inform_at_rich_loc): Rename to...
8728 (inform): ...this overload.
8729 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8730 (diagnostic_n_impl_richloc): Rename to...
8731 (diagnostic_n_impl): ...this rich_location *-based decl.
8732 (inform_at_rich_loc): Rename to...
8733 (inform): ...this, and add an assertion.
8734 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8735 (warning_at_rich_loc): Rename to...
8736 (warning_at): ...this, and add an assertion.
8737 (warning_at_rich_loc_n): Rename to...
8738 (warning_n): ...this, and add an assertion.
8739 (warning_n): Update location_t-based implementation for removal of
8740 location_t-based diagnostic_n_impl.
8741 (pedwarn_at_rich_loc): Rename to...
8742 (pedwarn): ...this, and add an assertion.
8743 (permerror_at_rich_loc): Rename to...
8744 (permerror): ...this, and add an assertion.
8745 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8746 (error_at_rich_loc): Rename to...
8747 (error_at): ...this, and add an assertion.
8748 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8749 (driver::do_spec_on_infiles): Likewise.
8750 * substring-locations.c (format_warning_va): Update for renaming
8751 of inform_at_rich_loc.
8753 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8755 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8756 _Float<N>X built-in functions so that the variant without the
8757 "__builtin_" prefix is only enabled for the GNU C and Objective C
8758 languages when they are in non-strict ANSI/ISO mode.
8759 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8760 * target.def (floatn_builtin_p): Add a target hook to control
8761 whether _Float<N> and _Float<N>X built-in functions without the
8762 "__builtin_" prefix are enabled, and return true for C and
8763 Objective C in the default hook. Include langhooks.h in
8765 * targhooks.h (default_floatn_builtin_p): Likewise.
8766 * targhooks.c (default_floatn_builtin_p): Likewise.
8767 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8768 floatn_builtin_p target hook.
8769 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8771 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8772 Eric Botcazou <ebotcazou@adacore.com>
8774 PR rtl-optimization/81803
8775 * lra-constraints.c (curr_insn_transform): Also reload the whole
8776 register for a strict subreg no wider than a word if this is for
8777 a WORD_REGISTER_OPERATIONS target.
8779 2017-10-31 Jason Merrill <jason@redhat.com>
8781 * gdbinit.in: Skip over inlines from timevar.h.
8783 2017-10-31 Martin Liska <mliska@suse.cz>
8785 * doc/gcov.texi: Document new option.
8786 * gcov.c (print_usage): Likewise print it.
8787 (process_args): Support the argument.
8788 (format_count): New function.
8789 (format_gcov): Use the function.
8791 2017-10-31 Martin Liska <mliska@suse.cz>
8793 * gcov.c (struct name_map): do not use typedef.
8794 Define operator== and operator<.
8795 (name_search): Remove.
8796 (name_sort): Remove.
8797 (main): Do not allocate names.
8798 (process_file): Add vertical space.
8799 (generate_results): Use std::find.
8800 (release_structures): Do not release memory.
8801 (find_source): Use std::find.
8803 2017-10-31 Martin Liska <mliska@suse.cz>
8805 * gcov.c (struct line_info): Remove it's typedef.
8806 (line_info::line_info): Add proper ctor.
8807 (line_info::has_block): Do not use a typedef.
8808 (struct source_info): Do not use typedef.
8809 (circuit): Likewise.
8810 (get_cycles_count): Likewise.
8811 (output_intermediate_file): Iterate via vector iterator.
8812 (add_line_counts): Use std::vector methods.
8813 (accumulate_line_counts): Likewise.
8814 (output_lines): Likewise.
8816 2017-10-31 Martin Liska <mliska@suse.cz>
8818 * gcov.c (struct source_info): Remove typedef.
8819 (source_info::source_info): Add proper ctor.
8820 (accumulate_line_counts): Use struct, not it's typedef.
8821 (output_gcov_file): Likewise.
8822 (output_lines): Likewise.
8823 (main): Do not allocate an array.
8824 (output_intermediate_file): Use size of vector container.
8825 (process_file): Resize the vector.
8826 (generate_results): Do not preallocate, use newly added vector
8828 (release_structures): Do not release sources.
8829 (find_source): Use vector methods.
8830 (add_line_counts): Do not use typedef.
8832 2017-10-31 Martin Liska <mliska@suse.cz>
8834 * doc/gcov.texi: Document that.
8835 * gcov.c (add_line_counts): Mark lines with a non-executed
8837 (output_line_beginning): Handle such lines.
8838 (output_lines): Pass new argument.
8839 (output_intermediate_file): Print it in intermediate format.
8841 2017-10-31 Martin Liska <mliska@suse.cz>
8843 * color-macros.h: New file.
8844 * diagnostic-color.c: Factor out color related to macros to
8846 * doc/gcov.texi: Document -k option.
8847 * gcov.c (INCLUDE_STRING): Include string.h.
8848 (print_usage): Add -k option.
8849 (process_args): Parse it.
8850 (pad_count_string): New function.
8851 (output_line_beginning): Likewise.
8852 (DEFAULT_LINE_START): New macro.
8853 (output_lines): Support color output.
8855 2017-10-31 Martin Liska <mliska@suse.cz>
8857 PR gcov-profile/82633
8858 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8859 their interaction with GCOV infrastructure.
8860 * configure.ac: Add -fkeep-{inline,static}-functions to
8862 * configure: Regenerate.
8864 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8867 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8869 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8872 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8873 into a register if it does not fit into an immediate offset field.
8875 2017-10-31 Olivier Hainque <hainque@adacore.com>
8877 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8879 2017-10-31 Julia Koval <julia.koval@intel.com>
8881 * config.gcc: Add gfniintrin.h.
8882 * config/i386/gfniintrin.h: New.
8883 * config/i386/i386-builtin-types.def
8884 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8885 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8886 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8887 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8888 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8889 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8890 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8891 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8892 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8893 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8894 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8895 * config/i386/immintrin.h: Include gfniintrin.h.
8896 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8898 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8900 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8902 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8904 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8905 (arm_ashldi3_1bit): Remove pattern.
8906 (ashrdi3): Remove shift by 1 expansion.
8907 (arm_ashrdi3_1bit): Remove pattern.
8908 (lshrdi3): Remove shift by 1 expansion.
8909 (arm_lshrdi3_1bit): Remove pattern.
8910 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8911 cost of ashldi3 by 1.
8912 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8913 (<shift>di3_neon): Likewise.
8915 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8917 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8918 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8919 and (*aarch64_simd_mov<VQ:mode>).
8920 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8921 pattern alternative.
8922 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8923 attributes to match pattern alternative.
8925 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8927 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8928 * config/rs6000/emmintrin.h: New file.
8929 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8931 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8933 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8934 * (movdi_vfp_cortexa8): Remove pattern.
8936 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8938 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8940 (Specific, alpha*-dec-osf5.1): Remove.
8941 (Specific, mips-sgi-irix5): Remove.
8942 (Specific, mips-sgi-irix6): Remove.
8944 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8947 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8948 arguments to clear_bit_region_be.
8950 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8952 * gimplify.c: Include memmodel.h.
8954 2017-10-30 Martin Jambor <mjambor@suse.cz>
8956 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8957 condition whether loop should be executed at all.
8959 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8961 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8962 gimple folding of vec_madd() intrinsics.
8963 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8964 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8965 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8967 2017-10-30 Richard Biener <rguenther@suse.de>
8969 PR tree-optimization/82762
8971 2017-10-23 Richard Biener <rguenther@suse.de>
8973 PR tree-optimization/82129
8975 2017-08-01 Richard Biener <rguenther@suse.de>
8977 PR tree-optimization/81181
8978 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8979 (compute_antic): ... end of iteration here.
8981 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8983 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8984 -std=iso9899:2017 and -std=gnu17.
8985 * doc/standards.texi (C Language): Document C17 support.
8986 * doc/cpp.texi (Overview): Mention -std=c17.
8987 (Standard Predefined Macros): Document C11 and C17 values of
8988 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8989 * doc/extend.texi (Inline): Do not list individual options for
8990 standards newer than C99.
8991 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8993 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8996 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8998 * asan.c (asan_finish_file): Align asan globals array by shadow
9001 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9004 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9005 (struct store_immediate_info): Add bitregion_start and bitregion_end
9007 (store_immediate_info::store_immediate_info): Add brs and bre
9008 arguments and initialize bitregion_{start,end} from those.
9009 (struct merged_store_group): Add bitregion_start, bitregion_end,
9010 align_base and mask fields. Drop unnecessary struct keyword from
9011 struct store_immediate_info. Add do_merge method.
9012 (clear_bit_region_be): Use memset instead of loop storing zeros.
9013 (merged_store_group::do_merge): New method.
9014 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9015 stores as long as the surrounding bitregions have no gaps.
9016 (merged_store_group::merge_overlapping): Use do_merge.
9017 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9018 is byte aligned, rather than requiring that start and width are
9019 byte aligned. Drop unnecessary struct keyword from
9020 struct store_immediate_info. Allocate and populate also mask array.
9021 Make start of the arrays relative to bitregion_start rather than
9022 start and size them according to bitregion_{end,start} difference.
9023 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9024 struct store_immediate_info.
9025 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9026 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9027 struct keyword from struct store_immediate_info.
9028 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9029 between stores as long as the surrounding bitregions have no gaps.
9031 (struct split_store): Add orig non-static data member.
9032 (split_store::split_store): Initialize orig to false.
9033 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9034 if there is exactly a single original stmt. Change stmts argument
9035 to pointer from reference, if NULL, don't push anything to it. Add
9036 first argument, use it to optimize skipping over orig stmts that
9037 are known to be before bitpos already. Simplify.
9038 (split_group): Return unsigned int count how many stores are or
9039 would be needed rather than a bool. Add allow_unaligned argument.
9040 Change split_stores argument from reference to pointer, if NULL,
9041 only do a dry run computing how many stores would be produced.
9042 Rewritten algorithm to use both alignment and misalign if
9043 !allow_unaligned and handle bitfield stores with gaps.
9044 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9045 from bitregion_start instead of start. Compute allow_unaligned
9046 here, if true, do 2 split_group dry runs to compute which one
9047 produces fewer stores and prefer aligned if equal. Punt if
9048 new count is bigger or equal than original before emitting any
9049 statements, rather than during that. Remove no longer needed
9050 new_ssa_names tracking. Replace num_stmts with
9051 split_stores.length (). Use 32-bit stack allocated entries
9052 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9053 if possible. Handle bitfields with gaps.
9054 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9055 Compute bitregion_{start,end} for the stores and construct
9056 store_immediate_info with that. Formatting fixes.
9058 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9061 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9062 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9064 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9066 * gimplify.c: Include tm_p.h.
9068 * common.opt (gcoff): Re-add as ignored option.
9069 (gcoff1, gcoff2, gcoff3): Likewise.
9071 * Makefile.in (OBJS): Delete sdbout.o.
9072 (GTFILES): Delete $(srcdir)/sdbout.c.
9073 * debug.h: Delete sdb_debug_hooks.
9074 * final.c: Delete sdbout.h include.
9075 (final_scan_insn): Delete SDB_DEBUG check.
9076 (rest_of_clean_state): Likewise.
9077 * output.h: Delete sdb_begin_function_line.
9080 * toplev.c: Delete sdbout.h include.
9081 (process_options): Delete SDB_DEBUG check.
9082 * tree-core.h (tree_type_common): Delete pointer field of
9084 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9085 TYPE_SYMTAB_POINTER.
9086 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9087 (TYPE_SYMTAB_IS_POINTER): Delete.
9088 (TYPE_SYMTAB_IS_DIE): Renumber.
9089 * xcoffout.c: Refer to former sdbout.c file.
9090 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9092 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9093 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9094 (Debugging Options): Delete -gcoff.
9095 (-gstabs): Delete SDB reference.
9097 (-gcoff@var{level}): Delete.
9098 * doc/passes.texi (Debugging information output): Delete SDB and
9099 sdbout.c references.
9100 * doc/tm.texi: Regenerate.
9101 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9102 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9104 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9105 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9106 (SDB_DEBUGGING_INFO): Delete.
9107 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9108 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9109 * target.def (output_source_filename): Delete COFF reference.
9111 * common.opt (gcoff): Delete.
9112 (gxcoff+): Update Negative chain.
9113 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9115 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9116 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9117 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9119 (expand_function_start): Change sdb reference to past tense.
9120 (expand_function_end): Change sdb reference to past tense.
9121 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9122 * opts.c (debug_type_names): Delete coff entry.
9123 (common_handle_option): Delete OPT_gcoff case.
9124 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9126 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9127 * config/cris/cris.h: Delete SDB reference in comment.
9128 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9129 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9130 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9131 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9133 (ix86_expand_prologue): Likewise.
9134 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9135 * config/ia64/ia64.h: Likewise.
9136 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9137 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9139 * config/mmix/mmix.h: Likewise.
9140 * config/nds32/nds32.c: Likewise.
9141 * config/stormy/storym16.h: Likewise.
9142 * config/visium/visium.h: Likewise.
9143 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9145 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9147 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9148 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9149 FRAME_POINTER_REGNUM point at high end of local var area.
9151 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9153 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9154 Move comment around. Do not reset best_edge for a copiable
9155 destination if the copy would cause a partition change.
9156 (better_edge_p): Remove redundant check.
9158 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9160 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9162 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9164 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9165 for math functions that have _Float<N> and _Float<N>X variants.
9166 (mathfn_built_in_2): Add support for math functions that have
9167 _Float<N> and _Float<N>X variants.
9168 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9169 (expand_builtin_mathfn_ternary): Add support for fma with
9170 _Float<N> and _Float<N>X variants.
9171 (expand_builtin): Likewise.
9172 (fold_builtin_3): Likewise.
9173 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9174 create math function _Float<N> and _Float<N>X variants as external
9176 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9177 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9178 the __builtin_ prefix and if not strict ansi, without the prefix.
9179 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9180 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9181 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9182 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9183 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9184 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9185 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9186 function signatures for fma _Float<N> and _Float<N>X variants.
9187 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9188 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9189 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9190 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9191 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9192 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9193 * gencfn-macros.c (print_case_cfn): Add support for math functions
9194 that have _Float<N> and _Float<N>X variants.
9195 (print_define_operator_list): Likewise.
9196 (fltfn_suffixes): Likewise.
9198 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9199 for math functions that have _Float<N> and _Float<N>X variants.
9200 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9201 and _Float<N>X variants.
9202 (COPYSIGN): Likewise.
9205 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9206 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9208 (integer_valued_read_call_p): Likewise.
9209 * fold-const-call.c (fold_const_call_ss): Likewise.
9210 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9211 _Float<N> and _Float<N>X variants.
9212 (fold_const_call_ssss): Add support for fma _Float<N> and
9213 _Float<N>X variants.
9214 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9215 support for copysign and fma _Float<N> and _Float<N>X variants.
9216 (backprop::process_builtin_call_use): Likewise.
9217 * tree-call-cdce.c (can_test_argument_range); Add support for
9218 sqrt _Float<N> and _Float<N>X variants.
9219 (edom_only_function): Likewise.
9220 (get_no_error_domain): Likewise.
9221 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9222 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9223 copysign _Float<N> and _Float<N>X variants.
9224 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9225 handled by machine independent code.
9226 (FMAF128): Likewise.
9227 * doc/cpp.texi (Common Predefined Macros): Document defining
9228 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9229 fma _Float<N> and _Float<N>X variants.
9231 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9234 * config/i386/i386-modes.def (CCFPU): Remove definition.
9235 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9236 (ix86_cc_modes_compatible): Ditto.
9237 (ix86_expand_carry_flag_compare): Ditto.
9238 (ix86_expand_int_movcc): Ditto.
9239 (ix86_expand_int_addcc): Ditto.
9240 (ix86_reverse_condition): Ditto.
9241 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9242 Return true/false for unordered/ordered fp comparisons.
9243 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9244 (ix86_prepare_fp_compare_args): Update for rename.
9245 (ix86_expand_fp_compare): Update for rename. Generate unordered
9246 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9247 (ix86_expand_sse_compare_and_jump): Ditto.
9248 * config/i386/predicates.md (fcmov_comparison_operator):
9249 Remove CCFPU mode handling.
9250 (ix86_comparison_operator): Ditto.
9251 (ix86_carry_flag_operator): Ditto.
9252 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9253 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9254 (*cmpu<mode>_cc_i387): Ditto.
9255 (FPCMP): Remove mode iterator.
9256 (unord): Remove mode attribute.
9257 (unord_subst): New define_subst transformation
9258 (unord): New define_subst attribute.
9260 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9261 (*cmpi<unord>xf_i387): Ditto.
9262 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9263 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9264 using unord_subst transformation.
9265 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9266 (round_saeonly): Also handle CCFP mode.
9267 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9268 Remove UNSPEC_SAHF unspec handling.
9270 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9272 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9274 2017-10-27 Jeff Law <law@redhat.com>
9276 * gimple-ssa-sprintf.c: Include domwalk.h.
9277 (class sprintf_dom_walker): New class, derived from dom_walker.
9278 (sprintf_dom_walker::before_dom_children): New function.
9279 (struct call_info): Moved into sprintf_dom_walker class
9280 (compute_formath_length, handle_gimple_call): Likewise.
9281 (sprintf_length::execute): Call the dominator walker rather
9282 than walking the statements.
9284 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9285 gimple statement locations.
9286 (check_array_bounds): Corresponding changes. Get the statement's
9287 location directly from wi->stmt.
9289 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9292 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9294 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9296 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9297 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9299 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9302 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9303 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9304 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9305 maybe_get_pool_constant.
9306 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9309 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9311 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9312 2.26 caveat. Update gas and gld versions.
9313 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9316 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9318 * cgraph.h (set_malloc_flag): Declare.
9319 * cgraph.c (set_malloc_flag_1): New function.
9320 (set_malloc_flag): Likewise.
9321 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9322 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9324 (read_ipa_call_summary): Add support for reading is_return_callee.
9325 (write_ipa_call_summary): Stream is_return_callee.
9326 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9327 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9328 ipa-prop.h, ipa-fnsummary.h.
9329 (pure_const_names): Change to static.
9330 (malloc_state_e): Define.
9331 (malloc_state_names): Define.
9332 (funct_state_d): Add field malloc_state.
9333 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9334 (check_retval_uses): New function.
9335 (malloc_candidate_p): Likewise.
9336 (analyze_function): Add support for malloc attribute.
9337 (pure_const_write_summary): Stream malloc_state.
9338 (pure_const_read_summary): Add support for reading malloc_state.
9339 (dump_malloc_lattice): New function.
9340 (propagate_malloc): New function.
9341 (warn_function_malloc): New function.
9342 (ipa_pure_const::execute): Call propagate_malloc and
9343 ipa_free_fn_summary.
9344 (pass_local_pure_const::execute): Add support for malloc attribute.
9345 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9346 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9348 2017-10-27 Martin Liska <mliska@suse.cz>
9350 PR gcov-profile/82457
9351 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9352 for fork-like functions to be properly instrumented.
9354 2017-10-27 Richard Biener <rguenther@suse.de>
9357 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9358 info when we redirected EH.
9360 2017-10-26 Michael Collison <michael.collison@arm.com>
9362 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9364 (<optab>_trunchf<GPI:mode>2: New pattern.
9365 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9366 * config/aarch64/iterators.md (wv): New mode attribute.
9367 (vf, VF): New mode attributes.
9368 (vgp, VGP): New mode attributes.
9369 (s): Update attribute with SImode and DImode prefixes.
9371 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9373 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9374 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9375 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9376 (nios2_option_overide): Initialize it. Don't allow R0-relative
9377 addressing with PIC.
9378 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9379 (nios2_symbolic_constant_p): Likewise.
9380 (nios2_legitimate_address_p): Likewise.
9381 (nios2_r0rel_section_name_p): New.
9382 (nios2_symbol_ref_in_r0rel_data_p): New.
9383 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9384 (r0rel_constant_p): New.
9385 (nios2_print_operand_address): Handle r0rel_constant_p.
9386 (nios2_cdx_narrow_form_p): Likewise.
9387 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9388 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9389 (Nios II Options): Document -mr0rel-sec.
9391 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9393 * config/nios2/nios2.c: Include xregex.h.
9394 (nios2_gprel_sec_regex): New.
9395 (nios2_option_overide): Initialize it. Don't allow GP-relative
9396 addressing with PIC.
9397 (nios2_small_section_name_p): Check for regex match.
9398 * config/nios2/nios2.opt (mgprel-sec=): New option.
9399 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9400 (Nios II Options): Document -mgprel-sec.
9402 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9404 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9406 2017-10-26 Tom de Vries <tom@codesourcery.com>
9408 PR tree-optimization/82707
9409 * gimple.c (gimple_copy): Fix unsharing of
9410 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9412 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9414 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9415 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9416 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9417 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9418 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9419 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9420 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9421 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9422 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9423 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9424 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9425 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9426 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9427 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9428 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9429 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9430 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9432 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9434 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9436 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9437 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9438 warning to rs6000.c. Remove the Undocumented flag, since it has
9440 (-mabi=ibmlongdouble): Likewise.
9441 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9442 already set, set the default format for long double.
9443 (rs6000_debug_reg_global): Print whether long double is IBM or
9445 (rs6000_option_override_internal): Rework setting long double
9446 format. Only warn if the user is changing the long double default
9447 and they did not use -Wno-psabi.
9448 * doc/invoke.texi (PowerPC options): Update the documentation for
9449 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9451 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9452 Alan Hayward <alan.hayward@arm.com>
9453 David Sherwood <david.sherwood@arm.com>
9455 * rtl.h (wider_subreg_mode): New function.
9456 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9457 rather than an unsigned int *.
9458 * ira-color.c (regno_max_ref_width): Replace with...
9459 (regno_max_ref_mode): ...this new variable.
9460 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9461 Use wider_subreg_mode.
9462 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9463 rather than an unsigned int *.
9464 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9465 (process_alt_operands): Likewise.
9466 (invariant_p): Likewise.
9467 * lra-spills.c (assign_mem_slot): Likewise.
9468 (add_pseudo_to_slot): Likewise.
9469 * lra.c (collect_non_operand_hard_regs): Likewise.
9470 (add_regs_to_insn_regno_info): Likewise.
9471 * reload1.c (regno_max_ref_width): Replace with...
9472 (regno_max_ref_mode): ...this new variable.
9473 (reload): Update accordingly. Update call to
9474 ira_sort_regnos_for_alter_reg.
9475 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9476 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9477 (scan_paradoxical_subregs): Likewise.
9479 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9481 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9482 (aarch64_frame): Add emit_frame_chain boolean.
9483 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9484 Move eh_return case to aarch64_layout_frame.
9485 (aarch64_layout_frame): Initialize emit_frame_chain.
9486 (aarch64_expand_prologue): Use emit_frame_chain.
9488 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9490 * config/aarch64/aarch64.c (aarch64_layout_frame):
9491 Ensure LR is always stored at the bottom of the callee-saves.
9492 Remove rarely used frame layout which saves callee-saves at top of
9493 frame, so the store of LR can be used as a valid probe in all cases.
9495 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9497 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9498 Improve unaligned TImode/TFmode base/offset split.
9500 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9501 Alan Hayward <alan.hayward@arm.com>
9502 David Sherwood <david.sherwood@arm.com>
9504 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9505 * combine.c (find_single_use_1): Likewise.
9506 (expand_field_assignment): Likewise.
9507 (move_deaths): Likewise.
9508 * lra-constraints.c (simplify_operand_subreg): Likewise.
9509 (curr_insn_transform): Likewise.
9510 * lra.c (collect_non_operand_hard_regs): Likewise.
9511 (add_regs_to_insn_regno_info): Likewise.
9512 * rtlanal.c (reg_referenced_p): Likewise.
9513 (covers_regno_no_parallel_p): Likewise.
9515 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9517 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9518 Don't print any bits outside the precision of the value.
9519 * wide-int.cc (test_printing): Add some new tests.
9521 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9523 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9524 supports -xbrace_comment option.
9525 * configure: Regenerate.
9526 * config.in: Regenerate.
9527 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9528 (ASM_CPU_SPEC): Use it.
9530 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9532 * target.def (static_rtx_alignment): New hook.
9533 * targhooks.h (default_static_rtx_alignment): Declare.
9534 * targhooks.c (default_static_rtx_alignment): New function.
9535 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9536 * doc/tm.texi: Regenerate.
9537 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9538 instead of targetm.constant_alignment. Remove call to
9540 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9541 (cris_preferred_mininum_alignment): New function, split out from...
9542 (cris_constant_alignment): ...here.
9543 (cris_static_rtx_alignment): New function.
9544 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9546 (ix86_constant_alignment): ...here.
9547 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9548 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9549 (mmix_static_rtx_alignment): New function.
9550 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9551 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9553 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9556 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9557 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9559 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9561 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9562 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9563 mark as a sign-extending load.
9564 (local_pic_load_u): Define.
9566 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9569 * fold-const.c (operand_equal_for_comparison_p): Also return true
9570 if ARG0 is a simple variant of ARG1 with narrower precision.
9571 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9573 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9575 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9577 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9578 scatter_static, scatter_per_elt.
9579 * x86-tune-costs.h: Add new cost entries.
9581 2017-10-25 Richard Biener <rguenther@suse.de>
9583 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9584 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9585 class pass_fre): Move to ...
9586 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9587 class pass_fre): ... here and adjust for statistics.
9589 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9592 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9593 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9594 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9595 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9598 2017-10-25 Richard Biener <rguenther@suse.de>
9600 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9601 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9602 eliminate_push_avail, eliminate_insert): Move inside...
9603 (class eliminate_dom_walker): ... this class in preparation
9605 (fini_eliminate): Remove by merging with ...
9606 (eliminate): ... this function. Adjust for class changes.
9607 (pass_pre::execute): Remove fini_eliminate call.
9608 (pass_fre::execute): Likewise.
9610 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9613 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9614 (VPERMI2, VPERMI2I): New mode iterators.
9615 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9616 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9618 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9619 mode iterator. Remove 3 old define_insn patterns.
9620 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9621 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9622 VPERMI2 mode iterator, remove the other two expanders.
9623 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9624 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9625 instructions, remove the other two patterns.
9626 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9627 mode iterator, remove the other two patterns.
9628 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9629 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9630 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9631 and adjust argument order accordingly.
9632 (ix86_expand_vec_perm): Adjust caller.
9633 (expand_vec_perm_1): Likewise.
9634 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9635 (expand_vec_perm_vpermt2_vpshub2): ... this.
9636 (ix86_expand_vec_perm_const_1): Adjust caller.
9637 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9640 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9641 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9642 (vec_shl_<mode>): Remove unused expander.
9643 (avx512bw_<shift_insn><mode>3): New define_insn.
9644 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9645 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9647 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9650 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9653 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9655 PR rtl-optimization/82396
9656 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9657 (autopref_multipass_init): Simplify initialization.
9658 (autopref_rank_data): Simplify sort order.
9659 * gcc/sched-int.h (autopref_multipass_data_): Remove
9660 multi_mem_insn_p, min_offset and max_offset.
9662 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9665 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9666 Check special value of flag_omit_frame_pointer.
9667 (aarch64_can_eliminate): Likewise.
9668 (aarch64_override_options_after_change_1): Simplify handling of
9669 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9671 2017-10-24 Richard Biener <rguenther@suse.de>
9673 PR tree-optimization/82697
9674 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9675 zero for conditional load and unconditional store.
9677 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9679 * doc/install.texi: Document bootstrap-cet.
9681 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9684 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9685 ENDBR instruction at function entrance if function is only
9688 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9691 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9692 patterns to better describe from which operation the CF is computed.
9693 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9694 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9695 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9696 is 0, use _0 suffixed expanders instead of emitting a comparison
9699 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9701 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9702 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9704 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9707 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9708 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9709 * loop-iv.c (iv_get_reaching_def): Likewise.
9710 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9711 variable is promoted and the partition contains undefined values.
9713 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9715 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9717 (nios2_address_cost): Define.
9718 (nios2_legitimize_address): Recognize (exp + constant) directly.
9719 (TARGET_ADDRESS_COST): Define.
9721 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9723 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9724 (nios2_symbolic_memory_operand_p): Declare.
9725 (nios2_split_large_constant): Declare.
9726 (nios2_split_symbolic_memory_operand): Declare.
9727 * config/nios2/nios2.c: Adjust includes.
9728 (nios2_symbolic_constant_allowed): New.
9729 (nios2_symbolic_constant_p): New.
9730 (nios2_plus_symbolic_constant_p): New.
9731 (nios2_valid_addr_expr_p): Recognize addresses involving
9733 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9734 (nios2_symbolic_memory_operand_p): New.
9735 (nios2_large_constant_p): New.
9736 (nios2_split_large_constant): New.
9737 (nios2_split_plus_large_constant): New.
9738 (nios2_split_symbolic_memory_operand): New.
9739 (nios2_legitimize_address): Code refactoring. Handle addresses
9740 involving symbolic constants.
9741 (nios2_emit_move_sequence): Likewise.
9742 (nios2_print_operand): Improve error output.
9743 (nios2_print_operand_address): Handle LO_SUM.
9744 (nios2_cdx_narrow_form_p): Likewise.
9745 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9746 operands involving symbolic constants.
9747 (movhi_internal, movsi_internal): Likewise.
9748 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9749 (extendhisi2, extendqi<mode>2): Likewise.
9751 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9753 * tree-pass.h (PROP_rtl_split_insns): Define.
9754 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9756 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9758 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9760 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9763 * target.def (const_not_ok_for_debug_p): Default to
9764 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9765 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9766 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9767 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9768 which targetm.const_not_ok_for_debug_p returned true.
9769 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9771 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9773 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9774 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9775 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9777 (ix86_const_not_ok_for_debug_p): New function.
9778 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9779 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9781 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9784 * system.h: Conditionally include "unique-ptr.h" if
9785 INCLUDE_UNIQUE_PTR is defined.
9786 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9787 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9789 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9791 * config/rl78/rl78.md: New define_expand "subdi3".
9793 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9796 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9797 DF_REF_INSN if DF_REF_INSN_INFO is false.
9799 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9801 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9802 xmm_move instead of sse_move.
9803 (sse_store_index): New function.
9804 (ix86_register_move_cost): Be more sensible about mismatch stall;
9805 model AVX moves correctly; make difference between sse->integer and
9807 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9808 moves; make difference between SSE and AVX.
9809 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9810 and zmm_move. Increase size of sse load and store tables;
9811 add unaligned load and store tables; add ssemmx_to_integer.
9812 * x86-tune-costs.h: Update all entries according to real
9813 move latencies from Agner Fog's manual and chip documentation.
9815 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9818 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9819 * config/i386/constraints.md (Wf): New constraint.
9820 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9821 (cmp<dwi>_doubleword): Removed.
9822 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9823 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9824 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9825 expand with cmp<dwi>_doubleword. For LTU and GEU use
9826 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9828 * common.opt (gcolumn-info): Enable by default.
9829 * doc/invoke.texi (gcolumn-info): Document new default.
9831 2017-10-23 Richard Biener <rguenther@suse.de>
9833 PR tree-optimization/82672
9834 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9835 Fold the stmt if we propagated into it.
9837 2017-10-23 Richard Biener <rguenther@suse.de>
9839 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9840 (bitmap_remove_expr_from_set): ... this. All callers call this
9841 for non-constant values.
9842 (bitmap_set_subtract): Rename to...
9843 (bitmap_set_subtract_expressions): ... this. Adjust and
9845 (bitmap_set_contains_value): Remove superfluous check.
9846 (bitmap_set_replace_value): Inline into single caller ...
9847 (bitmap_value_replace_in_set): ... here and simplify.
9848 (dependent_clean): Merge into ...
9849 (clean): ... this using an overload. Adjust.
9850 (prune_clobbered_mems): Adjust.
9851 (compute_antic_aux): Likewise.
9852 (compute_partial_antic_aux): Likewise.
9854 2017-10-23 Richard Biener <rguenther@suse.de>
9856 PR tree-optimization/82129
9858 2017-08-01 Richard Biener <rguenther@suse.de>
9860 PR tree-optimization/81181
9861 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9862 (compute_antic): ... end of iteration here.
9864 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9866 * target.def (starting_frame_offset): New hook.
9867 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9868 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9869 * doc/tm.texi.in: Regenerate.
9870 * hooks.h (hook_hwi_void_0): Declare.
9871 * hooks.c (hook_hwi_void_0): New function.
9872 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9873 STARTING_FRAME_OFFSET.
9874 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9875 * reload1.c (reload): Likewise.
9876 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9877 instead of STARTING_FRAME_OFFSET.
9878 * function.c (try_fit_stack_local): Likewise.
9879 (assign_stack_local_1): Likewise
9880 (instantiate_virtual_regs): Likewise.
9881 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9882 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9883 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9884 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9885 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9886 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9887 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9888 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9889 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9890 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9891 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9892 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9893 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9894 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9895 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9896 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9897 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9898 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9899 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9900 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9901 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9902 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9903 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9904 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9905 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9906 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9907 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9908 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9909 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9910 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9911 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9912 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9913 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9914 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9915 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9916 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9917 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9918 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9919 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9920 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9921 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9922 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9923 return a HOST_WIDE_INT.
9924 (avr_builtin_setjmp_frame_value): Use it instead of
9925 STARTING_FRAME_OFFSET.
9926 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9927 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9928 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9930 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9931 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9932 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9933 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9934 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9935 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9936 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9937 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9938 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9939 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9940 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9941 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9943 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9944 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9945 * config/mips/mips.c (mips_compute_frame_info): Refer to
9946 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9947 (mips_starting_frame_offset): New function.
9948 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9949 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9950 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9951 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9952 and return a HOST_WIDE_INT.
9953 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9954 (mmix_initial_elimination_offset): Refer to
9955 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9956 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9957 * config/pa/pa.c (pa_starting_frame_offset): New function.
9958 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9959 (pa_expand_prologue): Likewise.
9960 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9961 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9962 !FRAME_GROWS_DOWNWARD handling to...
9963 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9964 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9965 !FRAME_GROWS_DOWNWARD handling to...
9966 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9967 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9968 !FRAME_GROWS_DOWNWARD handling to...
9969 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9970 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9972 (rs6000_starting_frame_offset): New function.
9973 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9974 !FRAME_GROWS_DOWNWARD handling to...
9975 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9976 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9977 !FRAME_GROWS_DOWNWARD handling to...
9978 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9979 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9980 !FRAME_GROWS_DOWNWARD handling to...
9981 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9982 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9983 (rs6000_starting_frame_offset): New function.
9984 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9985 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9986 * config/vax/vax.c (vax_starting_frame_offset): New function.
9987 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9988 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9989 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9990 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9991 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9992 * system.h (STARTING_FRAME_OFFSET): Poison.
9994 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9996 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9997 SCALAR_TYPE_MODE instead of TYPE_MODE.
9999 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10000 Alan Hayward <alan.hayward@arm.com>
10001 David Sherwood <david.sherwood@arm.com>
10003 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10005 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10006 Alan Hayward <alan.hayward@arm.com>
10007 David Sherwood <david.sherwood@arm.com>
10009 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10011 2017-10-23 Richard Biener <rguenther@suse.de>
10013 PR tree-optimization/82129
10014 * tree-ssa-pre.c (bitmap_set_and): Remove.
10015 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10016 canonicalizing expressions in the set to those with lowest
10017 ID rather than taking that from the first edge.
10019 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10021 * combine.c (rtx_equal_for_field_assignment_p): Use
10022 byte_lowpart_offset.
10024 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10025 Alan Hayward <alan.hayward@arm.com>
10026 David Sherwood <david.sherwood@arm.com>
10028 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10029 to a SUBREG_PROMOTED_VAR.
10031 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10032 Alan Hayward <alan.hayward@arm.com>
10033 David Sherwood <david.sherwood@arm.com>
10035 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10036 (expand_debug_source_expr): Likewise.
10037 * combine.c (combine_simplify_rtx): Likewise.
10038 * cse.c (fold_rtx): Likewise.
10039 * optabs.c (expand_float): Likewise.
10040 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10041 (simplify_binary_operation_1): Likewise.
10043 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10044 Alan Hayward <alan.hayward@arm.com>
10045 David Sherwood <david.sherwood@arm.com>
10047 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10048 (record_promoted_value): Likewise.
10049 * expr.c (expand_expr_real_2): Likewise.
10050 * ree.c (update_reg_equal_equiv_notes): Likewise.
10051 (combine_set_extension): Likewise.
10052 * rtlanal.c (low_bitmask_len): Likewise.
10053 * simplify-rtx.c (neg_const_int): Likewise.
10054 (simplify_binary_operation_1): Likewise.
10056 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10057 Alan Hayward <alan.hayward@arm.com>
10058 David Sherwood <david.sherwood@arm.com>
10060 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10061 * regcprop.c (maybe_mode_change): Likewise.
10062 * reload1.c (alter_reg): Likewise.
10064 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10066 * inchash.h (inchash::hash::add_wide_int): New function.
10067 * lto-streamer-out.c (hash_tree): Use it.
10069 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10071 * inchash.h (inchash::hash::add_wide_int): Rename to...
10072 (inchash::hash::add_hwi): ...this.
10073 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10074 (polymorphic_call_target_hasher::hash): Likewise.
10075 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10076 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10077 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10078 * lto-streamer-out.c (hash_tree): Likewise.
10079 * optc-save-gen.awk: Likewise.
10080 * tree.c (add_expr): Likewise.
10082 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10085 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10086 for ordered inequality comparisons even with TARGET_IEEE_FP.
10088 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10091 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10092 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10093 Expand with cmp<dwi>_doubleword.
10095 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10097 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10098 List CET intrinsics.
10099 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10100 specific to -fcf-protection option.
10102 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10104 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10105 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10106 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10107 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10108 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10109 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10110 (extra_objs): Add cet.o for Linux/x86 targets.
10111 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10112 * config/i386/cet.c: New file.
10113 * config/i386/cetintrin.h: Likewise.
10114 * config/i386/t-cet: Likewise.
10115 * config/i386/cpuid.h (bit_SHSTK): New.
10116 (bit_IBT): Likewise.
10117 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10118 pass IBT and SHSTK bits.
10119 * config/i386/i386-builtin-types.def
10120 (VOID_FTYPE_UNSIGNED_PVOID): New.
10121 (VOID_FTYPE_UINT64_PVOID): Likewise.
10122 * config/i386/i386-builtin.def: Add CET intrinsics.
10123 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10124 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10125 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10126 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10128 * config/i386/i386.c (rest_of_insert_endbranch): New.
10129 (pass_data_insert_endbranch): Likewise.
10130 (pass_insert_endbranch): Likewise.
10131 (make_pass_insert_endbranch): Likewise.
10132 (ix86_notrack_prefixed_insn_p): Likewise.
10133 (ix86_target_string): Add -mibt, -mshstk flags.
10134 (ix86_option_override_internal): Add flag_cf_protection
10136 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10137 (ix86_print_operand): Add 'notrack' prefix output.
10138 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10139 (ix86_expand_builtin): Expand CET intrinsics.
10140 (x86_output_mi_thunk): Add 'endbranch' instruction.
10141 * config/i386/i386.h (TARGET_IBT): New.
10142 (TARGET_IBT_P): Likewise.
10143 (TARGET_SHSTK): Likewise.
10144 (TARGET_SHSTK_P): Likewise.
10145 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10146 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10147 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10148 (builtin_setjmp_setup): New pattern.
10149 (builtin_longjmp): Likewise.
10150 (rdssp<mode>): Likewise.
10151 (incssp<mode>): Likewise.
10152 (saveprevssp): Likewise.
10153 (rstorssp): Likewise.
10154 (wrss<mode>): Likewise.
10155 (wruss<mode>): Likewise.
10156 (setssbsy): Likewise.
10157 (clrssbsy): Likewise.
10158 (nop_endbr): Likewise.
10159 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10161 * config/i386/immintrin.h: Include <cetintrin.h>.
10162 * config/i386/linux-common.h
10163 (file_end_indicate_exec_stack_and_cet): New prototype.
10164 (TARGET_ASM_FILE_END): New.
10166 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10168 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10169 latencies instead of having separate table; make difference between
10170 integer and float costs.
10171 * i386.h (processor_costs): Remove scalar_stmt_cost,
10172 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10173 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10175 * x86-tune-costs.h: Remove entries which has been removed in
10176 procesor_costs from all tables; make cond_taken_branch_cost
10177 and cond_not_taken_branch_cost COST_N_INSNS based.
10179 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10181 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10183 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10185 * config/i386/i386.md (isa): Remove fma_avx512f.
10186 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10187 <avx512>_fmadd_<mode>_mask3<round_name>,
10188 <avx512>_fmsub_<mode>_mask<round_name>,
10189 <avx512>_fmsub_<mode>_mask3<round_name>,
10190 <avx512>_fnmadd_<mode>_mask<round_name>,
10191 <avx512>_fnmadd_<mode>_mask3<round_name>,
10192 <avx512>_fnmsub_<mode>_mask<round_name>,
10193 <avx512>_fnmsub_<mode>_mask3<round_name>,
10194 <avx512>_fmaddsub_<mode>_mask<round_name>,
10195 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10196 <avx512>_fmsubadd_<mode>_mask<round_name>,
10197 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10198 (*vec_widen_umult_even_v16si<mask_name>,
10199 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10200 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10202 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10204 * extend.texi: Add 'nocf_check' documentation.
10205 * gimple.texi: Add second parameter to
10206 gimple_build_call_from_tree.
10207 * invoke.texi: Add -fcf-protection documentation.
10208 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10210 2017-10-20 Richard Biener <rguenther@suse.de>
10212 PR tree-optimization/82473
10213 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10214 the largest input type.
10216 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10218 * c-attribs.c (handle_nocf_check_attribute): New function.
10219 (c_common_attribute_table): Add 'nocf_check' handling.
10220 * gimple-parser.c: Add second argument NULL to
10221 gimple_build_call_from_tree.
10222 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10223 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10225 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10226 * common.opt: Add fcf-protection flag.
10227 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10228 * flag-types.h: Add enum cf_protection_level.
10229 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10230 Add 'nocf_check' attribute propagation to gimple call.
10231 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10232 (gimple_build_call_from_tree): Update prototype.
10233 (gimple_call_nocf_check_p): New function.
10234 (gimple_call_set_nocf_check): Likewise.
10235 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10236 * ipa-icf.c: Add nocf_check attribute in statement hash.
10237 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10238 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10239 * toplev.c (process_options): Add flag_cf_protection handling.
10241 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10243 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10245 2017-10-20 Richard Biener <rguenther@suse.de>
10247 PR tree-optimization/82603
10248 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10249 remove false predicated stores.
10251 2017-10-20 Richard Biener <rguenther@suse.de>
10253 * graphite-isl-ast-to-gimple.c
10254 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10255 Remove return value and simplify, dump copied stmt after lhs
10257 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10258 Reduce dump verbosity.
10259 (gsi_insert_earliest): Likewise.
10260 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10261 * graphite.c (print_global_statistics): Adjust dumping.
10262 (print_graphite_scop_statistics): Likewise.
10263 (print_graphite_statistics): Do not dump loops here.
10264 (graphite_transform_loops): But here.
10266 2017-10-20 Nicolas Roche <roche@adacore.com>
10268 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10269 * configure: Regenerate.
10271 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10274 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10275 functions when optimizing replace GIMPLE_RETURN stmts with
10276 calls to __builtin_unreachable ().
10279 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10280 for -fsanitize=thread link of executables.
10281 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10282 link of executables.
10285 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10286 New mode iterators.
10287 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10288 define_insns for logical vector shifts to use VI248_AVX512BW
10289 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10290 condition, useless isa and prefix attributes. Change the first
10291 2 of these define_insns to ...
10292 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10293 define_insn for avx512vl.
10294 (<shift_insn><mode>3): ... and this, new define_insn without
10295 masking for non-avx512vl.
10298 * config/i386/sse.md (*andnot<mode>3,
10299 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10300 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10301 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10302 not applied use empty suffix even for TARGET_AVX512VL.
10303 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10304 is applied, supply evex,evex or evex,evex,evex instead of just
10307 2017-10-20 Julia Koval <julia.koval@intel.com>
10309 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10310 (OPTION_MASK_ISA_GFNI_UNSET): New.
10311 (ix86_handle_option): Handle OPT_mgfni.
10312 * config/i386/cpuid.h (bit_GFNI): New.
10313 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10314 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10315 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10316 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10317 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10318 * config/i386/i386.opt: Add mgfni.
10320 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10322 * config/msp430/msp430.c (msp430_option_override): Disable
10323 -fdelete-null-pointer-checks.
10324 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10326 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10328 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10329 of x87 and SSE instructions.
10331 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10333 * asan.c (create_cond_insert_point): Do not update edge count.
10334 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10335 (afdo_propagate_circuit): Likewise.
10336 (afdo_calculate_branch_prob): Likewise.
10337 (afdo_annotate_cfg): Likewise.
10338 * basic-block.h (struct edge_def): Remove count.
10339 (edge_def::count): New accessor.
10340 * bb-reorder.c (rotate_loop): Update.
10341 (find_traces_1_round): Update.
10342 (connect_traces): Update.
10343 (sanitize_hot_paths): Update.
10344 * cfg.c (unchecked_make_edge): Update.
10345 (make_single_succ_edge): Update.
10346 (check_bb_profile): Update.
10347 (dump_edge_info): Update.
10348 (update_bb_profile_for_threading): Update.
10349 (scale_bbs_frequencies_int): Update.
10350 (scale_bbs_frequencies_gcov_type): Update.
10351 (scale_bbs_frequencies_profile_count): Update.
10352 (scale_bbs_frequencies): Update.
10353 * cfganal.c (connect_infinite_loops_to_exit): Update.
10354 * cfgbuild.c (compute_outgoing_frequencies): Update.
10355 (find_many_sub_basic_blocks): Update.
10356 * cfgcleanup.c (try_forward_edges): Update.
10357 (try_crossjump_to_edge): Update
10358 * cfgexpand.c (expand_gimple_cond): Update
10359 (expand_gimple_tailcall): Update
10360 (construct_exit_block): Update
10361 * cfghooks.c (verify_flow_info): Update
10362 (redirect_edge_succ_nodup): Update
10363 (split_edge): Update
10364 (make_forwarder_block): Update
10365 (duplicate_block): Update
10366 (account_profile_record): Update
10367 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10368 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10369 * cfgloopmanip.c (scale_loop_profile): Update.
10371 (lv_adjust_loop_entry_edge): Update.
10372 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10373 (force_nonfallthru_and_redirect): Update.
10374 (purge_dead_edges): Update.
10375 (rtl_flow_call_edges_add): Update.
10376 * cgraphunit.c (init_lowered_empty_function): Update.
10377 (cgraph_node::expand_thunk): Update.
10378 * gimple-pretty-print.c (dump_probability): Update.
10379 (dump_edge_probability): Update.
10380 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10381 * haifa-sched.c (sched_create_recovery_edges): Update.
10382 * hsa-gen.c (convert_switch_statements): Update.
10383 * ifcvt.c (dead_or_predicable): Update.
10384 * ipa-inline-transform.c (inline_transform): Update.
10385 * ipa-split.c (split_function): Update.
10386 * ipa-utils.c (ipa_merge_profiles): Update.
10387 * loop-doloop.c (add_test): Update.
10388 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10389 * lto-streamer-in.c (input_cfg): Update.
10390 (input_function): Update.
10391 * lto-streamer-out.c (output_cfg): Update.
10392 * modulo-sched.c (sms_schedule): Update.
10393 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10394 * predict.c (maybe_hot_edge_p): Update.
10395 (unlikely_executed_edge_p): Update.
10396 (probably_never_executed_edge_p): Update.
10397 (dump_prediction): Update.
10398 (drop_profile): Update.
10399 (propagate_unlikely_bbs_forward): Update.
10400 (determine_unlikely_bbs): Update.
10401 (force_edge_cold): Update.
10402 * profile.c (compute_branch_probabilities): Update.
10403 * reg-stack.c (better_edge): Update.
10404 * shrink-wrap.c (handle_simple_exit): Update.
10405 * tracer.c (better_p): Update.
10406 * trans-mem.c (expand_transaction): Update.
10407 (split_bb_make_tm_edge): Update.
10408 * tree-call-cdce.c: Update.
10409 * tree-cfg.c (gimple_find_sub_bbs): Update.
10410 (gimple_split_edge): Update.
10411 (gimple_duplicate_sese_region): Update.
10412 (gimple_duplicate_sese_tail): Update.
10413 (gimple_flow_call_edges_add): Update.
10414 (insert_cond_bb): Update.
10415 (execute_fixup_cfg): Update.
10416 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10417 * tree-complex.c (expand_complex_div_wide): Update.
10418 * tree-eh.c (lower_resx): Update.
10419 (unsplit_eh): Update.
10420 (cleanup_empty_eh_move_lp): Update.
10421 * tree-inline.c (copy_edges_for_bb): Update.
10422 (freqs_to_counts): Update.
10423 (copy_cfg_body): Update.
10424 * tree-ssa-dce.c (remove_dead_stmt): Update.
10425 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10426 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10427 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10428 (unloop_loops): Update.
10429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10430 * tree-ssa-loop-split.c (connect_loops): Update.
10431 (split_loop): Update.
10432 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10433 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10434 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10435 * tree-ssa-reassoc.c (branch_fixup): Update.
10436 * tree-ssa-tail-merge.c (replace_block_by): Update.
10437 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10438 (compute_path_counts): Update.
10439 (update_profile): Update.
10440 (recompute_probabilities): Update.
10441 (update_joiner_offpath_counts): Update.
10442 (estimated_freqs_path): Update.
10443 (freqs_to_counts_path): Update.
10444 (clear_counts_path): Update.
10445 (ssa_fix_duplicate_block_edges): Update.
10446 (duplicate_thread_path): Update.
10447 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10448 (case_bit_test_cmp): Update.
10449 (collect_switch_conv_info): Update.
10450 (gen_inbound_check): Update.
10451 (do_jump_if_equal): Update.
10452 (emit_cmp_and_jump_insns): Update.
10453 * tree-tailcall.c (decrease_profile): Update.
10454 (eliminate_tail_call): Update.
10455 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10456 (vect_do_peeling): Update.
10457 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10458 * ubsan.c (ubsan_expand_null_ifn): Update.
10459 (ubsan_expand_ptr_ifn): Update.
10460 * value-prof.c (gimple_divmod_fixed_value): Update.
10461 (gimple_mod_pow2): Update.
10462 (gimple_mod_subtract): Update.
10463 (gimple_ic): Update.
10464 (gimple_stringop_fixed_value): Update.
10466 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10469 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10471 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10473 PR rtl-optimization/82395
10474 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10475 based on non_spilled_static_chain_regno_p.
10477 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10479 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10480 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10481 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10483 2017-10-19 Martin Sebor <msebor@redhat.com>
10485 PR tree-optimization/82596
10486 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10488 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10490 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10491 (get_mem_refs_of_builtin_call): Likewise.
10492 * builtins.c (expand_builtin_apply): Adjust call to
10493 allocate_dynamic_stack_space.
10494 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10495 the third argument to allocate_dynamic_stack_space, otherwise -1.
10496 (expand_builtin): Deal with all alloca variants.
10497 (is_inexpensive_builtin): Likewise.
10498 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10499 * calls.c (special_function_p): Deal with all alloca variants.
10500 (initialize_argument_information): Adjust call to
10501 allocate_dynamic_stack_space.
10502 (expand_call): Likewise.
10503 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10504 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10505 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10506 use it for the stack usage computation.
10507 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10508 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10509 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10510 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10511 (in_loop_p): Remove first argument and useless check.
10512 (pass_walloca::execute): Remove useless test and adjust call to above.
10513 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10514 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10515 (gimplify_call_expr): Deal with all alloca variants.
10516 * hsa-gen.c (gen_hsa_alloca): Likewise.
10517 (gen_hsa_insns_for_call): Likewise.
10518 * ipa-pure-const.c (special_builtin_state): Likewise.
10519 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10520 * tree-object-size.c (alloc_object_size): Likewise.
10521 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10522 (call_may_clobber_ref_p_1): Likewise.
10523 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10524 (ccp_fold_stmt): Likewise.
10525 (optimize_stack_restore): Likewise.
10526 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10527 (mark_all_reaching_defs_necessary_1): Likewise.
10528 (propagate_necessity): Likewise.
10529 (eliminate_unnecessary_stmts): Likewise.
10530 * tree.c (build_common_builtin_nodes): Build
10531 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10532 (build_alloca_call_expr): New function.
10533 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10534 (CASE_BUILT_IN_ALLOCA): Likewise.
10535 (build_alloca_call_expr): Declare.
10536 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10538 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10541 * dwarf2out.c (new_die_raw): New static inline function.
10542 (new_die): Use it to create the DIE.
10543 (add_AT_external_die_ref): Likewise.
10544 (clone_die): Likewise.
10545 (clone_as_declaration): Likewise.
10546 (dwarf2out_vms_debug_main_pointer): Likewise.
10547 (base_type_die): Likewise. Remove early return for corner cases.
10548 Do not call add_pubtype on the DIE here.
10549 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10550 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10551 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10552 native order exists for base types, attach the DIE manually and call
10553 add_pubtype on it. Do not equate a reverse order DIE to the type.
10555 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10557 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10558 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10560 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10562 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10564 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10565 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10566 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10567 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10568 * opts.c (sanitizer_opts): Add builtin.
10569 * ubsan.c (instrument_builtin): New function.
10570 (pass_ubsan::execute): Call it.
10571 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10572 * doc/invoke.texi: Document -fsanitize=builtin.
10574 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10575 builtins, store max (log2 (align), 0) into uchar field instead of
10576 align into uptr field.
10577 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10578 store uchar 0 field instead of uptr 0 field.
10579 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10580 instead of passing one address of struct with 2 locations pass
10581 two addresses of structs with 1 location each.
10582 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10583 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10584 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10585 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10586 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10587 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10588 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10589 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10591 2017-10-19 Martin Liska <mliska@suse.cz>
10594 * file-find.c (remove_prefix): Remove.
10595 * file-find.h (remove_prefix): Likewise.
10596 * gcc-ar.c: Remove smartness of lookup.
10598 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10600 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10601 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10602 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10604 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10607 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10608 (setcc + and to xor + setcc): New peephole2.
10610 2017-10-19 Tom de Vries <tom@codesourcery.com>
10612 * doc/sourcebuild.texi (Test Directives, Variants of
10613 dg-require-support): Add dg-require-stack-size.
10615 2017-10-19 Martin Liska <mliska@suse.cz>
10618 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10619 that have a large alignment.
10620 (gimplify_target_expr): Likewise.
10622 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10624 PR rtl-optimization/82602
10625 * ira.c (rtx_moveable_p): Return false for volatile asm.
10627 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10630 * config/i386/i386-modes.def (CCGZ): New CC mode.
10631 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10632 * config/i386/predicates.md (ix86_comparison_operator):
10634 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10635 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10636 with double-word subtraction.
10637 (put_condition_code): Handle CCGZmode.
10639 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10641 * wide-int.cc (debug (const wide_int &)): New.
10642 (debug (const wide_int *)): New.
10643 (debug (const widest_int &)): New.
10644 (debug (const widest_int *)): New.
10646 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10648 PR middle-end/82556
10649 * lra-constraints.c (curr_insn_transform): Use non-input operand
10650 instead of output one for matched reload.
10652 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10654 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10655 (tree-ssa-loop-ivopts.h): New header file.
10656 (struct builtin_info): New fields.
10657 (classify_builtin_1): Compute and record base and offset parts for
10658 memset builtin partition by calling strip_offset.
10659 (offset_cmp, fuse_memset_builtins): New functions.
10660 (finalize_partitions): Fuse adjacent memset partitions by calling
10662 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10663 Expose the interface.
10664 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10666 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10668 PR tree-optimization/82574
10669 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10670 that data reference must be executed exactly once per iteration
10671 against the outermost loop in nest.
10672 (classify_partition): Update call to above function.
10674 2017-10-18 Richard Biener <rguenther@suse.de>
10676 PR tree-optimization/82591
10677 * graphite.c (graphite_transform_loops): Move code gen message
10679 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10680 Here. Handle scop_to_isl_ast failing.
10681 (scop_to_isl_ast): Limit the number of ISL operations.
10683 2017-10-18 Richard Biener <rguenther@suse.de>
10685 * graphite-isl-ast-to-gimple.c
10686 (translate_isl_ast_to_gimple::set_rename): Simplify.
10687 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10688 (graphite_copy_stmts_from_block): ... here.
10689 (copy_bb_and_scalar_dependences): Simplify.
10690 (add_parameters_to_ivs_params): Canonicalize.
10691 (generate_entry_out_of_ssa_copies): Simplify.
10692 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10693 by passing in ISL dimension.
10694 (parameter_index_in_region_1): Rename to ...
10695 (parameter_index_in_region): ... this.
10696 (extract_affine): Adjust assert, pass down parameter index.
10697 (add_param_constraints): Use range-info when available.
10698 (build_scop_context): Adjust.
10699 * sese.c (new_sese_info): Adjust.
10700 (free_sese_info): Likewise.
10701 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10702 Remove unused typedefs.
10703 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10705 2017-10-18 Martin Liska <mliska@suse.cz>
10707 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10709 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10711 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10712 (s390_sched_init): Do not reset s390_sched_state if we entered the
10713 current basic block via a fallthru edge and all others are unlikely.
10715 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10717 * config/s390/s390.c (NUM_SIDES): New variable.
10718 (LONGRUNNING_THRESHOLD): New variable.
10719 (LATENCY_FACTOR): New variable.
10720 (s390_sched_score): Decrease score for long-running instructions on
10722 (s390_sched_variable_issue): Perform bookkeeping for long-running
10725 2017-10-18 Richard Biener <rguenther@suse.de>
10727 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10728 Simplify with removal of the parameter rename map.
10729 (set_rename): Likewise.
10730 (should_copy_to_new_region): Likewise.
10731 (graphite_copy_stmts_from_block): Likewise.
10732 (copy_bb_and_scalar_dependences): Remove initialization of
10733 unused copied_bb_map.
10734 (copy_def): Remove.
10735 (copy_internal_parameters): Likewise.
10736 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10737 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10738 Use INTEGRAL_TYPE_P.
10739 (parameter_index_in_region_1): Rename to ...
10740 (assign_parameter_index_in_region): ... this. Assert we have
10741 a parameter we handle.
10742 (scan_tree_for_params): Adjust.
10743 * sese.h (parameter_rename_map_t): Remove.
10744 (struct sese_info_t): Remove unused parameter_rename_map and
10745 copied_bb_map members.
10746 * sese.c (new_sese_info): Adjust.
10747 (free_sese_info): Likewise.
10749 2017-10-18 Martin Liska <mliska@suse.cz>
10752 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10753 on an abnormal edge.
10755 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10757 * doc/invoke.texi (ffunction-sections and fdata-sections):
10760 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10762 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10763 the use statement can throw internally.
10765 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10767 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10768 any RTX present on the RHS of a SET.
10769 * compare-elim.c (try_eliminate_compare): Restore comment.
10771 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10773 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10774 may be also called on tcc_type nodes.
10775 * langhooks.c (lhd_tree_size): Likewise.
10777 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10779 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10780 format_warning_at_substring.
10781 (maybe_warn): Convert source_range * param to a location_t. Pass
10782 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10783 (format_directive): Remove code to extract source_ranges and
10784 source_range * in favor of just a location_t.
10785 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10787 * substring-locations.c (format_warning_va): Convert
10788 source_range * param to a location_t.
10789 (format_warning_at_substring): Likewise.
10790 * substring-locations.h (format_warning_va): Likewise.
10791 (format_warning_at_substring): Likewise.
10793 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10795 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10797 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10798 and scatter/gather ops.
10800 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10801 vec_gather_load and vec_scatter_store.
10802 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10803 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10804 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10805 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10806 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10807 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10809 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10811 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10812 Detect FTST instruction and handle its register pops. Only pop
10813 second operand if can_pop_second_op is true.
10814 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10815 set can_pop_second_op to false in the compare_for_stack_reg call.
10817 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10818 output_fp_compare for stack register operands.
10819 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10820 instructions here. Do not emit stack register pops here. Assert
10821 that FCOMPP pops next to top stack register. Rewrite function.
10823 2017-10-17 Nathan Sidwell <nathan@acm.org>
10825 PR middle-end/82577
10826 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10827 use DECL_ASSEMBLER_NAME_RAW.
10829 PR middle-end/82546
10830 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10833 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10834 Wilco Dijkstra <wilco.dijkstra@arm.com>
10836 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10837 converstion to Pmode from the buf_addr.
10839 2017-10-17 Richard Biener <rguenther@suse.de>
10841 * graphite-dependences.c (scop_get_reads_and_writes): Change
10842 output parameters to references.
10844 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10846 PR 71026/tree-optimization
10847 * fold-const.c (distribute_real_division): Removed.
10848 (fold_binary_loc): Remove calls to distribute_real_divison.
10850 2017-10-17 Richard Biener <rguenther@suse.de>
10852 * graphite-scop-detection.c
10853 (scop_detection::stmt_has_simple_data_refs_p): Always use
10854 the full nest as region.
10855 (try_generate_gimple_bb): Likewise.
10856 * sese.c (scalar_evolution_in_region): Simplify now that
10857 SCEV can handle instantiation in regions.
10858 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10859 in the non-loop part of a function if requested.
10861 2017-10-17 Richard Biener <rguenther@suse.de>
10863 PR tree-optimization/82563
10864 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10866 (graphite_regenerate_ast_isl): Call it.
10867 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10869 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10871 PR tree-optimization/82549
10872 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10873 Formatting fixes. Instead of calling make_bit_field_ref with negative
10876 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10878 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10879 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10880 _mm_maskz_reduce_ss): New.
10881 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10882 __builtin_ia32_reducess_mask): Ditto..
10883 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10884 * config/i386/sse.md (reduces<mode>): Renamed to ...
10885 (reduces<mode><mask_scalar_name>): ... this.
10886 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10888 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10889 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10891 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10893 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10894 * selftest-run-tests.c (selftest::run_tests): Call
10895 selftest::unique_ptr_tests_cc_tests.
10896 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10897 * unique-ptr-tests.cc: New file.
10899 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10902 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10904 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10905 (make_hard_regno_born, make_hard_regno_dead): Update
10906 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10908 2017-10-16 Jeff Law <law@redhat.com>
10910 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10912 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10914 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10916 2017-10-16 Olivier Hainque <hainque@adacore.com>
10918 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10919 with_cpu if we were configured for an e500v2 target cpu name.
10921 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10923 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10924 * doc/invoke.texi: Document +nodsp as a valid extension for
10927 2017-10-16 Martin Liska <mliska@suse.cz>
10929 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10930 (test_set_range): Likewise.
10931 (test_range_functions): Rename to ...
10932 (test_bit_in_range): ... this.
10933 (sbitmap_c_tests): Add new test.
10935 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10937 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10939 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10940 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10942 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10944 * config/aarch64/aarch64-builtins.c
10945 (aarch64_types_quadopu_lane_qualifiers): New.
10946 (TYPES_QUADOPU_LANE): New.
10947 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10948 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10949 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10950 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10951 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10952 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10953 (Vdottype, DOTPROD): New.
10954 (sur): Add SDOT and UDOT.
10956 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10958 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10959 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10960 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10961 Add TARGET_DOTPROD.
10962 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10963 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10964 Enable TARGET_DOTPROD.
10965 (cortex-a75.cortex-a55): Likewise.
10966 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10968 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10970 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10971 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10973 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10975 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10976 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10977 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10978 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10979 * config/arm/types.md (neon_dot, neon_dot_q): New.
10980 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10982 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10984 * config/arm/arm.h (TARGET_DOTPROD): New.
10985 * config/arm/arm.c (arm_arch_dotprod): New.
10986 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10987 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10988 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10989 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10990 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10991 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10992 * doc/invoke.texi (armv8.2-a): Document dotprod
10994 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10996 * i386.c (ix86_vec_cost): New function.
10997 (ix86_rtx_costs): Handle vector operations better.
10998 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10999 * x86-tune-costs.h: Add new costs to all tables.
11001 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11003 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11005 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11006 divsd, sqrtss and sqrtsd
11007 * x86-tune-costs.h: Add new entries to all costs.
11008 (znver1_cost): Fix to match real instruction latencies.
11010 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11011 Michael Collison <michael.collison@arm.com>
11013 * compare-elim.c: Include emit-rtl.h.
11014 (can_merge_compare_into_arith): New function.
11015 (try_validate_parallel): Likewise.
11016 (try_merge_compare): Likewise.
11017 (try_eliminate_compare): Call the above when no previous clobber
11019 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11022 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11024 PR middle-end/62263
11025 PR middle-end/82498
11026 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11027 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11029 PR middle-end/62263
11030 PR middle-end/82498
11031 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11032 to be any operand_equal_p operands. For & (B - 1) require
11033 B to be power of 2. Recognize
11034 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11036 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11039 * optabs.c (expand_memory_blockage): Fix call of
11040 targetm.have_memory_blockage.
11042 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11045 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11046 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11047 objects to extra_objs instead of overwriting it.
11049 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11051 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11052 Use any_fp_register_operand as operand[3] predicate. Simplify
11053 equality test for operands[2] and operands[4] memory location.
11054 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11055 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11056 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11057 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11058 any_fp_register_operand as operand[1] predicate. Simplify
11059 equality test for operands[0] and operands[3] memory location.
11060 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11061 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11062 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11064 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11066 * target-insns.def: Add memory_blockage.
11067 * optabs.c (expand_memory_blockage): New function.
11068 (expand_asm_memory_barrier): Rename ...
11069 (expand_asm_memory_blockage): ... to this.
11070 (expand_mem_thread_fence): Call expand_memory_blockage
11071 instead of expand_asm_memory_barrier.
11072 (expand_mem_singnal_fence): Ditto.
11073 (expand_atomic_load): Ditto.
11074 (expand_atomic_store): Ditto.
11075 * doc/md.texi (Standard Pattern Names For Generation):
11076 Document memory_blockage instruction pattern.
11078 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11080 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11081 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11082 * config/rl78/rl78.md: New define_expand "adddi3".
11084 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11086 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11089 2017-10-13 Jeff Law <law@redhat.com>
11091 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11093 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11096 * internal-fn.c (expand_mul_overflow): If both operands have
11097 the same highpart of -1 or 0 and the topmost bit of lowpart
11098 is different, overflow is if res <= 0 rather than res < 0.
11100 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11102 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11103 TARGET_P9_VECTOR code for unaligned_load case.
11105 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11107 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11109 2017-10-13 Nathan Sidwell <nathan@acm.org>
11111 * tree-core.h (tree_contains_struct): Make bool.
11112 * tree.c (tree_contains_struct): Likewise.
11113 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11114 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11115 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11116 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11117 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11119 2017-10-13 Richard Biener <rguenther@suse.de>
11121 * graphite-isl-ast-to-gimple.c
11122 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11123 parameters and dominance check.
11124 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11125 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11126 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11127 Do not update SSA form here or do intermediate IL verification.
11128 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11129 (graphite_initialize): Remove check on the number of loops in
11130 the function and inline into graphite_transform_loops.
11131 (graphite_finalize): Inline into graphite_transform_loops.
11132 (graphite_transform_loops): Perform SSA update and IL verification
11134 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11136 2017-10-13 Richard Biener <rguenther@suse.de>
11138 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11139 graphite_expression_type_precision): Avoid global constructor
11141 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11142 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11143 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11144 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11145 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11146 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11148 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11151 * config/i386/i386.h (ix86_red_zone_size): New.
11152 * config/i386/i386.md (push peephole2s): Replace
11153 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11155 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11156 Alan Hayward <alan.hayward@arm.com>
11157 David Sherwood <david.sherwood@arm.com>
11159 * combine.c (can_change_dest_mode): Reject changes in
11160 REGMODE_NATURAL_SIZE.
11162 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11163 Alan Hayward <alan.hayward@arm.com>
11164 David Sherwood <david.sherwood@arm.com>
11166 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11167 (expand_debug_source_expr): Likewise.
11168 * combine.c (combine_simplify_rtx): Likewise.
11169 * cse.c (fold_rtx): Likewise.
11170 * fwprop.c (canonicalize_address): Likewise.
11171 * targhooks.c (default_shift_truncation_mask): Likewise.
11173 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11174 Alan Hayward <alan.hayward@arm.com>
11175 David Sherwood <david.sherwood@arm.com>
11177 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11178 (widened_mode): Likewise.
11179 (expand_unop): Likewise.
11180 * ree.c (transform_ifelse): Likewise.
11181 (merge_def_and_ext): Likewise.
11182 (combine_reaching_defs): Likewise.
11183 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11185 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11190 * combine.c (gen_lowpart_for_combine): Likewise.
11191 * dwarf2out.c (rtl_for_decl_location): Likewise.
11192 * final.c (alter_subreg): Likewise.
11193 * rtlhooks.c (gen_lowpart_general): Likewise.
11194 (gen_lowpart_if_possible): Likewise.
11196 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11197 Alan Hayward <alan.hayward@arm.com>
11198 David Sherwood <david.sherwood@arm.com>
11200 * calls.c (expand_call): Use subreg_lowpart_offset.
11201 * cse.c (cse_insn): Likewise.
11202 * regcprop.c (copy_value): Likewise.
11203 (copyprop_hardreg_forward_1): Likewise.
11205 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11208 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11209 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11210 =Q constraints to +Q and into insn condition add check
11211 that operands[0] and operands[1] are equal.
11212 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11213 =Q constraints to +Q and into insn condition add check
11214 that operands[0] is equal to either operands[1] or operands[2].
11217 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11218 instead of handling MINUS_EXPR twice (once for each argument),
11219 canonicalize operand order and handle just once, use rtype where
11220 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11223 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11224 any values of __C while still being pattern recognizable as a simple
11225 rotate instruction.
11227 2017-10-13 Richard Biener <rguenther@suse.de>
11229 PR tree-optimization/82451
11231 2017-10-02 Richard Biener <rguenther@suse.de>
11233 PR tree-optimization/82355
11234 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11235 a mapping for the enclosing loop but avoid generating one for
11236 the loop tree root.
11237 (copy_bb_and_scalar_dependences): Remove premature codegen
11238 error on PHIs in blocks duplicated into multiple places.
11239 * graphite-scop-detection.c
11240 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11241 in the region use it as loop and nest to analyze the DR in.
11242 (try_generate_gimple_bb): Likewise.
11243 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11244 (add_loop_constraints): For blocks in a loop not in the region
11245 create a dimension with a single iteration.
11246 * sese.h (gbb_loop_at_index): Remove assert.
11248 * cfgloop.c (loop_preheader_edge): For the loop tree root
11249 return the single successor of the entry block.
11250 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11251 Reset the SCEV hashtable and niters.
11252 * graphite-scop-detection.c
11253 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11254 assert that we only have POLYNOMIAL_CHREC that vary in loops
11255 contained in the region.
11256 (scop_detection::graphite_can_represent_expr): Adjust.
11257 (scop_detection::stmt_has_simple_data_refs_p): For loops
11258 not in the region set loop to NULL. The nest is now the
11259 entry edge to the region.
11260 (try_generate_gimple_bb): Likewise.
11261 * sese.c (scalar_evolution_in_region): Adjust for
11262 instantiate_scev change.
11263 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11264 Make nest parameter the edge into the region.
11265 (create_data_ref): Likewise.
11266 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11267 entry edge into a region and adjust instantiate_scev calls.
11268 (create_data_ref): Likewise.
11269 (graphite_find_data_references_in_stmt): Likewise.
11270 (find_data_references_in_stmt): Pass the loop preheader edge
11271 from the nest argument.
11272 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11273 parameter the edge into the region.
11274 (instantiate_parameters): Use the loop preheader edge as entry.
11275 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11277 (get_instantiated_value_entry): Make instantiate_below parameter
11278 the edge into the region.
11279 (instantiate_scev_name): Likewise. Adjust dominance checks,
11280 when we cannot use loop-based instantiation instantiate by
11281 walking use-def chains.
11282 (instantiate_scev_poly): Adjust.
11283 (instantiate_scev_binary): Likewise.
11284 (instantiate_scev_convert): Likewise.
11285 (instantiate_scev_not): Likewise.
11286 (instantiate_array_ref): Remove.
11287 (instantiate_scev_3): Likewise.
11288 (instantiate_scev_2): Likewise.
11289 (instantiate_scev_1): Likewise.
11290 (instantiate_scev_r): Do not blindly handle N-operand trees.
11291 Do not instantiate array-refs. Handle all constants and invariants.
11292 (instantiate_scev): Make instantiate_below parameter
11293 the edge into the region.
11294 (resolve_mixers): Use the loop preheader edge for the region
11295 parameter to instantiate_scev_r.
11296 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11298 2017-10-13 Richard Biener <rguenther@suse.de>
11300 PR tree-optimization/82525
11301 * graphite-isl-ast-to-gimple.c
11302 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11304 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11305 Fail code generation when we cannot represent the isl integer.
11306 (binary_op_to_tree): Elide modulo operations that are no-ops
11307 in the type we code generate. Remove now superfluous code
11310 2017-10-13 Richard Biener <rguenther@suse.de>
11312 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11313 (scop_detection::harmful_loop_in_region): Remove premature
11314 IV type restriction.
11315 (scop_detection::graphite_can_represent_scev): We can handle
11316 pointer IVs just fine.
11318 2017-10-13 Alan Modra <amodra@gmail.com>
11320 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11321 "Clobbers and Scratch Registers". Add paragraph on
11322 alternative to clobbers for scratch registers and OpenBLAS
11325 2017-10-13 Alan Modra <amodra@gmail.com>
11327 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11328 example of a memory input for a string of known length. Move
11329 commentary out of table. Add a number of new examples
11330 covering array memory inputs.
11332 2017-10-12 Martin Liska <mliska@suse.cz>
11334 PR tree-optimization/82493
11335 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11336 (test_range_functions): New function.
11337 (sbitmap_c_tests): Likewise.
11338 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11339 * selftest.h (sbitmap_c_tests): New function.
11341 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11343 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11345 * config/rs6000/amo.h: Fix spacing issue.
11347 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11350 * config/i386/i386.md (*ashl<mode>3_mask_1,
11351 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11352 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11355 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11357 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11358 (profile_probability): Set max_probability
11359 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11361 * profile-count.c (profile_probability::differs_from_p): Do not
11362 rely on max_probaiblity == 10000
11364 2017-10-12 Jeff Law <law@redhat.com>
11366 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11369 2017-10-12 Martin Sebor <msebor@redhat.com>
11373 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11374 (handle_alias_pairs): Call it.
11375 * common.opt (-Wattribute-alias): New option.
11376 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11377 * doc/invoke.texi (-Wattribute-alias): Document.
11379 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11382 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11384 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11386 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11387 (make_hard_regno_born, make_hard_regno_dead): Update
11388 bb_killed_pseudos and bb_gen_pseudos.
11390 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11392 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11394 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11396 * config/alpha/alpha.c (alpha_split_conditional_move):
11397 Use std::swap instead of manually swapping.
11398 (alpha_stdarg_optimize_hook): Ditto.
11399 (alpha_canonicalize_comparison): Ditto.
11401 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11403 * tree-loop-distribution.c (struct builtin_info): New struct.
11404 (struct partition): Refactor fields into struct builtin_info.
11405 (partition_free): Free struct builtin_info.
11406 (build_size_arg_loc, build_addr_arg_loc): Delete.
11407 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11408 information from struct builtin_info.
11409 (find_single_drs): New function refactored from classify_partition.
11410 Also moved builtin validity checks to this function.
11411 (compute_access_range, alloc_builtin): New functions.
11412 (classify_builtin_st, classify_builtin_ldst): New functions.
11413 (classify_partition): Refactor code into functions find_single_drs,
11414 classify_builtin_st and classify_builtin_ldst.
11415 (distribute_loop): Don't do runtime alias check when distributing
11417 (find_seed_stmts_for_distribution): New function.
11418 (pass_loop_distribution::execute): Refactor code finding seed
11419 stmts into above function. Support distribution for the innermost
11420 two-level loop nest. Adjust dump information.
11422 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11424 * tree-loop-distribution.c: Adjust the general comment.
11425 (NUM_PARTITION_THRESHOLD): New macro.
11426 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11427 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11428 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11429 in call to build_partition_graph.
11430 (finalize_partitions): New parameter. Make loop distribution more
11431 conservative by fusing more partitions.
11432 (distribute_loop): Don't do runtime alias check in case of loop nest
11434 (find_seed_stmts_for_distribution): New function.
11435 (prepare_perfect_loop_nest): New function.
11436 (pass_loop_distribution::execute): Refactor code finding seed stmts
11437 and loop nest into above functions. Support loop nest distribution.
11438 Adjust dump information accordingly.
11440 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11442 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11443 and set PTYPE_SEQUENTIAL for merged partition.
11445 2017-10-12 Richard Biener <rguenther@suse.de>
11447 PR tree-optimization/69728
11449 2017-09-19 Richard Biener <rguenther@suse.de>
11451 PR tree-optimization/69728
11452 * graphite-sese-to-poly.c (schedule_error): New global.
11453 (add_loop_schedule): Handle empty domain by failing the
11455 (build_original_schedule): Handle schedule_error.
11457 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11458 domain by returning an unchanged schedule.
11460 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11462 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11463 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11465 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11467 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11470 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11473 * expr.c (store_field): Don't optimize away bitsize == 0 store
11474 from CALL_EXPR with addressable return type.
11476 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11478 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11479 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11480 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11481 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11482 TARGET_ISEL instead of TARGET_ISEL<sel>.
11484 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11486 * config/rs6000/rs6000.c
11487 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11489 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11491 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11493 (reg_or_cint_operand): Fix comment.
11494 (reg_or_zero_operand): New predicate.
11495 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11496 * config/rs6000/rs6000.c (output_isel): Delete.
11497 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11498 instead of reg_or_cint_operand. Output instruction directly (not via
11500 (isel_unsigned_<mode>): Ditto.
11501 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11502 gpc_reg_operand. Add an instruction alternative for this. Output
11503 instruction directly.
11504 (*isel_reversed_unsigned_<mode>): Ditto.
11506 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11508 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11509 (TARGET_CANONICALIZE_COMPARISON): Define.
11511 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11514 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11515 Check whether the dest is REG before adding REG_EQUIV note.
11517 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11520 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11522 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11523 (make_hard_regno_born, make_hard_regno_dead): Update
11524 bb_killed_pseudos and bb_gen_pseudos.
11526 2017-10-11 Nathan Sidwell <nathan@acm.org>
11528 * incpath.h (enum incpath_kind): Name enum, prefix values.
11529 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11530 * incpath.c (heads, tails): Use INC_MAX.
11531 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11532 (merge_include_chains, split_quote_chain,
11533 register_include_chains): Update incpath_kind names.
11534 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11535 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11537 (add_framework_path, darwin_register_objc_includes): Likewise.
11538 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11540 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11542 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11543 Do not use float_operator operator predicate.
11544 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11545 * config/i386/predicates.md (float_operator): Remove predicate.
11547 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11549 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11550 (*jccxf_i387): Ditto.
11551 (*jcc<mode>_i387): Ditto.
11552 (*jccu<mode>_i387): Ditto.
11553 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11554 (*jcc_*_i387 splitters): Remove.
11555 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11556 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11557 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11560 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11562 * profile-count.h (slow_safe_scale_64bit): New function.
11563 (safe_scale_64bit): New inline.
11564 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11565 * profile-count.c: Include wide-int.h
11566 (slow_safe_scale_64bit): New.
11568 2017-10-11 Nathan Sidwell <nathan@acm.org>
11570 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11571 HAS_DECL_ASSEMBLER_NAME_P.
11572 * gimple-expr.c (gimple_decl_printable_name: Check
11573 HAS_DECL_ASSEMBLER_NAME_P too.
11574 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11575 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11576 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11577 * passes.c (rest_of_decl_compilation): Check
11578 HAS_DECL_ASSEMBLER_NAME_P too.
11579 * recog.c (verify_changes): Likewise.
11580 * tree-pretty-print.c (dump_decl_name): Likewise.
11581 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11583 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11584 (SET_DECL_ASSEMBLER_NAME): Use it.
11585 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11586 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11587 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11589 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11591 * config.gcc (i386, x86_64): Add extra objects.
11592 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11593 (ix86_min_insn_size): Declare.
11594 (ix86_issue_rate): Declare.
11595 (ix86_adjust_cost): Declare.
11596 (ia32_multipass_dfa_lookahead): Declare.
11597 (ix86_macro_fusion_p): Declare.
11598 (ix86_macro_fusion_pair_p): Declare.
11599 (ix86_bd_has_dispatch): Declare.
11600 (ix86_bd_do_dispatch): Declare.
11601 (ix86_core2i7_init_hooks): Declare.
11602 (ix86_atom_sched_reorder): Declare.
11603 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11604 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11605 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11606 (rip_relative_addr_p): Rename to ...
11607 (ix86_rip_relative_addr_p): ... this one; export.
11608 (memory_address_length): Update.
11609 (ix86_issue_rate): Move to x86-tune-sched.c.
11610 (ix86_flags_dependent): Move to x86-tune-sched.c.
11611 (ix86_agi_dependent): Move to x86-tune-sched.c.
11612 (exact_dependency_1): Move to x86-tune-sched.c.
11613 (exact_store_load_dependency): Move to x86-tune-sched.c.
11614 (ix86_adjust_cost): Move to x86-tune-sched.c.
11615 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11616 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11617 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11618 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11619 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11620 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11621 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11622 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11623 (min_insn_size): Rename to ...
11624 (ix86_min_insn_size): ... this one; export.
11625 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11626 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11627 (core2i7_first_cycle_multipass_backtrack): Move to
11628 x86-tune-sched-core.c.
11629 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11630 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11631 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11632 (ix86_avoid_jump_mispredicts): Update.
11633 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11634 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11635 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11636 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11637 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11638 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11639 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11640 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11641 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11642 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11643 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11644 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11645 (BIG): Move to ix86-tune-sched-bd.c.
11646 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11647 (enum insn_path): Move to ix86-tune-sched-bd.c.
11648 (get_mem_group): Move to ix86-tune-sched-bd.c.
11649 (is_cmp): Move to ix86-tune-sched-bd.c.
11650 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11651 (is_branch): Move to ix86-tune-sched-bd.c.
11652 (is_prefetch): Move to ix86-tune-sched-bd.c.
11653 (init_window): Move to ix86-tune-sched-bd.c.
11654 (allocate_window): Move to ix86-tune-sched-bd.c.
11655 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11656 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11657 (process_end_window): Move to ix86-tune-sched-bd.c.
11658 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11659 (find_constant): Move to ix86-tune-sched-bd.c.
11660 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11661 (has_immediate): Move to ix86-tune-sched-bd.c.
11662 (get_insn_path): Move to ix86-tune-sched-bd.c.
11663 (get_insn_group): Move to ix86-tune-sched-bd.c.
11664 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11665 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11666 (add_insn_window): Move to ix86-tune-sched-bd.c.
11667 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11668 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11669 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11670 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11671 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11672 (do_dispatch): Move to ix86-tune-sched-bd.c.
11673 (has_dispatch): Move to ix86-tune-sched-bd.c.
11674 * config/i386/t-i386: Add new object files.
11675 * config/i386/x86-tune-costs.h: New file.
11676 * config/i386/x86-tune-sched-atom.c: New file.
11677 * config/i386/x86-tune-sched-bd.c: New file.
11678 * config/i386/x86-tune-sched-core.c: New file.
11679 * config/i386/x86-tune-sched.c: New file.
11681 2017-10-11 Liu Hao <lh_mouse@126.com>
11683 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11684 the generic version below instead.
11685 (should_colorize): Recognize Windows consoles as terminals
11687 * pretty-print.c [__MINGW32__] (write_all): New function.
11688 [__MINGW32__] (find_esc_head): Likewise.
11689 [__MINGW32__] (find_esc_terminator): Likewise.
11690 [__MINGW32__] (eat_esc_sequence): Likewise.
11691 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11693 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11696 2017-10-11 Richard Biener <rguenther@suse.de>
11698 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11699 Properly call analyze_scalar_evolution with the loop of the stmt.
11701 2017-10-11 Richard Biener <rguenther@suse.de>
11703 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11704 * tree-core.h (tree_base): Add chrec_var union member.
11705 * tree.h (CHREC_VAR): Remove.
11706 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11707 * tree-chrec.h (build_polynomial_chrec): Adjust.
11708 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11709 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11711 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11713 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11714 * match.pd: ... here.
11715 ((T) X == (T) Y): Relax condition.
11717 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11719 PR tree-optimization/82472
11720 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11722 (break_alias_scc_partitions): Update postorder number.
11724 2017-10-11 Martin Liska <mliska@suse.cz>
11727 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11729 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11731 2017-10-11 Martin Liska <mliska@suse.cz>
11733 * print-rtl.c (print_insn): Move declaration of idbuf
11734 to same scope as name.
11736 2017-10-11 Martin Liska <mliska@suse.cz>
11741 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11742 volatile arguments.
11744 2017-10-11 Martin Liska <mliska@suse.cz>
11747 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11748 volatile arguments.
11750 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11752 * config.gcc (default_gnu_indirect_function): Default to yes for
11753 arm*-*-linux* with glibc.
11755 2017-10-11 Richard Biener <rguenther@suse.de>
11757 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11758 default-defs and types we do not want to analyze.
11759 (interpret_loop_phi): Replace unreachable code with an assert.
11760 (compute_scalar_evolution_in_loop): Remove and inline ...
11761 (analyze_scalar_evolution_1): ... here, replacing condition with
11762 what makes the intent clearer. Remove handling of cases
11763 get_scalar_evolution now handles.
11765 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11767 PR rtl-optimization/81434
11768 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11769 comment for main loop. In sched_group_found if, also add checks for
11770 pass and min_cost_group.
11772 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11774 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11775 (rs6000_insn_cost): New function.
11776 * config/rs6000/rs6000.md (cost): New attribute.
11778 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11779 H.J. Lu <hongjiu.lu@intel.com>
11783 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11784 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11785 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11786 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11787 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11788 OPTION_MASK_ISA_MMX and some other bit set require both
11789 mmx and the ISAs without the mmx bit.
11790 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11791 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11792 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11793 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11794 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11795 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11796 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11797 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11798 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11799 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11800 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11801 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11802 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11803 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11804 Add OPTION_MASK_ISA_MMX.
11806 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11808 * config.gcc (armv7*-*-freebsd*): New target.
11809 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11811 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11813 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11814 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11817 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11819 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11820 a template parameter.
11821 (WIDE_INT_REF_FOR): Update accordingly.
11822 * tree.h (wi::int_traits <const_tree>): Delete.
11823 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11824 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11825 (wi::tree_to_wide_ref): New typedef.
11826 (wi::to_wide): New function.
11827 * calls.c (get_size_range): Use wi::to_wide when operating on
11828 trees as wide_ints.
11829 * cgraph.c (cgraph_node::create_thunk): Likewise.
11830 * config/i386/i386.c (ix86_data_alignment): Likewise.
11831 (ix86_local_alignment): Likewise.
11832 * dbxout.c (stabstr_O): Likewise.
11833 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11834 * expr.c (const_vector_from_tree): Likewise.
11835 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11836 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11837 (fold_negate_expr_1, int_const_binop_1, const_binop)
11838 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11839 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11840 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11841 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11842 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11843 (fold_not_const, round_up_loc): Likewise.
11844 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11845 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11846 (alloca_call_type): Likewise.
11847 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11848 * godump.c (go_output_typedef): Likewise.
11849 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11850 * internal-fn.c (get_min_precision): Likewise.
11851 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11852 * ipa-polymorphic-call.c
11853 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11854 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11855 (ipa_modify_call_arguments): Likewise.
11856 * match.pd: Likewise.
11857 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11858 * print-tree.c (print_node_brief, print_node): Likewise.
11859 * stmt.c (expand_case): Likewise.
11860 * stor-layout.c (layout_type): Likewise.
11861 * tree-affine.c (tree_to_aff_combination): Likewise.
11862 * tree-cfg.c (group_case_labels_stmt): Likewise.
11863 * tree-data-ref.c (dr_analyze_indices): Likewise.
11864 (prune_runtime_alias_test_list): Likewise.
11865 * tree-dump.c (dequeue_and_dump): Likewise.
11866 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11867 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11868 * tree-pretty-print.c (dump_generic_node): Likewise.
11869 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11870 (simple_iv_with_niters): Likewise.
11871 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11872 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11873 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11874 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11875 (refine_value_range_using_guard, number_of_iterations_ne_max)
11876 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11877 (get_cst_init_from_scev, record_nonwrapping_iv)
11878 (scev_var_range_cant_overflow): Likewise.
11879 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11880 * tree-ssa-pre.c (compute_avail): Likewise.
11881 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11882 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11883 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11884 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11885 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11886 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11887 (dump_case_nodes, try_switch_expansion): Likewise.
11888 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11889 (vect_do_peeling): Likewise.
11890 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11891 * tree-vect-stmts.c (vectorizable_load): Likewise.
11892 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11893 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11894 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11895 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11896 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11897 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11898 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11899 (evrp_dom_walker::before_dom_children): Likewise.
11900 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11901 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11902 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11903 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11904 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11905 (get_range_pos_neg): Likewise.
11906 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11907 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11908 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11909 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11910 * config/avr/avr.c (avr_fold_builtin): Likewise.
11911 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11912 * config/msp430/msp430.c (msp430_attr): Likewise.
11913 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11914 * config/powerpcspe/powerpcspe-c.c
11915 (altivec_resolve_overloaded_builtin): Likewise.
11916 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11917 (rs6000_expand_ternop_builtin): Likewise.
11918 * config/rs6000/rs6000-c.c
11919 (altivec_resolve_overloaded_builtin): Likewise.
11920 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11921 (rs6000_expand_ternop_builtin): Likewise.
11922 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11924 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11926 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11927 when copying loop nest with only one inner loop.
11929 2017-10-10 Richard Biener <rguenther@suse.de>
11931 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11932 blocks if SCEV is active.
11933 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11935 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11936 (scev_initialize): Assert we are not yet initialized.
11938 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11940 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11941 inner loop's exit stmt by making it always exit the loop, otherwise
11942 we would generate an infinite empty loop.
11944 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11946 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11947 renaming variables in new preheader if it's deleted.
11949 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11951 * tree-loop-distribution.c (struct partition): Remove unused field
11952 loops of the structure.
11953 (partition_alloc, partition_free): Ditto.
11954 (build_rdg_partition_for_vertex): Ditto.
11956 2017-10-09 Jeff Law <law@redhat.com>
11958 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11959 return type to match prototype and documentation.
11961 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11963 * config/rs6000/rs6000.c (processor_costs): Move to ...
11964 * config/rs6000/rs6000.h: ... here.
11965 (rs6000_cost): Declare.
11967 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11969 * except.c (setjmp_fn): New global variable.
11970 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11971 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11972 if DONT_USE_BUILTIN_SETJMP is defined.
11974 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11976 * target.def (insn_cost): New hook.
11977 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11978 * doc/tm.texi: Regenerate.
11979 * rtlanal.c (insn_cost): Use the new hook.
11981 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11983 * combine.c (combine_validate_cost): Compute the new insn_cost,
11984 not just pattern_cost.
11985 (try_combine): Adjust comment.
11987 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11989 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11991 * combine.c (uid_insn_cost): Adjust comment.
11992 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11994 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11995 * dse.c (find_shift_sequence): Ditto.
11996 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11997 (bb_valid_for_noce_process_p): Use pattern_cost.
11998 * rtl.h (insn_rtx_cost): Delete.
11999 (pattern_cost): New prototype.
12000 (insn_cost): New prototype.
12001 * rtlanal.c (insn_rtx_cost): Rename to...
12002 (pattern_cost): ... this.
12005 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12007 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12008 (*jcc<mode>_0_r_i387): Ditto.
12009 (*jccxf_r_i387): Ditto.
12010 (*jcc<mode>_r_i387): Ditto.
12011 (*jccu<mode>_r_i387): Ditto.
12012 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12013 (*jcc): Rename from *jcc_1.
12015 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12017 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12018 deferred rescans after the lvx/stvx recombination pre-pass.
12020 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12022 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12023 memory operation instruction support.
12024 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12025 (rs6000-ibm-aix[789]*): Likewise.
12026 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12027 Document new functions.
12029 2017-10-09 Richard Biener <rguenther@suse.de>
12031 PR tree-optimization/82397
12032 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12033 equality only for semantically equal trees.
12035 2017-10-09 Richard Biener <rguenther@suse.de>
12037 PR tree-optimization/82449
12038 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12039 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12040 allow constant addresses.
12041 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12044 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12046 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12049 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12052 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12055 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12058 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12060 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12063 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12064 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12066 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12068 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12069 (WI_BINARY_PREDICATE_RESULT): Likewise.
12070 (wi::binary_traits::operator_result): New type.
12071 (wi::binary_traits::predicate_result): Likewise.
12072 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12073 (generic_wide_int::operator==, generic_wide_int::operator!=)
12074 (generic_wide_int::operator&, generic_wide_int::and_not)
12075 (generic_wide_int::operator|, generic_wide_int::or_not)
12076 (generic_wide_int::operator^, generic_wide_int::operator+
12077 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12079 (operator~, unary operator-, operator==, operator!=, operator&)
12080 (operator|, operator^, operator+, binary operator-, operator*): New
12082 * expr.c (get_inner_reference): Use wi::bit_and_not.
12083 * fold-const.c (fold_binary_loc): Likewise.
12084 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12085 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12086 (bit_value_binop): Likewise.
12087 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12088 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12089 (extract_range_from_binary_expr_1): Likewise.
12090 (masked_increment): Likewise.
12091 (simplify_bit_ops_using_ranges): Likewise.
12093 2017-10-09 Martin Jambor <mjambor@suse.cz>
12096 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12097 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12098 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12100 (hsa_fixup_mov_insn_type): New function.
12101 (hsa_op_with_type::get_in_type): Use it.
12102 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12103 immediates in an assert.
12104 (hsa_op_with_type::extend_int_to_32bit): New method.
12105 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12106 types. Convert to dest type if necessary.
12107 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12108 (reg_for_gimple_ssa): Pass false as min32int to
12109 hsa_type_for_scalar_tree_type.
12110 (gen_hsa_addr): Fixup type when creating addresable temporary.
12111 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12112 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12113 necessary. Call hsa_fixup_mov_insn_type.
12114 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12115 extend operands and convert to dest type if necessary.
12116 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12117 to dest type if necessary.
12118 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12119 if conversion nt necessary and size matches.
12120 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12121 to dest type if necessary.
12122 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12123 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12125 (gen_hsa_clrsb): Likewise.
12126 (gen_hsa_ffs): Likewise.
12127 (gen_hsa_divmod): Extend operands and convert to dest type if
12129 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12131 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12133 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12134 Remove empty default arguments. Use a brace block as output
12136 (conditional return): Ditto.
12138 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12139 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12141 (group_ending_nop): Ditto.
12142 (doloop_end): Ditto.
12143 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12144 (splitters for those): Ditto.
12146 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12148 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12149 a conditional jump (and the compare for it) so that pc_rtx is the
12151 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12152 for the deleted and renamed ctr<mode>_internal[234] patterns.
12153 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12154 Delete second conditional return pattern.
12155 (ctr<mode>_internal2): Delete this second bdnz pattern.
12156 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12157 (ctr<mode>_internal4): Delete this second bdz pattern.
12159 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12161 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12162 (always_initialized_rtx_for_ssa_name_p): New predicate.
12163 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12164 (finish_out_of_ssa): Free new field of SA.
12165 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12166 * tree-ssa-coalesce.c: Include tree-ssa.h.
12167 (get_parm_default_def_partitions): Remove extern keyword.
12168 (get_undefined_value_partitions): New function.
12169 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12170 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12171 uninitialized bits.
12172 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12174 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12176 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12178 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12180 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12181 for targets that preffer 128bit.
12183 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12185 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12187 2017-10-08 Olivier Hainque <hainque@adacore.com>
12189 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12190 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12192 (thumb_set_return_address): Likewise.
12194 2017-10-08 Olivier Hainque <hainque@adacore.com>
12196 * common/config/arm/arm-common.c (arm_except_unwind_info):
12197 Handle DWARF2_UNWIND_INFO.
12199 2017-10-07 Michael Collison <michael.collison@arm.com>
12201 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12204 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12206 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12207 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12208 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12209 defined, force the creation of a new block for a dispatch label.
12211 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12213 * invoke.texi (Wsuggest-attribute=cold): Document.
12214 * common.opt (Wsuggest-attribute=cold): New
12215 * ipa-pure-const.c (warn_function_cold): New function.
12216 * predict.c (compute_function_frequency): Use it.
12217 * predict.h (warn_function_cold): Declare.
12219 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12221 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12224 2017-10-06 Martin Liska <mliska@suse.cz>
12226 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12227 keyword for member functions.
12228 (struct sanopt_tree_couple): New struct.
12229 (struct sanopt_tree_couple_hash): New function.
12230 (struct sanopt_ctx): Add new hash_map.
12231 (has_dominating_ubsan_ptr_check): New function.
12232 (record_ubsan_ptr_check_stmt): Likewise.
12233 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12234 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12235 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12237 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12240 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12241 aarch64_simd_valid_immediate on CONST_VECTORs.
12242 (aarch64_reg_or_bic_imm): Likewise.
12244 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12246 PR rtl-optimization/82396
12247 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12249 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12251 * graphite-dependences.c (scop_get_reads): Move code to...
12252 (scop_get_must_writes): Move code to...
12253 (scop_get_may_writes): Move code to...
12254 (scop_get_reads_and_writes): ... here.
12255 (scop_get_dependences): Call scop_get_reads_and_writes.
12257 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12259 PR tree-optimization/82434
12260 * fold-const.h (can_native_encode_type_p,
12261 can_native_encode_string_p): Remove.
12262 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12263 don't encode anything, just return what would be otherwise returned.
12264 (native_encode_fixed, native_encode_complex, native_encode_vector):
12266 (native_encode_string): Likewise. Inline by hand
12267 can_native_encode_string_p.
12268 (can_native_encode_type_p): Remove.
12269 (can_native_encode_string_p): Remove.
12270 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12271 STRING_CSTs using can_native_encode_string_p, test all
12272 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12273 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12274 argument from native_encode_expr.
12275 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12276 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12279 2017-10-06 Richard Biener <rguenther@suse.de>
12281 PR tree-optimization/82397
12282 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12283 operand_equal_p but rely on data_ref_compare_tree for detecting
12285 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12286 to match up with dr_group_sort_cmp.
12288 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12291 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12293 * config/s390/s390-builtin-types.def: Regenerate.
12295 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12298 * config/s390/s390-builtin-types.def: Regenerate.
12299 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12300 Change flag from B_VXE to B_VX.
12301 (s390_vec_min_dbl): Remove B_VXE flag.
12303 2017-10-06 Richard Biener <rguenther@suse.de>
12305 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12306 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12307 translate_isl_ast_to_gimple::is_valid_rename,
12308 translate_isl_ast_to_gimple::get_rename,
12309 translate_isl_ast_to_gimple::get_def_bb_for_const,
12310 translate_isl_ast_to_gimple::get_new_name,
12311 translate_isl_ast_to_gimple::collect_all_ssa_names,
12312 translate_isl_ast_to_gimple::copy_loop_phi_args,
12313 translate_isl_ast_to_gimple::collect_all_ssa_names,
12314 translate_isl_ast_to_gimple::copy_loop_phi_args,
12315 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12316 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12317 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12318 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12319 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12320 translate_isl_ast_to_gimple::copy_cond_phi_args,
12321 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12322 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12323 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12324 translate_isl_ast_to_gimple::rename_uses,
12325 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12326 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12327 (set_rename_for_each_def): Likewise.
12328 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12329 here. Handle rewriting SCEV analyzable uses here.
12330 (copy_bb_and_scalar_dependences): Generate code for PHI
12332 (graphite_regenerate_ast_isl): Adjust.
12333 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12334 (add_write, add_read): New functions.
12335 (build_cross_bb_scalars_def): Use it and simplify.
12336 (build_cross_bb_scalars_use): Likewise.
12337 (graphite_find_cross_bb_scalar_vars): Inline into...
12338 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12339 simulating out-of-SSA. Compute liveout and add dependencies.
12340 (build_scops): Force an empty entry block.
12341 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12343 (sese_build_liveouts): Declare.
12344 (sese_trivially_empty_bb_p): Likewise.
12345 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12346 compute liveout and debug_liveout.
12347 (sese_bad_liveouts_use): Remove.
12348 (sese_reset_debug_liveouts_bb): Likewise.
12349 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12350 (sese_build_liveouts): Build liveout and debug_liveout and store
12352 (new_sese_info): Adjust.
12353 (free_sese_info): Likewise.
12354 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12355 do not build liveout here.
12356 (move_sese_in_condition): Adjust region entry.
12357 (scev_analyzable_p): Match up with chrec_apply requirements.
12358 (sese_trivially_empty_bb_p): New.
12359 * tree-into-ssa.c (get_reaching_def): Properly support generating
12360 default-defs for incremental rewrite of anonymous names.
12362 2017-10-06 Richard Biener <rguenther@suse.de>
12364 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12365 precision do not perform modulo reduction.
12367 2017-10-06 Richard Biener <rguenther@suse.de>
12369 PR tree-optimization/82436
12370 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12371 conservatively choose the vectorization factor when checking
12372 whether we can perform the required load permutation.
12373 (vect_transform_slp_perm_load): Assert when we may not fail.
12375 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12377 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12378 message for incompatible -msdata=* and -mcall-* options.
12380 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12382 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12383 rate for post-reload scheduling.
12385 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12387 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12389 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12391 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12392 to improve monte carlo in scimark.
12394 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12396 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12397 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12398 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12399 pentium4_cost, nocona_cost): Set reassociation width to 1.
12400 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12401 width to 2 for fp operations and 1 otherwise.
12402 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12404 (atom_cost): Set reassociation width to 2.
12405 (slm_cost, generic_cost): Set fp reassociation width
12406 to 2 and 1 otherwise.
12407 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12408 (core_cost): Set fp reassociation width to 4 and vector to 2.
12409 (ix86_reassociation_width): Rewrite using cost table; special case
12410 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12411 and TARGET_AVX128_OPTIMAL.
12412 * config/i386/i386.h (processor_costs): Add
12413 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12414 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12415 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12416 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12417 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12418 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12420 2017-10-05 Nathan Sidwell <nathan@acm.org>
12422 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12424 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12426 * config/arm/arm.c (arm_test_fpu_data): New.
12427 (arm_run_selftests): Call arm_test_fpu_data.
12429 2017-10-04 Nathan Sidwell <nathan@acm.org>
12431 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12433 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12435 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12437 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12438 check type for aarch64_simd_valid_immediate.
12439 (aarch64_output_simd_mov_immediate): Update prototype.
12440 (aarch64_simd_valid_immediate): Update prototype.
12441 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12442 support for ORR-immediate.
12443 (and<mode>3): modified pattern to add support for BIC-immediate.
12444 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12445 now checks for valid immediate for BIC and ORR based on new enum
12447 (aarch64_output_simd_mov_immediate): Function now used to output
12448 BIC/ORR imm as well based on new enum argument.
12449 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12451 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12452 (aarch64_reg_or_bic_imm): Likewise.
12454 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12456 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12457 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12459 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12463 PR rtl-optimization/82396
12464 * haifa-sched.c (autopref_multipass_init): Simplify
12466 (autopref_rank_data): Simplify sort order.
12467 * sched-int.h (autopref_multipass_data_): Remove
12468 multi_mem_insn_p, min_offset and max_offset.
12470 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12472 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12474 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12476 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12477 vect_doubleint_cvt.
12479 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12481 * doc/sourcebuild.texi: Document vect_long_mult.
12483 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12485 PR tree-optimization/82413
12486 * fold-const.c (build_range_check): Use widest_int when comparing
12487 the maximum ETYPE value with HIGH.
12489 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12491 PR rtl-optimization/82396
12492 * haifa-sched.c (autopref_multipass_init): Simplify
12494 (autopref_rank_data): Simplify sort order.
12495 * sched-int.h (autopref_multipass_data_): Remove
12496 multi_mem_insn_p, min_offset and max_offset.
12498 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12500 PR tree-optimization/82381
12501 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12502 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12505 PR tree-optimization/82374
12506 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12507 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12508 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12509 current_function_decl to the new decl.
12511 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12513 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12514 helper macro for IEEE float128 hardware built-in functions.
12515 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12517 (TRUNCF128_ODD): Likewise.
12518 (ADDF128_ODD): Likewise.
12519 (SUBF128_ODD): Likewise.
12520 (MULF128_ODD): Likewise.
12521 (DIVF128_ODD): Likewise.
12522 (FMAF128_ODD): Likewise.
12523 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12524 UNSPEC_TRUNC_ROUND_TO_ODD.
12525 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12526 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12527 floating point round to odd instructions.
12528 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12529 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12530 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12531 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12532 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12533 (trunc<mode>sf2_hw): Change the truncate with round to odd
12534 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12535 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12536 to odd hardware instructions.
12537 (sub<mode>3_odd): Likewise.
12538 (mul<mode>3_odd): Likewise.
12539 (div<mode>3_odd): Likewise.
12540 (sqrt<mode>2_odd): Likewise.
12541 (fma<mode>4_odd): Likewise.
12542 (fms<mode>4_odd): Likewise.
12543 (nfma<mode>4_odd): Likewise.
12544 (nfms<mode>4_odd): Likewise.
12545 (trunc<mode>df2_odd): Change the truncate with round to odd
12546 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12548 * doc/extend.texi (PowerPC built-in functions): Update documentation
12549 for existing IEEE float128-bit built-in functions. Add built-in
12550 functions that generate the IEEE 128-bit floating point round to
12553 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12555 PR rtl-optimization/77729
12556 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12557 to (X&(C1&~C2))|C2 transformations.
12559 2017-10-03 Martin Jambor <mjambor@suse.cz>
12561 PR tree-optimization/82363
12562 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12563 mismatch, mark lacc written regardless of racc.
12565 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12567 PR tree-optimization/82381
12568 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12569 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12570 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12571 fallthrough into reassoc_stmt_dominates_stmt_p.
12574 * combine.c (combine_instructions): Don't combine in unreachable
12577 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12580 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12581 function to not use the have_cpu variable. Do not set cpu_index,
12582 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12583 or the default cpu.
12584 (rs6000_valid_attribute_p): Remove duplicate initializations of
12585 old_optimize and func_optimize.
12586 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12587 (rs6000_activate_target_options): Make global.
12588 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12591 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12593 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12594 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12595 Set *poffset to 0 and *psize and *pmax_size to -1 if
12596 *poffset + *psize overflows in HOST_WIDE_INT.
12598 PR tree-optimization/82387
12599 PR tree-optimization/82388
12600 PR tree-optimization/82389
12601 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12602 instead of live_bytes non-NULL.
12604 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12607 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12610 2017-10-02 Richard Biener <rguenther@suse.de>
12612 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12613 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12614 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12616 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12618 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12619 requested precision matches the type's.
12620 * calls.c (alloc_max_size): Calculate the new candidate size as
12621 a widest_int and use wi::to_widest when comparing it with the
12622 current candidate size.
12623 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12624 zero rather than integer_zero_node.
12625 * match.pd: Check for a no-op conversion before using wi::add
12626 rather than after. Use tree_to_uhwi when summing small shift
12627 counts into an unsigned int.
12629 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12634 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12635 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12636 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12637 POINTER_AND_FP_REGS.
12639 2017-10-02 Richard Biener <rguenther@suse.de>
12641 PR tree-optimization/82355
12642 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12643 a mapping for the enclosing loop but avoid generating one for
12644 the loop tree root.
12645 (copy_bb_and_scalar_dependences): Remove premature codegen
12646 error on PHIs in blocks duplicated into multiple places.
12647 * graphite-scop-detection.c
12648 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12649 in the region use it as loop and nest to analyze the DR in.
12650 (try_generate_gimple_bb): Likewise.
12651 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12652 (add_loop_constraints): For blocks in a loop not in the region
12653 create a dimension with a single iteration.
12654 * sese.h (gbb_loop_at_index): Remove assert.
12656 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12658 * omp-expand.c (adjust_context_scope): New function.
12659 (expand_parallel_call): Call adjust_context_scope.
12661 2017-10-01 Jeff Law <law@redhat.com>
12663 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12664 dom_opt_dom_walker class with direct access to private members.
12665 Add comments. Call test_for_singularity.
12666 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12667 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12668 m_dummy_cond anymore.
12669 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12671 (pass_dominator:execute): Build the dummy_cond here and pass it
12672 to the dom_opt_dom_walker ctor.
12673 (test_for_singularity): New function.
12675 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12676 Maya Rashish <coypu@sdf.org>
12678 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12679 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12680 (alpha*-*-netbsd*) Use nbsd_tm_file.
12681 (arm*-*-netbsdelf*) Likewise.
12682 (i[34567]86-*-netbsdelf*) Likewise.
12683 (x86_64-*-netbsd*) Likewise.
12684 (mips*-*-netbsd*) Likewise.
12685 (powerpc-*-netbsd*) Likewise.
12686 (sh*-*-netbsd*) Likewise.
12687 (sparc-*-netbsdelf*) Likewise.
12688 (sparc64-*-netbsd*) Likewise.
12689 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12691 (vax-*-netbsdelf*) Likewise.
12692 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12693 (UINT_FAST8_TYPE) Likewise.
12694 (INT_FAST16_TYPE) Check CHAR_FAST16.
12695 (UINT_FAST16_TYPE) Likewise.
12697 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12700 * config/i386/i386.md
12701 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12702 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12703 *divmodsi4_zext_2): New define_insn_and_split.
12704 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12705 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12706 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12707 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12708 New define_insn_and_split.
12709 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12710 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12711 operands[1] having DImode when mode is SImode.
12713 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12714 always SImode for DIV and MOD in REG_EQUAL notes.
12716 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12718 PR middle-end/82319
12719 * match.pd: Fix handling of NaNs in pattern.
12721 2017-09-29 Jeff Law <law@redhat.com>
12723 * sbitmap.c (bitmap_bit_in_range_p): New function.
12724 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12725 * tree-ssa-dse.c (live_bytes_read): New function.
12726 (dse_classify_store): Ignore reads of dead bytes.
12728 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12729 typos and whitespace errors.
12730 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12731 * config/s390/s390.c (s390_emit_prologue): Likewise.
12733 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12736 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12737 with a symbol for LRA.
12739 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12741 PR rtl-optimization/82338
12742 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12744 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12746 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12747 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12748 (qsort_chk): Declare.
12749 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12750 (qsort_chk): New function.
12752 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12754 PR tree-optimization/82337
12755 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12756 phi definition if the PHI result appears in an abnormal PHI.
12757 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12758 basis appears in an abnormal PHI.
12760 2017-09-29 Richard Biener <rguenther@suse.de>
12762 * graphite-isl-ast-to-gimple.c
12763 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12764 (binary_op_to_tree): Use it.
12765 (get_rename_from_scev): Likewise.
12766 (copy_loop_phi_nodes): Likewise.
12767 (copy_bb_and_scalar_dependences): Likewise.
12768 (translate_pending_phi_nodes): Likewise.
12770 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12773 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12774 for movabsq $(i32 << shift), r64.
12776 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12778 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12779 index when encoding %esp as %rsp to avoid 0x67 prefix.
12781 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12783 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12784 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12786 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12788 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12789 Extensions with more than 16 double VFP registers.
12790 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12791 to_clear_mask and all code related to it. Replace the remaining
12792 entry by a sbitmap and adapt code accordingly.
12794 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12796 * brig-builtins.def: Change pure attributes to const.
12798 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12800 * config.gcc (default_gnu_indirect_function): Default to yes for
12801 sparc*-*-linux* with glibc.
12803 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12805 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12806 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12807 when creating .init_array and .fini_array sections with priority
12810 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12813 * config/aarch64/aarch64.c
12814 (aarch64_builtin_support_vector_misalignment): Always return false
12815 when misalignment is unknown.
12817 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12819 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12820 this function to return false if the definition used by the swap
12821 instruction is artificial, or if the memory address from which the
12822 constant value is loaded is not represented by a base address held
12823 in a register or if the base address register is a frame or stack
12824 pointer. Additionally, return false if the base address of the
12825 loaded constant is a SYMBOL_REF but is not considered to be a
12827 (replace_swapped_load_constant): New function.
12828 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12829 loaded constant vector with a load of a swapped constant vector.
12831 2017-09-27 Carl Love <cel@us.ibm.com>
12833 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12834 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12835 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12838 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12840 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12841 first, always call autopref_rank_data otherwise.
12843 2017-09-27 Richard Biener <rguenther@suse.de>
12845 * graphite-scop-detection.c (find_scop_parameters): Move
12846 loop bound handling ...
12847 (gather_bbs::before_dom_children): ... here, avoiding the need
12848 to build scop_info->loop_nest.
12849 (record_loop_in_sese): Remove.
12850 * sese.h (sese_info_t::loop_nest): Remove.
12851 * sese.c (new_sese_info): Do not allocate loop_nest.
12852 (free_sese_info): Do not free loop_nest.
12854 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12857 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12858 lhs from calls if the lhs has addressable type.
12860 2017-09-27 Richard Biener <rguenther@suse.de>
12862 * graphite.h (scop::max_alias_set): New member.
12863 * graphite-scop-detection.c: Remove references to non-existing
12864 --param in comments.
12865 (build_alias_sets): Record the maximum alias set used for drs.
12866 (build_scops): Support zero as unlimited for
12867 --param graphite-max-arrays-per-scop.
12868 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12869 and inline into ...
12870 (build_poly_sr_1): ... here. Compute alias set based on the
12871 maximum alias set used for drs rather than
12872 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12874 2017-09-27 Richard Biener <rguenther@suse.de>
12876 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12877 --param loop-block-tile-size=0 to disable tiling.
12879 2017-09-27 Richard Biener <rguenther@suse.de>
12881 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12882 (graphite-max-nb-scop-params): Document special value zero.
12883 * domwalk.h (dom_walker::STOP): New symbolical constant.
12884 (dom_walker::dom_walker): Add optional parameter for bb to
12886 (dom_walker::~dom_walker): Declare.
12887 (dom_walker::before_dom_children): Document STOP return value.
12888 (dom_walker::m_user_bb_to_rpo): New member.
12889 (dom_walker::m_bb_to_rpo): Likewise.
12890 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12891 mapping here if not provided by the user.
12892 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12893 provided by the user.
12894 (dom_walker::STOP): Define.
12895 (dom_walker::walk): Do not compute bb to RPO mapping here.
12896 Support STOP return value from before_dom_children to stop
12898 * graphite-optimize-isl.c (optimize_isl): If the schedule
12899 is the same still generate code if -fgraphite-identity
12900 or -floop-parallelize-all are given.
12901 * graphite-scop-detection.c: Include cfganal.h.
12902 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12904 (gather_bbs::before_dom_children): Return STOP for BBs
12906 (build_scops): Compute bb to RPO mapping and pass it to
12907 the domwalk. Treat --param graphite-max-nb-scop-params=0
12908 as not limiting the number of params.
12909 * graphite.c (graphite_initialize): Remove limit on the
12910 number of basic-blocks in a function.
12911 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12912 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12913 default value of 10.
12915 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12917 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12918 Adjust code to eliminate needing to do the shift right 32-bits
12919 operation after XSCVDPSPN.
12921 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12923 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12924 ((X / Y) != 0 -> X >= Y): Likewise.
12926 2017-09-26 Carl Love <cel@us.ibm.com>
12928 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12929 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12930 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12931 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12932 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12933 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12934 definitions and overloading.
12935 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12936 statement for P9V_BUILTIN_XST_LEN_R.
12937 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12938 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12939 define_expand and define_insn for the instructions and builtins.
12940 * doc/extend.texi: Update the built-in documentation file for the new
12941 built-in functions.
12942 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12943 define_insn for the instructions
12945 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12948 * gcc/config/netbsd-protos.h: New file.
12949 * gcc/config/netbsd.c: New file.
12950 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12951 * gcc/config/t-netbsd: New file.
12952 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12953 (tmake_file) Add t-netbsd.
12954 (extra_objs) Add netbsd.o.
12956 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12960 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12962 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12964 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12965 sign extension from a vector register to a GPR by doing a 32-bit
12966 direct move and then an EXTSW.
12967 (extendsi<mode>2 splitter): Likewise.
12968 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12969 right or vector extract after doing XSCVDPSPN. Use
12970 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12972 (movdi_from_sf_zero_ext): Likewise.
12973 (reload_gpr_from_vsxsf): Likewise.
12974 (p8_mfvsrd_4_disf): Delete, no longer used.
12975 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12976 then needing to move the SFmode to a GPR to use the XSCVDPSP
12977 instruction instead of FRSP and XSCVDPSPN.
12978 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12979 it is adjacent to the other XSCVSPDP insns.
12980 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12981 SFmode to be in traditional Altivec registers.
12982 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12983 (vsx_xscvspdpn): Likewise.
12984 (vsx_xscvspdpn_scalar): Likewise.
12986 2017-09-26 Martin Jambor <mjambor@suse.cz>
12988 * tree-sra.c (compare_access_positions): Put integral types first,
12989 stabilize sorting of integral types, remove conditions putting
12990 non-full-precision integers last.
12991 (sort_and_splice_var_accesses): Disable scalarization if a
12992 non-integert would be represented by a non-full-precision integer.
12994 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12996 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12997 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12998 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12999 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13000 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13001 conditionals inside the function instead of around it. Call
13002 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13003 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13005 2017-09-26 Richard Biener <rguenther@suse.de>
13007 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13009 (scop_detection::build_scop_breadth): ... this. Removed.
13010 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13011 (scop_detection::harmful_stmt_in_bb): Likewise.
13012 (scop_detection::graphite_can_represent_stmt): Likewise.
13013 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13014 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13015 (scop_detection::can_represent_loop_1): ... this. Removed.
13016 (scop_detection::harmful_loop_in_region): Simplify after inlining
13017 the above and remove more quadraticness.
13018 (build_scops): Adjust.
13019 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13022 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13025 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13026 REGNO (base) == SP_REG if base is a REG.
13028 PR middle-end/35691
13029 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13030 if it is different SSA_NAME.
13031 (optimize_range_tests_cmp_bitwise): New function.
13032 (optimize_range_tests): Call it.
13034 2017-09-26 Richard Biener <rguenther@suse.de>
13036 PR tree-optimization/82321
13037 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13038 for the def being inside the loop.
13040 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13042 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13044 * config/s390/s390-builtins.def: Fix constraint on op4.
13046 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13048 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13049 independent expanders.
13050 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13051 ("vec_ordered", "vec_unordered"): New expanders.
13053 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13055 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13058 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13060 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13061 vec_unpacks_lo_v16qi.
13062 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13064 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13066 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13067 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13068 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13070 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13072 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13074 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13076 ("*vec_slb<mode>"): New insn pattern.
13077 ("vec_shr_<mode>"): New expander.
13078 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13079 and force the shift count operand to V16QImode.
13080 ("vec_srb<mode>"): Set shift count mode to V16QI.
13082 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13084 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13085 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13086 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13088 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13091 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13093 2017-09-26 Richard Biener <rguenther@suse.de>
13095 PR tree-optimization/82320
13096 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13099 2017-09-25 Jeff Law <law@redhat.com>
13101 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13102 prototype for new argument.
13103 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13104 mostly extracted from rs6000_emit_allocate_stack.
13105 (rs6000_emit_probe_stack_range_stack_clash): New function.
13106 (rs6000_emit_allocate_stack): Call
13107 rs6000_emit_probe_stack_range_stack_clash as needed.
13108 (rs6000_emit_probe_stack_range): Add additional argument
13109 to call to gen_probe_stack_range{si,di}.
13110 (output_probe_stack_range): New.
13111 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13112 (output_probe_stack_range_stack_clash): New.
13113 (rs6000_emit_prologue): Emit notes into dump file as requested.
13114 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13115 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13116 Add additional operand and pass it to output_probe_stack_range.
13118 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13120 PR tree-optimization/82163
13121 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13122 (checking_verify_loop_closed_ssa): New parameter.
13123 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13124 (check_loop_closed_ssa_stmt): Delete.
13125 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13126 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13127 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13130 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13132 * brig-builtins.def: Treat HSAIL barrier builtins as
13133 setjmp/longjump style functions.
13135 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13137 * target.def (constant_alignment): New hook.
13138 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13139 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13140 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13141 * doc/tm.texi: Regenerate.
13142 * targhooks.h (default_constant_alignment): Declare.
13143 (constant_alignment_word_strings): Likewise.
13144 * targhooks.c (default_constant_alignment): New function.
13145 (constant_alignment_word_strings): Likewise.
13146 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13147 instead of CONSTANT_ALIGNMENT.
13148 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13149 (force_const_mem): Likewise.
13150 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13151 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13152 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13153 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13154 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13156 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13157 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13158 constant_alignment_word_strings.
13159 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13160 (CONSTANT_ALIGNMENT): Likewise.
13161 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13162 (arm_constant_alignment): New function.
13163 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13164 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13165 constant_alignment_word_strings.
13166 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13167 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13168 constant_alignment_word_strings.
13169 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13170 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13171 (cris_constant_alignment): New function.
13172 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13173 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13174 (epiphany_constant_alignment): New function.
13175 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13176 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13177 constant_alignment_word_strings.
13178 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13179 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13180 constant_alignment_word_strings.
13181 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13182 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13183 constant_alignment_word_strings.
13184 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13185 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13186 * config/i386/i386.c (ix86_constant_alignment): Make static.
13187 Use the same interface as the target hook.
13188 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13189 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13190 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13191 constant_alignment_word_strings.
13192 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13193 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13194 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13195 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13196 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13197 constant_alignment_word_strings.
13198 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13199 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13200 constant_alignment_word_strings.
13201 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13202 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13203 constant_alignment_word_strings.
13204 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13205 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13207 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13208 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13209 * config/mips/mips.c (mips_constant_alignment): New function.
13210 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13211 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13212 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13213 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13214 (mmix_constant_alignment): Make static. Use the same interface
13215 as the target hook.
13216 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13217 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13218 constant_alignment_word_strings.
13219 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13220 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13221 constant_alignment_word_strings.
13222 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13223 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13224 constant_alignment_word_strings.
13225 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13226 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13227 (rs6000_constant_alignment): New function.
13228 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13229 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13231 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13232 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13233 (rs6000_constant_alignment): New function.
13234 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13235 * config/s390/s390.c (s390_constant_alignment): New function.
13236 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13237 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13238 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13239 constant_alignment_word_strings.
13240 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13241 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13242 (sparc_constant_alignment): New function.
13243 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13244 * config/spu/spu.c (spu_constant_alignment): New function.
13245 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13246 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13247 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13248 constant_alignment_word_strings.
13249 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13250 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13251 constant_alignment_word_strings.
13252 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13253 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13254 constant_alignment_word_strings.
13255 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13256 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13257 (visium_constant_alignment): New function.
13258 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13259 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13260 (xtensa_constant_alignment): New function.
13261 * system.h (CONSTANT_ALIGNMENT): Poison.
13263 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13266 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13267 (rs6000_builtin_valid_without_lhs): New helper function.
13268 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13269 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13271 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13273 * target.h (vec_perm_indices): Use unsigned short rather than
13275 (auto_vec_perm_indices): Likewise.
13276 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13277 Use unsigned int rather than unsigned char.
13278 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13280 2017-09-25 Richard Biener <rguenther@suse.de>
13282 * cfgloop.h (sort_sibling_loops): Declare.
13283 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13284 (sort_sibling_loops): New function sorting the sibling loop list
13286 * graphite.c (graphite_transform_loops): Sort sibling loops.
13288 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13290 * target.def (vec_perm_const_ok): Change sel parameter to
13292 * optabs-query.c (can_vec_perm_p): Update accordingly.
13293 * doc/tm.texi: Regenerate.
13294 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13295 auto_vec_perm_indices and remove separate nelt field.
13296 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13297 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13298 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13299 (aarch64_expand_vec_perm_const): Update accordingly.
13300 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13301 to vec_perm_indices.
13302 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13303 auto_vec_perm_indices and remove separate nelt field.
13304 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13305 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13306 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13308 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13309 to vec_perm_indices.
13310 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13311 sel to vec_perm_indices.
13312 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13313 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13314 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13316 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13319 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13322 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13323 on the FUNCTION_DECL function context if it has a DIE that is a
13326 2017-09-25 Richard Biener <rguenther@suse.de>
13328 PR tree-optimization/82285
13329 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13332 2017-09-25 Tom de Vries <tom@codesourcery.com>
13336 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13339 2017-09-25 Richard Biener <rguenther@suse.de>
13341 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13342 ISL errors other than isl_error_quota happen. Dump if the
13343 schedule is the same.
13344 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13345 errors instead of aborting inside ISL.
13347 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13350 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13351 of libgcc_eh for m64.
13352 * config/i386/darwin64.h: Likewise.
13354 2017-09-25 Richard Biener <rguenther@suse.de>
13356 PR middle-end/82144
13357 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13358 attribute for incomplete types nor twice for complete ones.
13360 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13363 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13364 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13367 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13370 * config/i386/i386.opt (mprefer-avx256): Use
13371 ix86_target_flags variable.
13372 * config/i386/i386.c (ix86_target_string): Move
13373 -mprefer-avx256 to flag2_opts.
13375 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13377 PR middle-end/35691
13378 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13379 and x != -1 | y != -1 into (x & y) != -1.
13381 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13383 * config.gcc: Add new case statement to set
13384 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13385 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13386 s390x-*-linux* case statements. Added aarch64 to the list of
13387 supported architectures.
13389 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13391 PR tree-optimization/82289
13392 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13393 STMT_VINFO_RELEVANT_P.
13395 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13396 Alan Hayward <alan.hayward@arm.com>
13397 David Sherwood <david.sherwood@arm.com>
13399 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13400 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13401 (extract_range_from_binary_expr_1): Don't call
13402 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13404 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13405 Alan Hayward <alan.hayward@arm.com>
13406 David Sherwood <david.sherwood@arm.com>
13408 * target.def (preferred_vector_alignment): New hook.
13409 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13411 * doc/tm.texi: Regenerate.
13412 * targhooks.h (default_preferred_vector_alignment): Declare.
13413 * targhooks.c (default_preferred_vector_alignment): New function.
13414 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13416 (DR_TARGET_ALIGNMENT): New macro.
13417 (aligned_access_p): Update commentary.
13418 (vect_known_alignment_in_bytes): New function.
13419 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13421 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13422 Calculate the misalignment based on the target alignment rather than
13424 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13425 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13426 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13427 the target alignment, rather than masking the element misalignment
13428 with the number of elements in a vector. Also use the target
13429 alignment when calculating the maximum number of peels.
13430 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13431 instead of TYPE_ALIGN_UNIT.
13432 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13433 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13434 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13435 (vect_create_data_ref_ptr): Likewise.
13436 (vect_setup_realignment): Realign by ANDing with
13437 -DR_TARGET_MISALIGNMENT.
13438 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13439 the number of peels based on DR_TARGET_ALIGNMENT.
13440 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13441 with the guaranteed alignment boundary when deciding whether
13443 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13444 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13445 (ensure_base_align): Remove stmt_info parameter. Get the
13446 target base alignment from DR_TARGET_ALIGNMENT.
13447 (vectorizable_store): Update call accordingly. Interpret
13448 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13450 (vectorizable_load): Likewise.
13452 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13453 Alan Hayward <alan.hayward@arm.com>
13454 David Sherwood <david.sherwood@arm.com>
13456 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13457 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13458 (vect_enhance_data_refs_alignment): Likewise.
13460 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13462 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13463 error. Only quit immediately if parsing is complete.
13464 (BEGIN): Initialize fatal_err and parse_done.
13465 (begin fpu, end fpu): Check number of arguments.
13466 (begin arch, end arch): Likewise.
13467 (begin cpu, end cpu): Likewise.
13468 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13469 (optalias): Likewise.
13471 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13473 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13474 * config/arm/arm-isa.h: Delete. Move definitions to ...
13475 * arm-cpus.in: ... here. Use new feature and fgroup values.
13476 * config/arm/arm.c (arm_option_override): Use lower case for feature
13478 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13479 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13480 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13482 (print_isa_bits_for): New function.
13483 (gen_isa): New function.
13484 (gen_comm_data): Use print_isa_bits_for.
13485 (define feature): New keyword.
13486 (define fgroup): New keyword.
13487 * config/arm/t-arm (TM_H): Remove.
13488 (GTM_H): Add arm-isa.h.
13489 (arm-isa.h): Add rule to generate file.
13490 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13491 case for feature bit names.
13493 2017-09-22 Richard Biener <rguenther@suse.de>
13495 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13497 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13498 print of no dependency loops ...
13499 * graphite.c (graphite_transform_loops): ... here.
13500 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13502 (same_close_phi_node, remove_duplicate_close_phi,
13503 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13504 (canonicalize_loop_closed_ssa): ... here and simplify.
13505 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13506 (optimize_isl): Use dump_printf_loc to tell when we stopped
13507 optimizing because of an ISL timeout.
13509 2017-09-22 Richard Biener <rguenther@suse.de>
13511 PR tree-optimization/82291
13512 * tree-if-conv.c (predicate_mem_writes): Make sure to
13513 remove writes in blocks predicated with false.
13515 2017-09-22 Richard Biener <rguenther@suse.de>
13517 * sese.c: Include cfganal.h.
13518 (if_region_set_false_region): Remove.
13519 (create_if_region_on_edge): Likewise.
13520 (move_sese_in_condition): Re-implement without destroying
13523 2017-09-22 Richard Biener <rguenther@suse.de>
13525 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13526 Verify both BBs contain loop PHI nodes before dispatching to
13527 copy_loop_phi_args.
13528 (graphite_regenerate_ast_isl): Do not recompute dominators,
13529 do not verify three times. Restructure for clarity.
13530 * graphite-scop-detection.c (same_close_phi_node,
13531 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13532 defined_in_loop_p, canonicalize_loop_closed_ssa,
13533 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13534 checking and SSA rewrite, move to ...
13535 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13536 (graphite_initialize): Do not pass in ctx, do not reset the
13537 SCEV cache, compute only dominators.
13538 (graphite_transform_loops): Allocate ISL ctx after
13539 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13540 Maintain post-dominators only around build_scops.
13541 * sese.c (if_region_set_false_region): Make static. Free
13542 and recompute dominators.
13543 (move_sese_in_condition): Assert we don't get called with
13544 post-dominators computed.
13545 * sese.h (if_region_set_false_region): Remove.
13547 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13549 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13550 mode attribute for TARGET_AVX512VL.
13552 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13554 * config/i386/i386.opt (mprefer-avx256): New option.
13555 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13557 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13558 for TARGET_PREFER_AVX256.
13559 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13561 2017-09-21 Jeff Law <law@redhat.com>
13563 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13564 Fix dump output if the only stack space is for pushed registers.
13566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13568 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13571 2017-09-21 Martin Sebor <msebor@redhat.com>
13574 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13575 code (in C++) or code that triggers warnings.
13577 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13579 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13581 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13583 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13584 * sched-rgn.c: Ditto.
13585 * sel-sched-ir.c: Ditto.
13587 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13589 * toplev.h (set_random_seed): Adjust return type.
13590 * toplev.c (init_local_tick): Move eager initialization of random_seed
13591 to get_random_seed. Adjust comment.
13592 (init_random_seed): Inline to get_random_seed, delete.
13593 (get_random_seed): Initialize random_seed lazily.
13594 (set_random_seed): Do not return previous value.
13595 (print_switch_value): Do not call get_random_seed.
13597 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13599 * cgraph.c (delete_function_version): New, broken out from...
13600 (cgraph_node::delete_function_version): ...here. Rename to
13601 cgraph_node::delete_function_version_by_decl. Update all uses.
13602 (cgraph_node::remove): Call delete_function_version.
13604 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13607 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13608 VAR_DECLs to which addressable non-volatile parameters are mapped
13609 and for id->retvar after the return value assignment. Clear
13610 id->retval and id->retbnd after inlining.
13612 2017-09-21 Richard Biener <rguenther@suse.de>
13614 PR tree-optimization/82276
13615 PR tree-optimization/82244
13616 * tree-vrp.c (build_assert_expr_for): Set
13617 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13619 (remove_range_assertions): Revert earlier change.
13621 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13624 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13626 2017-09-21 Richard Biener <rguenther@suse.de>
13628 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13629 Restore valid IL after code generation errors.
13630 * graphite.c (graphite_transform_loops): Diagnose code
13631 generation issues as MSG_MISSED_OPTIMIZATION and continue
13632 with processing SCOPs.
13634 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13635 Alan Hayward <alan.hayward@arm.com>
13636 David Sherwood <david.sherwood@arm.com>
13638 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13639 rather than choosing between plus_constant and gen_rtx_<CODE>.
13640 * expr.c (emit_push_insn): Likewise.
13641 (expand_expr_real_2): Likewise.
13643 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13644 Alan Hayward <alan.hayward@arm.com>
13645 David Sherwood <david.sherwood@arm.com>
13647 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13649 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13650 Alan Hayward <alan.hayward@arm.com>
13651 David Sherwood <david.sherwood@arm.com>
13653 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13654 calling tree_to_uhwi.
13656 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13657 Alan Hayward <alan.hayward@arm.com>
13658 David Sherwood <david.sherwood@arm.com>
13660 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13661 INTEGER_CST rather than a negative test for ADDR_EXPR.
13663 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13664 Alan Hayward <alan.hayward@arm.com>
13665 David Sherwood <david.sherwood@arm.com>
13667 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13668 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13670 2017-09-21 Richard Biener <rguenther@suse.de>
13672 PR tree-optimization/71351
13673 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13674 graphite_create_new_loop_guard): Remove, fold remaining parts
13676 (translate_isl_ast_node_for): ... here and simplify.
13678 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13681 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13682 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13683 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13684 alternative always use QI mode, for -Os imov (=R,R) alternative
13685 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13688 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13689 Jeff Law <law@redhat.com>
13691 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13692 (allocate_stack_space): New function, partially extracted from
13693 s390_emit_prologue.
13694 (s390_emit_prologue): Track offset to most recent stack probe.
13695 Code to allocate space moved into allocate_stack_space.
13696 Dump actions when no stack is allocated.
13697 (s390_prologue_plus_offset): New function.
13698 (s390_emit_stack_probe): Likewise.
13700 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13702 * common.opt (Wa, Wl, Wp, g, gz=): Add
13704 (gno-column-info): Remove.
13705 (gcolumn-info): Drop RejectNegative.
13706 (gno-): New prefix.
13707 (gno-record-gcc-switches): Remove.
13708 (grecord-gcc-switches): Drop RejectNegative.
13709 (gno-split-dwarf): Remove.
13710 (gsplit-dwarf): Drop RejectNegative.
13711 (gno-strict-dwarf): Remove.
13712 (gstrict-dwarf): Drop RejectNegative.
13713 * config/darwin.opt (gfull, gused): Add RejectNegative.
13714 * dwarf2out.c (gen_producer_string): Drop
13715 gno-record-gcc-switches handler.
13716 * optc-gen.awk: Add g to prefixes with negative forms.
13717 * opts-common.c (remapping_prefix_p): New.
13718 (find_opt): Check it.
13719 (generate_canonical_option): Test g prefix.
13720 (option_map): Add -gno- mapping.
13721 (add_misspelling_candidates): Check remapping_prefix_p.
13723 2017-09-20 Jeff Law <law@redhat.com>
13725 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13726 thinko in stack clash protection support.
13728 * explow.c (compute_stack_clash_protection_loop_data): Use
13729 CONST_INT_P instead of explicit test. Verify object is a
13730 CONST_INT_P before looking at INTVAL.
13731 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13732 instead of explicit test.
13734 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13737 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13738 address instead of to r1 and r11.
13740 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13742 * config.gcc: Support "knm".
13743 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13744 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13746 * config/i386/i386.c (m_KNM): Define.
13747 (processor_target_table): Add "knm".
13749 (ix86_option_override_internal): Add "knm".
13750 (ix86_issue_rate): Add PROCESSOR_KNM.
13751 (ix86_adjust_cost): Ditto.
13752 (ia32_multipass_dfa_lookahead): Ditto.
13753 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13754 (fold_builtin_cpu): Add M_INTEL_KNM.
13755 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13756 (processor_type): Add PROCESSOR_KNM.
13757 * config/i386/x86-tune.def: Add m_KNM.
13758 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13760 2017-09-20 Richard Biener <rguenther@suse.de>
13762 PR tree-optimization/80213
13763 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13764 are allowed in empty BBs as well.
13765 (canonicalize_loop_closed_ssa): Also look for other complex
13767 (scop_detection::get_sese): Include the loop-closed PHI block
13769 (scop_detection::merge_sese): Remove code adding extra blocks.
13770 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13771 (build_scops): Assert the final returned scop is invalid.
13773 2017-09-20 Richard Biener <rguenther@suse.de>
13775 PR tree-optimization/82264
13776 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13778 (vn_phi_lookup): Likewise.
13779 (vn_phi_insert): Likewise.
13781 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13783 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13784 that fits into uhwi or shwi, add DW_AT_const_value regardless
13785 of early_dwarf without going through RTL, using add_AT_unsigned
13788 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13789 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13790 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13791 (DEBUG_MACRO_SECTION): Likewise.
13792 (DEBUG_LTO_MACRO_SECTION): Likewise.
13793 (DEBUG_STR_DWO_SECTION): Likewise.
13794 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13795 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13796 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13797 (DEBUG_LTO_LINE_STR_SECTION): Define.
13798 (init_sections_and_labels): Initialize debug_line_str_section
13799 variable. Initialize debug_loc_section for -gdwarf-5 to
13800 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13802 2017-09-20 Richard Biener <rguenther@suse.de>
13804 * graphite-sese-to-poly.c (extract_affine): Properly handle
13805 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13807 2017-09-20 Richard Biener <rguenther@suse.de>
13809 PR tree-optimization/81373
13810 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13811 Force SESE live-out defs to be handled even if they are
13814 2017-09-19 Jeff Law <law@redhat.com>
13816 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13817 nothing for stack adjustments with REG_STACK_CHECK.
13818 * sched-deps.c (parse_add_or_inc): Reject insns with
13819 REG_STACK_CHECK from dependency breaking.
13820 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13821 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13822 * reg-notes.def (STACK_CHECK): New note.
13824 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13825 (ix86_expand_prologue): Dump stack clash info as needed.
13826 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13828 * function.c (dump_stack_clash_frame_info): New function.
13829 * function.h (dump_stack_clash_frame_info): Prototype.
13830 (enum stack_clash_probes): New enum.
13832 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13833 flag_stack_clash_protection.
13834 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13835 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13836 (arm_frame_pointer_required): Likewise.
13837 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13838 (ia64_expand_prologue): Likewise.
13839 * config/mips/mips.c (mips_expand_prologue): Likewise.
13840 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13841 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13842 (sparc_flat_expand_prologue): Likewise.
13843 * config/spu/spu.c (spu_expand_prologue): Likewise.
13845 * explow.c: Include "params.h".
13846 (anti_adjust_stack_and_probe_stack_clash): New function.
13847 (get_stack_check_protect): Likewise.
13848 (compute_stack_clash_protection_loop_data): Likewise.
13849 (emit_stack_clash_protection_loop_start): Likewise.
13850 (emit_stack_clash_protection_loop_end): Likewise.
13851 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13852 Use anti_adjust_stack_and_probe_stack_clash.
13853 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13854 (emit_stack_clash_protection_loop_start): Likewise.
13855 (emit_stack_clash_protection_loop_end): Likewise.
13856 * rtl.h (get_stack_check_protect): Prototype.
13857 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13858 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13859 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13861 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13863 * doc/tm.texi: Rebuilt.
13864 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13865 get_stack_check_protect.
13866 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13867 * config/arm/arm.c (arm_expand_prologue): Likewise.
13868 (arm_frame_pointer_required): Likewise.
13869 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13870 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13871 * config/mips/mips.c (mips_expand_prologue): Likewise.
13872 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13873 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13874 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13875 (sparc_flat_expand_prologue): Likewise.
13877 * common.opt (-fstack-clash-protection): New option.
13878 * flag-types.h (enum stack_check_type): Note difference between
13879 -fstack-check= and -fstack-clash-protection.
13880 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13881 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13882 * toplev.c (process_options): Issue warnings/errors for cases
13883 not handled with -fstack-clash-protection.
13884 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13885 (-fstack-check): Note additional problem with -fstack-check=generic.
13886 Note that -fstack-check is primarily for Ada and refer users
13887 to -fstack-clash-protection for stack-clash-protection.
13888 Document new params for stack clash protection.
13890 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13892 * config/i386/i386.c (ix86_split_long_move): Do not handle
13893 address used for LEA in a special way.
13895 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13897 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13899 2017-09-19 Martin Sebor <msebor@redhat.com>
13902 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13903 of incompatible types.
13905 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13908 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13909 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13910 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13912 2017-09-19 Richard Biener <rguenther@suse.de>
13914 PR tree-optimization/82244
13915 * tree-vrp.c (remove_range_assertions): Do not propagate
13916 a constant to abnormals but replace the assert with a copy.
13918 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13920 PR rtl-optimization/57878
13921 PR rtl-optimization/68988
13922 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13923 avoidance test involving non_reload_pseudos. Move frequency test
13924 below the general fragmentation avoidance test.
13926 2017-09-19 Richard Biener <rguenther@suse.de>
13928 PR tree-optimization/69728
13929 * graphite-sese-to-poly.c (schedule_error): New global.
13930 (add_loop_schedule): Handle empty domain by failing the
13932 (build_original_schedule): Handle schedule_error.
13934 2017-09-19 Richard Biener <rguenther@suse.de>
13936 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13937 Do not iterate to sibling loops but only to siblings of inner
13940 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13943 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13944 that the registers are different.
13946 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13948 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13949 to processor_model and "amdfam17h" to arch_names_table.
13950 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13952 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13955 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13957 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13958 Alan Hayward <alan.hayward@arm.com>
13959 David Sherwood <david.sherwood@arm.com>
13961 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13963 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13965 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13966 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13967 vect_schedule_slp_instance.
13968 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13969 Update call to vect_slp_analyze_node_operations. Simplify return
13971 (vect_slp_analyze_bb_1): Update call accordingly.
13972 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13973 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13974 (vect_schedule_slp): Update call accordingly.
13976 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13977 Alan Hayward <alan.hayward@arm.com>
13978 David Sherwood <david.sherwood@arm.com>
13980 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13981 with types that aren't in fact scalar.
13983 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13985 * tree-vect-slp.c (vect_record_max_nunits): New function,
13987 (vect_build_slp_tree_1): ...here.
13988 (vect_build_slp_tree_2): Call it for phis too.
13990 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13992 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13993 to vect_get_vec_def_for_operand when getting the mask operand.
13995 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13996 Alan Hayward <alan.hayward@arm.com>
13997 David Sherwood <david.sherwood@arm.com>
13999 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14002 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14003 Alan Hayward <alan.hayward@arm.com>
14004 David Sherwood <david.sherwood@arm.com>
14006 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14007 calculation for vector booleans.
14009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14010 Alan Hayward <alan.hayward@arm.com>
14011 David Sherwood <david.sherwood@arm.com>
14013 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14015 (vect_transform_stmt): ...here.
14016 (vect_analyze_stmt): Use it instead of calling
14017 vectorizable_live_operation directly.
14019 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14021 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14022 non-SIMT targets in acc vector loops.
14024 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14026 * configure.ac: Add arc and check if assembler supports gdwarf2.
14027 * configure: Regenerate.
14029 2017-09-18 Richard Biener <rguenther@suse.de>
14031 PR tree-optimization/82220
14032 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14033 epilogue niters from the min_profitable_iters compute.
14035 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14038 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14039 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14040 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14042 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14045 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14046 switching to a new text section.
14048 2017-09-18 Richard Biener <rguenther@suse.de>
14050 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14052 (build_alias_set): Reject aliases with no access function.
14054 2017-09-18 Richard Biener <rguenther@suse.de>
14056 PR tree-optimization/79622
14057 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14059 (build_cross_bb_scalars_use): Likewise.
14061 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14063 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14065 2017-09-18 Alan Modra <amodra@gmail.com>
14068 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14069 stack_pointer_rtx for count 0. Update comments. Break up
14070 large rtl expression.
14072 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14074 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14075 Increase to 20 bytes.
14076 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14077 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14078 or avx version of the stub.
14080 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14083 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14084 compute the minimum stack alignment. Also update preferred stack
14085 boundary for leaf functions.
14087 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14089 PR tree-optimization/82228
14090 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14093 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14095 * common/config/nds32/nds32-common.c
14096 (nds32_option_optimization_table): Refine formatting.
14097 (nds32_option_optimization_table): Use -fsched-pressure and
14098 -fomit-frame-pointer for specific optimization level.
14100 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14102 * config/nds32/nds32.c: Refine formatting and comments.
14103 * config/nds32/nds32.h: Likewise.
14104 * config/nds32/nds32.md: Likewise.
14105 * config/nds32/nds32-cost.c: Likewise.
14106 * config/nds32/nds32-isr.c: Likewise.
14107 * config/nds32/nds32-md-auxiliary.c: Likewise.
14108 * config/nds32/nds32-multiple.md: Likewise.
14109 * config/nds32/nds32-predicates.c: Likewise.
14111 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14112 Jakub Jelinek <jakub@redhat.com>
14114 Add support for -std=c++2a.
14115 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14117 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14119 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14122 * doc/extend.texi (Common Function Attributes): Add
14123 references to ARM, AArch64, and S/390 specific attributes.
14124 (Function Specific Option Pragmas): Add AArch64 and S/390
14125 to list of back ends that support the target pragma.
14127 2017-09-15 Nathan Sidwell <nathan@acm.org>
14129 * doc/standards.texi: Fix C++17 description. Update URLs for
14132 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14134 * common.opt (Wcast-align=strict): New warning option.
14135 * doc/invoke.texi: Document -Wcast-align=strict.
14137 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14139 * cgraph.h (cgraph_thunk_info): Add comments.
14140 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14141 assert for VIRTUAL_* arguments stricter.
14143 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14145 PR tree-optimization/71026
14146 * match.pd: Move RDIV patterns from fold-const.c
14147 * fold-const.c (distribute_real_division): Removed.
14148 (fold_binary_loc): Remove calls to distribute_real_divison.
14150 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14152 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14153 c++1z and gnu++1z as deprecated. Change other references to
14154 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14155 Change -Wc++1z-compat to -Wc++17-compat.
14156 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14157 * dwarf2out.c (highest_c_language): Handle C++17.
14158 (gen_compile_unit_die): Likewise.
14160 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14162 PR rtl-optimization/82192
14163 * combine.c (make_extraction): Don't look through non-paradoxical
14164 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14167 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14168 Alan Hayward <alan.hayward@arm.com>
14169 David Sherwood <david.sherwood@arm.com>
14171 * target.def (function_arg_offset): New hook.
14172 * targhooks.h (default_function_arg_offset): Declare.
14173 * targhooks.c (default_function_arg_offset): New function.
14174 * function.c (locate_and_pad_parm): Use
14175 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14176 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14177 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14178 * doc/tm.texi: Regenerate.
14179 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14180 * config/spu/spu.c (spu_function_arg_offset): New function.
14181 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14182 * system.h (FUNCTION_ARG_OFFSET): Poison.
14184 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14185 Alan Hayard <alan.hayward@arm.com>
14186 David Sherwood <david.sherwood@arm.com>
14188 * target.def (truly_noop_truncation): New hook.
14189 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14190 than TRULY_NOOP_TRUNCATION.
14191 * hooks.h (hook_bool_uint_uint_true): Declare.
14192 * hooks.c (hook_bool_uint_uint_true): New function.
14193 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14194 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14195 * doc/tm.texi: Regenerate.
14196 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14197 rather than TRULY_NOOP_TRUNCATION in comments.
14198 (simplify_comparison): Likewise.
14199 (record_truncated_value): Likewise.
14200 * expmed.c (extract_bit_field_1): Likewise.
14201 (extract_split_bit_field): Likewise.
14202 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14203 instead of TRULY_NOOP_TRUNCATION.
14204 * function.c (assign_parm_setup_block): Likewise.
14205 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14206 * rtlhooks.c: Include target.h.
14207 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14208 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14209 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14210 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14211 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14212 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14213 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14214 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14215 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14216 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14217 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14218 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14219 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14220 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14221 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14222 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14223 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14224 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14225 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14226 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14227 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14228 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14229 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14230 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14231 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14232 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14233 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14234 rather than TRULY_NOOP_TRUNCATION in comments.
14235 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14236 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14237 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14238 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14239 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14240 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14241 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14242 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14243 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14244 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14245 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14246 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14247 rather than TRULY_NOOP_TRUNCATION in comments.
14248 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14249 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14250 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14251 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14252 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14253 TRULY_NOOP_TRUNCATION condition.
14254 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14255 (TRULY_NOOP_TRUNCATION): Delete.
14256 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14257 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14258 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14259 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14260 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14261 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14262 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14263 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14264 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14265 rather than TRULY_NOOP_TRUNCATION in comments.
14266 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14267 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14268 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14269 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14270 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14271 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14273 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14276 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14277 (*cmp_ior): Likewise.
14278 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14279 (*ior_scc_scc_cmp): Likewise.
14280 (*and_scc_scc): Likewise.
14281 (*and_scc_scc_cmp): Likewise.
14283 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14284 Alan Hayard <alan.hayward@arm.com>
14285 David Sherwood <david.sherwood@arm.com>
14287 * target.def (can_change_mode_class): New hook.
14288 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14289 (hard_regno_nregs): Likewise.
14290 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14291 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14292 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14293 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14294 (LOAD_EXTEND_OP): Update accordingly.
14295 * doc/tm.texi: Regenerate.
14296 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14297 CANNOT_CHANGE_MODE_CLASS.
14298 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14299 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14300 * combine.c (simplify_set): Update accordingly.
14301 * emit-rtl.c (validate_subreg): Likewise.
14302 * recog.c (general_operand): Likewise.
14303 * regcprop.c (mode_change_ok): Likewise.
14304 * reload1.c (choose_reload_regs): Likewise.
14305 (inherit_piecemeal_p): Likewise.
14306 * rtlanal.c (simplify_subreg_regno): Likewise.
14307 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14308 instead of CANNOT_CHANGE_MODE_CLASS.
14309 (reload_cse_simplify_operands): Likewise.
14310 * reload.c (push_reload): Use targetm.can_change_mode_class
14311 instead of CANNOT_CHANGE_MODE_CLASS.
14312 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14313 REG_CANNOT_CHANGE_MODE_P.
14314 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14315 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14316 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14317 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14318 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14319 (arm_can_change_mode_class): New function.
14320 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14321 than CANNOT_CHANGE_MODE_CLASS in comments.
14322 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14323 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14324 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14325 (ix86_can_change_mode_class): ...this new function, inverting the
14326 sense of the return value.
14327 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14328 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14329 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14330 (ia64_can_change_mode_class): New function.
14331 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14332 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14333 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14334 (m32c_can_change_mode_class): ...this new function, inverting the
14335 sense of the return value.
14336 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14337 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14338 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14339 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14340 (mips_can_change_mode_class): ...this new function, inverting the
14341 sense of the return value.
14342 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14343 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14344 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14345 (msp430_can_change_mode_class): New function.
14346 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14347 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14348 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14349 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14350 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14351 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14352 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14353 (pa_cannot_change_mode_class): Replace with...
14354 (pa_can_change_mode_class): ...this new function, inverting the
14355 sense of the return value.
14356 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14357 than CANNOT_CHANGE_MODE_CLASS in comments.
14358 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14359 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14360 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14361 (pdp11_cannot_change_mode_class): Replace with...
14362 (pdp11_can_change_mode_class): ...this new function, inverting the
14363 sense of the return value.
14364 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14365 * config/powerpcspe/powerpcspe-protos.h
14366 (rs6000_cannot_change_mode_class_ptr): Delete.
14367 * config/powerpcspe/powerpcspe.c
14368 (rs6000_cannot_change_mode_class_ptr): Delete.
14369 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14370 (rs6000_option_override_internal): Assign to
14371 targetm.can_change_mode_class instead of
14372 rs6000_cannot_change_mode_class_ptr.
14373 (rs6000_cannot_change_mode_class): Replace with...
14374 (rs6000_can_change_mode_class): ...this new function, inverting the
14375 sense of the return value.
14376 (rs6000_debug_cannot_change_mode_class): Replace with...
14377 (rs6000_debug_can_change_mode_class): ...this new function.
14378 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14379 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14380 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14381 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14382 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14384 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14385 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14386 (rs6000_option_override_internal): Assign to
14387 targetm.can_change_mode_class instead of
14388 rs6000_cannot_change_mode_class_ptr.
14389 (rs6000_cannot_change_mode_class): Replace with...
14390 (rs6000_can_change_mode_class): ...this new function, inverting the
14391 sense of the return value.
14392 (rs6000_debug_cannot_change_mode_class): Replace with...
14393 (rs6000_debug_can_change_mode_class): ...this new function.
14394 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14395 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14396 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14397 (s390_can_change_mode_class): ...this new function, inverting the
14398 sense of the return value.
14399 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14400 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14401 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14402 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14403 (sh_cannot_change_mode_class): Replace with...
14404 (sh_can_change_mode_class): ...this new function, inverting the
14405 sense of the return value.
14406 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14407 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14408 (sparc_can_change_mode_class): New function.
14409 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14410 * config/spu/spu.c (spu_can_change_mode_class): New function.
14411 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14412 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14413 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14414 (visium_can_change_mode_class): New function.
14415 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14417 2017-09-15 Richard Biener <rguenther@suse.de>
14419 PR tree-optimization/82217
14420 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14421 but not undefined case.
14423 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14426 * postreload.c (reload_cse_simplify_operands): Skip
14427 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14429 2017-09-15 Richard Biener <rguenther@suse.de>
14431 PR tree-optimization/68823
14432 * graphite-scop-detection.c (build_alias_set): If we have a
14433 possible dependence check whether we can handle them by just
14434 looking at the DRs DR_ACCESS_FNs.
14435 (build_scops): If build_alias_set fails, fail the SCOP.
14437 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14439 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14440 to support float128 built-in functions that require the ISA 3.0
14442 (BU_FLOAT128_3_HW): Likewise.
14443 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14444 built-in functions.
14445 (FMAF128): Likewise.
14447 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14448 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14449 floating point instructions.
14450 (rs6000_invalid_builtin): Likewise.
14451 (rs6000_builtin_mask_names): Likewise.
14452 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14453 (RS6000_BTM_FLOAT128_HW): Likewise.
14454 (RS6000_BTM_COMMON): Likewise.
14455 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14457 * doc/extend.texi (RS/6000 built-in functions): Document the
14458 IEEE 128-bit floating point square root and fused multiply-add
14459 built-in functions.
14461 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14463 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14464 reg (r2) isn't in the set of registers defined in the prologue.
14466 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14467 Alan Hayward <alan.hayward@arm.com>
14468 David Sherwood <david.sherwood@arm.com>
14470 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14471 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14472 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14473 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14474 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14476 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14477 max_vectorization_factor.
14478 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14485 (vect_worthwhile_without_simd_p): Declare.
14486 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14487 (vectorizable_reduction): Use it.
14488 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14489 (vectorizable_operation): Likewise.
14491 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14492 Alan Hayward <alan.hayward@arm.com>
14493 David Sherwood <david.sherwood@arm.com>
14495 * tree-vectorizer.h (vect_get_num_copies): New function.
14496 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14497 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14498 (vectorizable_induction): Likewise.
14499 (vectorizable_live_operation): Likewise.
14500 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14501 (vectorizable_bswap): Likewise.
14502 (vectorizable_call): Likewise.
14503 (vectorizable_conversion): Likewise.
14504 (vectorizable_assignment): Likewise.
14505 (vectorizable_shift): Likewise.
14506 (vectorizable_operation): Likewise.
14507 (vectorizable_store): Likewise.
14508 (vectorizable_load): Likewise.
14509 (vectorizable_condition): Likewise.
14510 (vectorizable_comparison): Likewise.
14511 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14513 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14514 Alan Hayward <alan.hayward@arm.com>
14515 David Sherwood <david.sherwood@arm.com>
14517 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14518 of vect_init_vector.
14520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14521 Alan Hayward <alan.hayward@arm.com>
14522 David Sherwood <david.sherwood@arm.com>
14524 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14525 an inline wrapper that provides a location.
14526 (gimple_build_vector): Likewise.
14527 * gimple-fold.c (gimple_build_vector_from_val): New function.
14528 (gimple_build_vector): Likewise.
14529 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14530 functions to build the initial value. Always return a gimple value.
14531 (get_initial_defs_for_reduction): Likewise. Only compute
14533 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14534 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14535 (vectorizable_induction): Use gimple_build_vector rather than
14538 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14539 Alan Hayward <alan.hayward@arm.com>
14540 David Sherwood <david.sherwood@arm.com>
14542 * target.h (vec_perm_indices): New typedef.
14543 (auto_vec_perm_indices): Likewise.
14544 * optabs-query.h: Include target.h
14545 (can_vec_perm_p): Take a vec_perm_indices *.
14546 * optabs-query.c (can_vec_perm_p): Likewise.
14547 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14548 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14549 * tree-vect-generic.c (lower_vec_perm): Likewise.
14550 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14551 (vect_grouped_load_supported): Likewise.
14552 (vect_shift_permute_load_chain): Likewise.
14553 (vect_permute_store_chain): Use auto_vec_perm_indices.
14554 (vect_permute_load_chain): Likewise.
14555 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14556 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14557 Update uses of can_vec_perm_p.
14558 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14559 mode with a number of elements. Take a vec_perm_indices *.
14560 (vect_create_epilog_for_reduction): Update accordingly.
14561 Use auto_vec_perm_indices.
14562 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14563 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14564 (vect_transform_slp_perm_load): Likewise.
14565 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14566 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14567 (vect_gen_perm_mask_checked): Likewise.
14568 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14569 (vect_gen_perm_mask_checked): Likewise.
14570 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14571 (vectorizable_store): Likewise.
14572 (vectorizable_load): Likewise.
14573 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14574 (vectorizable_bswap): Likewise.
14576 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14577 Alan Hayward <alan.hayward@arm.com>
14578 David Sherwood <david.sherwood@arm.com>
14580 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14581 * tree.c (build_vector): Likewise.
14582 (build_vector_from_ctor): Update accordingly.
14583 (build_vector_from_val): Likewise.
14584 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14585 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14586 * tree-vect-generic.c (add_rshift): Likewise.
14587 (expand_vector_divmod): Likewise.
14588 (optimize_vector_constructor): Likewise.
14589 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14590 (vect_transform_slp_perm_load): Likewise.
14591 (vect_schedule_slp_instance): Likewise.
14592 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14593 (vectorizable_call): Likewise.
14594 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14595 * expmed.c (make_tree): Likewise.
14596 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14597 a vector passed to build_vector.
14598 (fold_convert_const): Likewise.
14599 (exact_inverse): Likewise.
14600 (fold_ternary_loc): Likewise.
14601 (fold_relational_const): Likewise.
14602 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14603 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14604 (const_unop): Likewise. Store the reduction accumulator in a
14605 variable rather than an array.
14606 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14607 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14608 the new vector, rather than constructing it after the input arrays.
14609 (native_interpret_vector): Use auto_vec<tree> when building
14610 a vector passed to build_vector. Add elements in order.
14611 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14612 auto_vec<tree> when building a vector passed to build_vector.
14613 (vect_create_epilog_for_reduction): Likewise.
14614 (vectorizable_induction): Likewise.
14615 (get_initial_def_for_reduction): Likewise. Fix indentation of
14617 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14619 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14620 passed to build_vector.
14622 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14623 Alan Hayward <alan.hayward@arm.com>
14624 David Sherwood <david.sherwood@arm.com>
14626 * tree-core.h (tree_base::u): Add an "nelts" field.
14627 (tree_vector): Use VECTOR_CST_NELTS as the length.
14628 * tree.c (tree_size): Likewise.
14629 (make_vector): Initialize VECTOR_CST_NELTS.
14630 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14631 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14632 TYPE_VECTOR_SUBPARTS.
14633 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14634 as the number of units, setting it from VECTOR_CST_NELTS.
14635 (const_vector_from_tree): Likewise.
14636 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14637 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14638 (fold_negate_expr_1): Likewise.
14639 (fold_convert_const): Likewise.
14640 (const_binop): Likewise. Differentiate the number of output and
14642 (const_unop): Likewise.
14643 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14644 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14645 in cases that did the opposite.
14647 2017-09-14 Richard Biener <rguenther@suse.de>
14649 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14652 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14654 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14656 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14659 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14660 if and where to split a bb, except for splitting before debug insn
14661 sequences followed by non-label real insn. Delete debug insns
14662 in between basic blocks.
14664 * combine.c (make_compound_operation_int): Formatting fixes.
14666 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14667 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14668 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14669 * config/sol2.h (LINK_EH_SPEC): Likewise.
14670 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14671 * config/s390/linux.h (LINK_SPEC): Likewise.
14672 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14673 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14674 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14675 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14676 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14677 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14678 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14679 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14681 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14683 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14685 (ENDFILE_LINUX_SPEC): Likewise.
14686 (LINK_EH_SPEC): Likewise.
14687 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14688 (LINK_OS_LINUX_SPEC32): Likewise.
14689 (LINK_OS_LINUX_SPEC64): Likewise.
14690 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14691 (LINK_OS_LINUX_SPEC): Likewise.
14693 2017-09-13 Martin Liska <mliska@suse.cz>
14695 PR middle-end/82154
14696 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14697 CASE_HIGH is NULL_TREE.
14699 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14700 Alan Hayward <alan.hayward@arm.com>
14701 David Sherwood <david.sherwood@arm.com>
14703 * target.def (secondary_memory_needed): New hook.
14704 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14705 instead of SECONDARY_MEMORY_NEEDED.
14706 (secondary_memory_needed_mode): Likewise.
14707 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14708 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14709 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14710 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14711 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14712 * doc/tm.texi: Regenerate.
14713 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14714 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14715 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14716 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14717 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14718 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14719 mode argument first and change the reg_class arguments to reg_class_t.
14720 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14721 Make static. Update the call to inline_secondary_memory_needed.
14722 (ix86_register_move_cost): Update the call to
14723 inline_secondary_memory_needed.
14724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14725 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14727 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14728 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14730 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14731 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14732 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14733 and match hook interface. Add comment from mips.h.
14734 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14735 * config/mmix/mmix.md (truncdfsf2): Refer to
14736 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14738 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14739 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14740 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14741 (pa_secondary_memory_needed): New function.
14742 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14743 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14744 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14745 (pdp11_secondary_memory_needed): Make static and match hook interface.
14746 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14747 * config/powerpcspe/powerpcspe-protos.h
14748 (rs6000_secondary_memory_needed_ptr): Delete.
14749 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14751 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14752 (rs6000_option_override_internal): Assign to
14753 targetm.secondary_memory_needed rather than
14754 rs6000_secondary_memory_needed_ptr.
14755 (rs6000_secondary_memory_needed): Match hook interface.
14756 (rs6000_debug_secondary_memory_needed): Likewise.
14757 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14758 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14759 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14760 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14761 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14762 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14764 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14765 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14766 (rs6000_option_override_internal): Assign to
14767 targetm.secondary_memory_needed rather than
14768 rs6000_secondary_memory_needed_ptr.
14769 (rs6000_secondary_memory_needed): Match hook interface.
14770 (rs6000_debug_secondary_memory_needed): Likewise.
14771 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14772 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14773 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14774 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14775 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14776 (sparc_secondary_memory_needed): New function.
14777 * lra-constraints.c (check_and_process_move): Refer to
14778 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14780 (curr_insn_transform): Likewise.
14781 (process_alt_operands): Use targetm.secondary_memory_needed
14782 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14783 (check_secondary_memory_needed_p): Likewise.
14784 (choose_split_class): Likewise.
14785 * reload.c: Unconditionally include code that was previously
14786 conditional on SECONDARY_MEMORY_NEEDED.
14787 (push_secondary_reload): Use targetm.secondary_memory_needed
14788 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14789 (push_reload): Likewise.
14790 * reload1.c: Unconditionally include code that was previously
14791 conditional on SECONDARY_MEMORY_NEEDED.
14792 (choose_reload_regs): Use targetm.secondary_memory_needed
14793 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14794 (gen_reload): Likewise.
14795 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14797 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14798 Alan Hayward <alan.hayward@arm.com>
14799 David Sherwood <david.sherwood@arm.com>
14801 * target.def (secondary_memory_needed_mode): New hook:
14802 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14803 * targhooks.h (default_secondary_memory_needed_mode): New function.
14804 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14805 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14806 * doc/tm.texi: Regenerate.
14807 * lra-constraints.c (check_and_process_move): Use
14808 targetm.secondary_memory_needed_mode instead of
14809 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14810 (curr_insn_transform): Likewise.
14811 * reload.c (get_secondary_mem): Likewise.
14812 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14813 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14815 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14816 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14817 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14818 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14819 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14821 * config/powerpcspe/powerpcspe-protos.h
14822 (rs6000_secondary_memory_needed_mode): Delete.
14823 * config/powerpcspe/powerpcspe.c
14824 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14825 (rs6000_secondary_memory_needed_mode): Make static.
14826 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14827 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14829 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14831 (rs6000_secondary_memory_needed_mode): Make static.
14832 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14833 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14834 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14835 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14836 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14838 (sparc_secondary_memory_needed_mode): New function.
14839 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14841 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14843 * config/aarch64/constraints.md (Umq): New constraint.
14844 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14846 (mov<mode>): Update condition.
14848 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14850 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14851 when bitposition is the same.
14853 2017-09-13 Richard Biener <rguenther@suse.de>
14855 * dwarf2out.c (output_die_symbol): Remove.
14856 (output_die): Do not output a DIEs symbol.
14858 2017-09-13 Richard Biener <rguenther@suse.de>
14860 PR middle-end/82128
14861 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14862 default-def to avoid breaking iterator update with the weird
14863 interaction with cgraph_update_edges_for_call_stmt_node.
14865 2017-09-13 Richard Biener <rguenther@suse.de>
14867 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14868 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14869 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14870 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14872 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14874 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14875 Disable pc relative literal load irrespective of
14876 TARGET_FIX_ERR_A53_84341 for default.
14878 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14880 * config/sparc/sparc.c (output_return): Output the source location of
14881 the insn in the delay slot, if any.
14882 (output_sibcall): Likewise.
14884 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14887 * common.opt (-static-pie): New alias.
14888 (shared): Negate static-pie.
14889 (-no-pie): Update help text.
14891 (static-pie): New option.
14892 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14893 -static-pie support.
14894 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14895 (LINK_EH_SPEC): Likewise.
14896 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14897 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14898 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14899 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14900 (init_gcc_specs): Likewise.
14901 (init_spec): Likewise.
14902 (display_help): Update help message for -pie.
14903 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14906 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14908 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14909 (movdi_aarch64): Likewise.
14910 (movti_aarch64): Likewise.
14912 2017-09-12 Simon Wright <simon@pushface.org>
14915 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14916 calculation of the minor version, always output as 0.
14918 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14922 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14923 on it early, rather than manual conversion late. For
14924 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14925 instead of performing manual conversion.
14927 2017-09-12 Carl Love <cel@us.ibm.com>
14929 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14930 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14932 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14933 VMULOSW): Add definitions.
14934 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14935 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14936 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14938 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14940 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14942 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14944 (movti_aarch64): Likewise.
14945 (movdf_aarch64): Likewise.
14946 (movtf_aarch64): Likewise.
14947 (load_pairdi): Likewise.
14948 (store_pairdi): Likewise.
14949 (load_pairdf): Likewise.
14950 (store_pairdf): Likewise.
14951 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14952 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14953 (ldr_got_small_<mode>): Likewise.
14954 (ldr_got_small_28k_<mode>): Likewise.
14955 (ldr_got_tiny): Likewise.
14956 * config/aarch64/iterators.md (ldst_sz): New.
14957 (ldpstp_sz): Likewise.
14958 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14960 (thunderx_load): Split load_8 to load_16.
14961 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14963 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14964 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14965 (xgene1_store_pair): Split store_8 to store_16.
14966 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14967 (falkor_st_0_st_sd): Split store_8 to store_16.
14969 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14971 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14972 and store1/2/3/4 to store_4/8/12/16.
14973 * config/aarch64/aarch64.md: Update for rename.
14974 * config/arm/arm.md: Likewise.: Likewise.
14975 * config/arm/arm.c: Likewise.
14976 * config/arm/thumb1.md: Likewise.
14977 * config/arm/thumb2.md: Likewise.
14978 * config/arm/vfp.md: Likewise.
14979 * config/arm/arm-generic.md: Likewise.
14980 * config/arm/arm1020e.md: Likewise.
14981 * config/arm/arm1026ejs.md: Likewise.
14982 * config/arm/arm1136jfs.md: Likewise.
14983 * config/arm/arm926ejs.md: Likewise.
14984 * config/arm/cortex-a15.md: Likewise.
14985 * config/arm/cortex-a17.md: Likewise.
14986 * config/arm/cortex-a5.md: Likewise.
14987 * config/arm/cortex-a53.md: Likewise.
14988 * config/arm/cortex-a57.md: Likewise.
14989 * config/arm/cortex-a7.md: Likewise.
14990 * config/arm/cortex-a8.md: Likewise.
14991 * config/arm/cortex-a9.md: Likewise.
14992 * config/arm/cortex-m4.md: Likewise.
14993 * config/arm/cortex-m7.md: Likewise.
14994 * config/arm/cortex-r4.md: Likewise.
14995 * config/arm/exynos-m1.md: Likewise.
14996 * config/arm/fa526.md: Likewise.
14997 * config/arm/fa606te.md: Likewise.
14998 * config/arm/fa626te.md: Likewise.
14999 * config/arm/fa726te.md: Likewise.
15000 * config/arm/fmp626.md: Likewise.
15001 * config/arm/iwmmxt.md: Likewise.
15002 * config/arm/ldmstm.md: Likewise.
15003 * config/arm/marvell-pj4.md: Likewise.
15004 * config/arm/xgene1.md: Likewise.
15005 * config/aarch64/thunderx.md: Likewise.
15006 * config/aarch64/thunderx2t99.md: Likewise.
15007 * config/aarch64/falkor.md: Likewise.
15009 2017-09-12 Martin Liska <mliska@suse.cz>
15011 * attribs.c (private_lookup_attribute): New function.
15012 * attribs.h (private_lookup_attribute): Declared here.
15013 (lookup_attribute): Called from this place.
15015 2017-09-12 Richard Biener <rguenther@suse.de>
15017 PR tree-optimization/82157
15018 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15019 stmts with side-effects.
15021 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15022 Alan Hayward <alan.hayward@arm.com>
15023 David Sherwood <david.sherwood@arm.com>
15025 * target.def (hard_regno_nregs): New hook.
15026 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15027 * targhooks.h (default_hard_regno_nregs): Declare.
15028 * targhooks.c (default_hard_regno_nregs): New function.
15029 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15030 (TARGET_HARD_REGNO_NREGS): ...this hook.
15031 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15032 (CLASS_MAX_NREGS): Likewise.
15033 * doc/tm.texi: Regenerate.
15034 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15035 instead of HARD_REGNO_NREGS.
15036 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15037 HARD_REGNO_NREGS in the comment.
15038 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15039 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15040 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15041 Return an unsigned int.
15042 (TARGET_HARD_REGNO_NREGS): Redefine.
15043 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15044 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15045 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15046 (arc_hard_regno_nregs): New function.
15047 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15048 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15049 (arm_hard_regno_nregs): New function.
15050 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15051 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15052 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15053 (TARGET_HARD_REGNO_NREGS): Redefine.
15054 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15055 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15056 (HARD_REGNO_NREGS): Delete.
15057 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15058 (cr16_hard_regno_nregs): New function.
15059 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15060 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15061 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15062 (cris_hard_regno_nregs): New function.
15063 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15064 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15065 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15066 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15067 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15068 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15069 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15070 (frv_hard_regno_nregs): Make static. Take and return an
15072 (frv_class_max_nregs): Remove outdated copy of documentation.
15073 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15074 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15075 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15076 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15077 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15078 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15079 (TARGET_HARD_REGNO_NREGS): Redefine.
15080 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15081 (CLASS_MAX_NREGS): Update comment.
15082 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15083 (ia64_hard_regno_nregs): New function.
15084 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15085 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15086 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15087 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15088 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15090 (m32c_hard_regno_nregs): Likewise. Make static.
15091 (TARGET_HARD_REGNO_NREGS): Redefine.
15092 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15093 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15094 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15095 (m68k_hard_regno_nregs): New function.
15096 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15097 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15098 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15099 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15100 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15101 Take and return an unsigned int.
15102 (TARGET_HARD_REGNO_NREGS): Redefine.
15103 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15104 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15105 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15106 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15107 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15108 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15109 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15110 (msp430_hard_regno_nregs): Make static. Take and return an
15112 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15113 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15114 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15115 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15116 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15117 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15118 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15119 (TARGET_HARD_REGNO_NREGS): Redefine.
15120 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15121 (PA_HARD_REGNO_NREGS): ...this.
15122 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15123 (PA_HARD_REGNO_NREGS): ...this.
15124 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15125 (pa_hard_regno_nregs): New function.
15126 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15127 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15128 (pdp11_hard_regno_nregs): New function.
15129 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15130 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15131 (rs6000_hard_regno_nregs_hook): New function.
15132 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15133 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15134 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15135 Take and return an unsigned int. Move earlier in file.
15136 (TARGET_HARD_REGNO_NREGS): Redefine.
15137 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15138 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15139 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15140 (rl78_hard_regno_nregs): Make static. Take and return an
15142 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15143 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15144 (rs6000_hard_regno_nregs_hook): New function.
15145 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15146 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15147 (TARGET_HARD_REGNO_NREGS): Redefine.
15148 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15149 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15150 instead of HARD_REGNO_NREGS.
15151 (s390_hard_regno_nregs): New function.
15152 (s390_hard_regno_mode_ok): Add comment from s390.h.
15153 (TARGET_HARD_REGNO_NREGS): Redefine.
15154 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15155 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15156 (sh_hard_regno_nregs): New function.
15157 (sh_pass_in_reg_p): Use it.
15158 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15159 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15160 (sparc_hard_regno_nregs): New function.
15161 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15162 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15163 (spu_function_arg_advance): Use it, supplying a valid register number.
15164 (TARGET_HARD_REGNO_NREGS): Redefine.
15165 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15166 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15167 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15168 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15169 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15170 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15171 (CLASS_MAX_NREGS): Remove copy of old documentation.
15172 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15173 (visium_hard_regno_nregs): New function.
15174 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15175 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15176 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15177 xtensa_hard_regno_nregs): New function.
15178 * system.h (HARD_REGNO_NREGS): Poison.
15180 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15182 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15183 hard_regno_nregs instead of HARD_REGNO_NREGS.
15184 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15185 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15186 (c6x_expand_epilogue): Likewise.
15187 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15188 (frv_read_iacc_argument): Likewise.
15189 * config/sh/sh.c: Include regs.h.
15190 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15191 (regs_used): Likewise.
15192 (output_stack_adjust): Likewise.
15193 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15194 * expmed.c: Include regs.h.
15195 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15196 * ree.c: Include regs.h.
15197 (combine_reaching_defs): Use hard_regno_nregs instead of
15199 (add_removable_extension): Likewise.
15201 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15203 * regs.h (hard_regno_nregs): Turn into a function.
15204 (end_hard_regno): Update accordingly.
15205 * caller-save.c (setup_save_areas): Likewise.
15206 (save_call_clobbered_regs): Likewise.
15207 (replace_reg_with_saved_mem): Likewise.
15208 (insert_restore): Likewise.
15209 (insert_save): Likewise.
15210 * combine.c (can_change_dest_mode): Likewise.
15211 (move_deaths): Likewise.
15212 (distribute_notes): Likewise.
15213 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15214 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15215 (rs6000_split_multireg_move): Likewise.
15216 (rs6000_register_move_cost): Likewise.
15217 (rs6000_memory_move_cost): Likewise.
15218 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15219 (rs6000_split_multireg_move): Likewise.
15220 (rs6000_register_move_cost): Likewise.
15221 (rs6000_memory_move_cost): Likewise.
15222 * cselib.c (cselib_reset_table): Likewise.
15223 (cselib_lookup_1): Likewise.
15224 * emit-rtl.c (set_mode_and_regno): Likewise.
15225 * function.c (aggregate_value_p): Likewise.
15226 * ira-color.c (setup_profitable_hard_regs): Likewise.
15227 (check_hard_reg_p): Likewise.
15228 (calculate_saved_nregs): Likewise.
15229 (assign_hard_reg): Likewise.
15230 (improve_allocation): Likewise.
15231 (calculate_spill_cost): Likewise.
15232 * ira-emit.c (modify_move_list): Likewise.
15233 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15234 (ira_hard_reg_in_set_p): Likewise.
15235 * ira.c (setup_reg_mode_hard_regset): Likewise.
15236 (clarify_prohibited_class_mode_regs): Likewise.
15237 (check_allocation): Likewise.
15238 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15239 (lra_setup_reg_renumber): Likewise.
15240 (setup_try_hard_regno_pseudos): Likewise.
15241 (spill_for): Likewise.
15242 (assign_hard_regno): Likewise.
15243 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15244 * lra-constraints.c (in_class_p): Likewise.
15245 (lra_constraint_offset): Likewise.
15246 (simplify_operand_subreg): Likewise.
15247 (lra_constraints): Likewise.
15248 (split_reg): Likewise.
15249 (split_if_necessary): Likewise.
15250 (invariant_p): Likewise.
15251 (inherit_in_ebb): Likewise.
15252 * lra-lives.c (process_bb_lives): Likewise.
15253 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15254 (get_hard_regs): Likewise.
15255 (do_remat): Likewise.
15256 * lra-spills.c (assign_spill_hard_regs): Likewise.
15257 * mode-switching.c (create_pre_exit): Likewise.
15258 * postreload.c (reload_combine_recognize_pattern): Likewise.
15259 * recog.c (peep2_find_free_register): Likewise.
15260 * regcprop.c (kill_value_regno): Likewise.
15261 (set_value_regno): Likewise.
15262 (copy_value): Likewise.
15263 (maybe_mode_change): Likewise.
15264 (find_oldest_value_reg): Likewise.
15265 (copyprop_hardreg_forward_1): Likewise.
15266 * regrename.c (check_new_reg_p): Likewise.
15267 (regrename_do_replace): Likewise.
15268 * reload.c (push_reload): Likewise.
15269 (combine_reloads): Likewise.
15270 (find_dummy_reload): Likewise.
15271 (operands_match_p): Likewise.
15272 (find_reloads): Likewise.
15273 (find_equiv_reg): Likewise.
15274 (reload_adjust_reg_for_mode): Likewise.
15275 * reload1.c (count_pseudo): Likewise.
15276 (count_spilled_pseudo): Likewise.
15277 (find_reg): Likewise.
15278 (clear_reload_reg_in_use): Likewise.
15279 (free_for_value_p): Likewise.
15280 (allocate_reload_reg): Likewise.
15281 (choose_reload_regs): Likewise.
15282 (reload_adjust_reg_for_temp): Likewise.
15283 (emit_reload_insns): Likewise.
15284 (delete_output_reload): Likewise.
15285 * rtlanal.c (subreg_get_info): Likewise.
15286 * sched-deps.c (sched_analyze_reg): Likewise.
15287 * sel-sched.c (init_regs_for_mode): Likewise.
15288 (mark_unavailable_hard_regs): Likewise.
15289 (choose_best_reg_1): Likewise.
15290 (verify_target_availability): Likewise.
15291 * valtrack.c (dead_debug_insert_temp): Likewise.
15292 * var-tracking.c (track_loc_p): Likewise.
15293 (emit_note_insn_var_location): Likewise.
15294 * varasm.c (make_decl_rtl): Likewise.
15295 * reginfo.c (choose_hard_reg_mode): Likewise.
15296 (init_reg_modes_target): Refer directly to
15297 this_target_regs->x_hard_regno_nregs.
15299 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15301 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15302 instead of hard_regno_nregs.
15304 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15306 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15307 end_hard_regno instead of hard_regno_nregs.
15308 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15309 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15310 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15311 * ira-color.c (improve_allocation): Likewise.
15312 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15313 * lra-lives.c (mark_regno_live): Likewise.
15314 (mark_regno_dead): Likewise.
15315 * lra-remat.c (operand_to_remat): Likewise.
15316 * lra.c (collect_non_operand_hard_regs): Likewise.
15317 * postreload.c (reload_combine_note_store): Likewise.
15318 (move2add_valid_value_p): Likewise.
15319 * reload.c (regno_clobbered_p): Likewise.
15321 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15323 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15325 * config/v850/v850.c (v850_reorg): Likewise.
15326 * reload.c (refers_to_regno_for_reload_p): Likewise.
15327 (find_equiv_reg): Likewise.
15328 * reload1.c (reload_reg_reaches_end_p): Likewise.
15330 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15332 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15334 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15335 * config/arm/arm.c (output_move_neon): Likewise.
15336 (arm_attr_length_move_neon): Likewise.
15337 (neon_split_vcombine): Likewise.
15338 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15339 (c6x_mark_reg_written): Likewise.
15340 (c6x_dwarf_register_span): Likewise.
15341 * config/i386/i386.c (ix86_save_reg): Likewise.
15342 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15343 (rws_access_reg): Likewise.
15344 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15345 * mode-switching.c (create_pre_exit): Likewise.
15346 * ree.c (combine_reaching_defs): Likewise.
15347 (add_removable_extension): Likewise.
15348 * regcprop.c (find_oldest_value_reg): Likewise.
15349 (copyprop_hardreg_forward_1): Likewise.
15350 * reload.c (reload_inner_reg_of_subreg): Likewise.
15351 (push_reload): Likewise.
15352 (combine_reloads): Likewise.
15353 (find_dummy_reload): Likewise.
15354 (reload_adjust_reg_for_mode): Likewise.
15355 * reload1.c (find_reload_regs): Likewise.
15356 (forget_old_reloads_1): Likewise.
15357 (reload_reg_free_for_value_p): Likewise.
15358 (reload_adjust_reg_for_temp): Likewise.
15359 (emit_reload_insns): Likewise.
15360 (delete_output_reload): Likewise.
15361 * sel-sched.c (choose_best_reg_1): Likewise.
15362 (choose_best_pseudo_reg): Likewise.
15364 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15365 Alan Hayward <alan.hayward@arm.com>
15366 David Sherwood <david.sherwood@arm.com>
15368 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15369 * target.def (slow_unaligned_access): New hook.
15370 * targhooks.h (default_slow_unaligned_access): Declare.
15371 * targhooks.c (default_slow_unaligned_access): New function.
15372 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15373 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15374 * doc/tm.texi: Regenerate.
15375 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15376 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15377 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15379 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15380 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15382 (rs6000_slow_unaligned_access): New function.
15383 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15384 (expand_block_compare): Likewise.
15385 (expand_strn_compare): Likewise.
15386 (rs6000_rtx_costs): Likewise.
15387 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15388 (riscv_slow_unaligned_access): Likewise.
15389 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15390 (riscv_slow_unaligned_access_p): ...this and make static.
15391 (riscv_option_override): Update accordingly.
15392 (riscv_slow_unaligned_access): New function.
15393 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15394 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15395 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15396 (rs6000_slow_unaligned_access): New function.
15397 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15398 (rs6000_rtx_costs): Likewise.
15399 * config/rs6000/rs6000-string.c (expand_block_compare)
15400 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15401 of SLOW_UNALIGNED_ACCESS.
15402 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15403 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15404 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15405 of SLOW_UNALIGNED_ACCESS.
15406 * expmed.c (simple_mem_bitfield_p): Likewise.
15407 * expr.c (alignment_for_piecewise_move): Likewise.
15408 (emit_group_load_1): Likewise.
15409 (emit_group_store): Likewise.
15410 (copy_blkmode_from_reg): Likewise.
15411 (emit_push_insn): Likewise.
15412 (expand_assignment): Likewise.
15413 (store_field): Likewise.
15414 (expand_expr_real_1): Likewise.
15415 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15416 * lra-constraints.c (simplify_operand_subreg): Likewise.
15417 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15418 * gimple-ssa-store-merging.c: Likewise in block comment at start
15420 * tree-ssa-strlen.c: Include target.h.
15421 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15422 of SLOW_UNALIGNED_ACCESS.
15423 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15425 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15427 PR rtl-optimization/82185
15428 * expmed.c (emit_store_flag_int): Only test tem if it has been
15431 2017-09-12 Richard Biener <rguenther@suse.de>
15433 PR middle-end/82149
15434 * match.pd ((FTYPE) N CMP CST): Fix typo.
15436 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15438 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15440 (mips_near_type_p): Add 'short_call' attribute as a synonym
15442 * doc/extend.texi (short_call): Document new function attribute.
15444 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15447 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15448 assertion check that in the condition.
15449 (get_atomic_generic_size): Likewise. Before testing if parameter
15450 has pointer type, if it has array type, call for C++
15451 default_conversion to perform array-to-pointer conversion.
15453 2017-09-12 Richard Biener <rguenther@suse.de>
15455 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15456 for operations we cannot scalarize.
15458 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15460 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15461 vectors heap vectors. Clean up comments.
15462 Make visited_bbs a reference.
15463 (profitable_jump_thread_path): Make GC
15464 vectors heap vectors. Clean up comments.
15466 (convert_and_register_jump_thread_path): Make GC vectors heap
15468 (check_subpath_and_update_thread_path): Same. Clean up comments.
15469 Make visited_bbs a reference.
15470 (handle_phi): Abstract common code to to
15471 register_jump_thread_path_if_profitable.
15472 Rename VAR_BB to DEF_BB.
15474 Make GC vectors heap vectors.
15475 Make visited_bbs a reference.
15476 (handle_assignment): Same.
15477 (register_jump_thread_path_if_profitable): New.
15478 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15480 Make GC vectors heap vectors. Clean up comments.
15481 Make visited_bbs a reference.
15482 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15483 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15486 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15489 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15490 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15492 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15494 Revert r251800 and r251799.
15496 2017-09-11 Martin Jambor <mjambor@suse.cz>
15499 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15500 arguments in advance.
15501 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15502 use it to find predecessor edges.
15503 (naive_outof_ssa): Collect vector of predecessors.
15505 2017-09-08 Jason Merrill <jason@redhat.com>
15507 PR c++/70029 - ICE with ref-qualifier and -flto
15508 * langhooks.h (struct lang_hooks_for_types): Add
15509 copy_lang_qualifiers.
15510 * attribs.c (build_type_attribute_qual_variant): Use it.
15511 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15513 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15514 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15516 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15519 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15520 (*mulsi3_sp64): New instruction.
15521 (mulsi3): New expander.
15523 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15525 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15527 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15529 * sancov.c: Include memmodel.h.
15531 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15534 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15537 2017-09-07 Carl Love <cel@us.ibm.com>
15539 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15540 the sldi instruction.
15542 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15544 * sancov.c: Include tm_p.h.
15546 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15549 * output.h (switch_to_other_text_partition): New declaration.
15550 * varasm.c (switch_to_other_text_partition): New function.
15551 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15552 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15553 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15554 to the other text partition before emitting LCL label and switch back
15555 after emitting the word after it.
15557 2017-09-07 Richard Biener <rguenther@suse.de>
15559 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15560 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15561 critical edges here, after loop init.
15562 (pass_data_pre): Remove PROP_no_crit_edges flags.
15563 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15564 for valueization of call args to avoid leaking VN_TOP.
15565 (visit_use): Assert we do not visit default defs.
15566 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15567 Use error_mark_node to more easily detect leaking VN_TOP.
15568 All default-defs are varying, not VN_TOP. Mark them visited.
15569 (run_scc_vn): Make code match comment.
15571 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15573 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15574 OPTION_MASK_FLOAT128_KEYWORD.
15575 (POWERPC_MASKS): Likewise.
15576 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15577 support for the -mfloat128-type option, and make -mfloat128
15578 default on PowerPC Linux systems. Define or undefine
15579 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15580 Define __float128 to be __ieee128 if IEEE 128-bit support is
15581 enabled, or undefine it.
15582 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15583 Delete defining __FLOAT128_TYPE__.
15584 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15585 -mfloat128-type option and make -mfloat128 default on PowerPC
15587 (TARGET_FLOAT128_TYPE): Likewise.
15588 (-mfloat128-type): Likewise.
15589 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15590 Delete the -mfloat128-type option and make -mfloat128 default on
15591 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15592 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15593 128-bit floating point is enabled. Change tests from using
15594 -mfloat128-type to -mfloat128.
15595 (rs6000_mangle_type): Use the correct mangling for the __float128
15596 type even if normal long double is restricted to 64-bits.
15597 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15599 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15600 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15601 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15603 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15604 documentation for -mfloat128.
15606 2017-09-06 Olivier Hainque <hainque@adacore.com>
15608 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15610 2017-09-06 Wish Wu <wishwu007@gmail.com>
15611 Jakub Jelinek <jakub@redhat.com>
15613 * asan.c (initialize_sanitizer_builtins): Add
15614 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15615 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15616 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15617 BT_FN_VOID_UINT64_PTR variables.
15618 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15619 (BT_FN_VOID_UINT16_UINT16): Likewise.
15620 (BT_FN_VOID_UINT32_UINT32): Likewise.
15621 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15622 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15623 (BT_FN_VOID_UINT64_PTR): Likewise.
15624 * common.opt (flag_sanitize_coverage): New variable.
15625 (fsanitize-coverage=trace-pc): Remove.
15626 (fsanitize-coverage=): Add.
15627 * flag-types.h (enum sanitize_coverage_code): New enum.
15628 * fold-const.c (fold_range_test): Disable non-short-circuit
15629 optimization if flag_sanitize_coverage.
15630 (fold_truth_andor): Likewise.
15631 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15632 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15633 (coverage_sanitizer_opts): New array.
15634 (get_closest_sanitizer_option): Add OPTS argument, handle also
15635 OPT_fsanitize_coverage_.
15636 (parse_sanitizer_options): Adjusted to also handle
15637 OPT_fsanitize_coverage_.
15638 (common_handle_option): Add OPT_fsanitize_coverage_.
15639 * sancov.c (instrument_comparison, instrument_switch): New function.
15640 (sancov_pass): Add trace-cmp support.
15641 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15642 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15643 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15644 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15645 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15646 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15647 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15648 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15649 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15650 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15652 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15654 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15655 error. Only quit immediately if parsing is complete.
15656 (BEGIN): Initialize fatal_err and parse_done.
15657 (begin fpu, end fpu): Check number of arguments.
15658 (begin arch, end arch): Likewise.
15659 (begin cpu, end cpu): Likewise.
15660 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15661 (optalias): Likewise.
15663 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15665 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15666 * config/arm/arm-isa.h: Delete. Move definitions to ...
15667 * arm-cpus.in: ... here. Use new feature and fgroup values.
15668 * config/arm/arm.c (arm_option_override): Use lower case for feature
15670 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15671 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15672 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15674 (print_isa_bits_for): New function.
15675 (gen_isa): New function.
15676 (gen_comm_data): Use print_isa_bits_for.
15677 (define feature): New keyword.
15678 (define fgroup): New keyword.
15679 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15680 (arm-isa.h): Add rule to generate file.
15681 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15682 case for feature bit names.
15684 2017-09-06 Richard Biener <rguenther@suse.de>
15686 * tree-ssa-pre.c (NECESSARY): Remove.
15687 (create_expression_by_pieces): Do not touch pass-local flags.
15688 (insert_into_preds_of_block): Likewise.
15689 (do_pre_regular_insertion): Likewise.
15690 (eliminate_insert): Likewise.
15691 (eliminate_dom_walker::before_dom_children): Likewise.
15692 (fini_eliminate): Do not look at inserted_exprs.
15693 (mark_operand_necessary): Remove.
15694 (remove_dead_inserted_code): Replace with simple work-list
15695 algorithm based on inserted_exprs and SSA uses.
15696 (pass_pre::execute): Re-order fini_eliminate and
15697 remove_dead_inserted_code.
15699 2017-09-06 Olivier Hainque <hainque@adacore.com>
15701 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15702 for VxWorks 7. Adjust surrounding comments.
15704 2017-09-06 Richard Biener <rguenther@suse.de>
15706 * gimple-ssa-strength-reduction.c
15707 (find_candidates_dom_walker::before_dom_children): Also allow
15710 2017-09-06 Richard Biener <rguenther@suse.de>
15712 PR tree-optimization/82108
15713 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15714 for gap in the non-permutation SLP case.
15716 2017-09-06 Martin Jambor <mjambor@suse.cz>
15718 PR tree-optimization/82078
15719 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15720 add_access_to_work_queue...
15721 (build_accesses_from_assign): ...here.
15722 (propagate_all_subaccesses): Make sure racc is the group
15723 representative, if there is one.
15725 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15727 PR middle-end/82095
15728 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15731 2017-09-06 Richard Biener <rguenther@suse.de>
15733 * gimple-ssa-strength-reduction.c
15734 (find_candidates_dom_walker::before_doom_children): Use a
15735 type and not a mode check.
15737 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15740 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15742 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15744 2017-09-05 Jeff Law <law@redhat.com>
15746 PR tree-optimization/64910
15747 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15748 cases where we have 3 or more operands.
15750 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15752 PR middle-end/81768
15753 * omp-low.c (lower_omp_for): Recompute tree invariant if
15754 gimple_omp_for_initial/final is ADDR_EXPR.
15756 PR middle-end/81768
15757 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15758 into gimple val before gimplification fo the COND_EXPR.
15760 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15762 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15763 REGION_COPY argument.
15764 (thread_through_all_blocks): Remove unused argument to
15765 duplicate_thread_path.
15767 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15768 Alan Hayward <alan.hayward@arm.com>
15769 David Sherwood <david.sherwood@arm.com>
15771 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15772 Take a scalar_mode rather than a machine_mode.
15773 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15774 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15775 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15776 (aarch64_gen_adjusted_ldpstp): Likewise.
15777 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15779 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15780 Alan Hayward <alan.hayward@arm.com>
15781 David Sherwood <david.sherwood@arm.com>
15783 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15784 Take a scalar_int_mode instead of a machine_mode.
15785 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15786 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15787 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15788 (aarch64_simd_attr_length_rglist): Delete.
15789 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15790 a scalar_int_mode instead of a machine_mode.
15791 (aarch64_add_offset): Likewise.
15792 (aarch64_internal_mov_immediate): Likewise
15793 (aarch64_add_constant_internal): Likewise.
15794 (aarch64_add_constant): Likewise.
15795 (aarch64_movw_imm): Likewise.
15796 (aarch64_rtx_arith_op_extract_p): Likewise.
15797 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15798 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15799 Remove assert that the mode isn't a vector.
15800 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15801 (aarch64_expand_mov_immediate): Update calls after above changes.
15802 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15803 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15804 (aarch64_move_imm): Likewise.
15805 (aarch64_can_const_movi_rtx_p): Likewise.
15806 (aarch64_strip_extend): Likewise.
15807 (aarch64_extr_rtx_p): Likewise.
15808 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15809 a CONST_INT when the mode parameter is VOIDmode.
15810 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15814 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15815 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15816 (bitwise_type_for_mode): Update accordingly.
15818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15820 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15821 * stor-layout.c (mode_for_size_tree): Likewise.
15822 (mode_for_array): Update accordingly.
15823 (layout_decl): Likewise.
15824 (compute_record_mode): Likewise. Only set the mode once.
15826 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15828 * target.def (get_mask_mode): Change return type to opt_mode.
15830 * doc/tm.texi: Regenerate.
15831 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15832 * targhooks.c (default_get_mask_mode): Likewise.
15833 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15834 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15835 targetm.get_mask_mode.
15836 * tree.c (build_truth_vector_type): Likewise.
15838 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15840 * machmode.h (mode_for_vector): Return an opt_mode.
15841 * stor-layout.c (mode_for_vector): Likewise.
15842 (mode_for_int_vector): Update accordingly.
15843 (layout_type): Likewise.
15844 * config/i386/i386.c (emit_memmov): Likewise.
15845 (ix86_expand_set_or_movmem): Likewise.
15846 (ix86_expand_vector_init): Likewise.
15847 (ix86_get_mask_mode): Likewise.
15848 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15850 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15851 * expmed.c (extract_bit_field_1): Likewise.
15852 * expr.c (expand_expr_real_2): Likewise.
15853 * optabs-query.c (can_vec_perm_p): Likewise.
15854 (can_vec_mask_load_store_p): Likewise.
15855 * optabs.c (expand_vec_perm): Likewise.
15856 * targhooks.c (default_get_mask_mode): Likewise.
15857 * tree-vect-stmts.c (vectorizable_store): Likewise.
15858 (vectorizable_load): Likewise.
15859 (get_vectype_for_scalar_type_and_size): Likewise.
15861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15863 * machmode.h (mode_for_int_vector): New function.
15864 * stor-layout.c (mode_for_int_vector): Likewise.
15865 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15866 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15867 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15868 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15869 (s390_expand_vcond): Likewise.
15871 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15873 * machmode.h (opt_machine_mode): New type.
15874 (opt_mode<T>): Allow construction from anything that can be
15876 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15877 (mode_for_size): Return an opt_machine_mode.
15878 * stor-layout.c (mode_for_size): Likewise.
15879 (mode_for_size_tree): Update call accordingly.
15880 (bitwise_mode_for_mode): Likewise.
15881 (make_fract_type): Likewise.
15882 (make_accum_type): Likewise.
15883 * caller-save.c (replace_reg_with_saved_mem): Update call
15885 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15886 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15887 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15888 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15889 * expmed.c (extract_bit_field_1): Likewise.
15890 * reload.c (get_secondary_mem): Likewise.
15891 * varasm.c (assemble_integer): Likewise.
15892 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15897 * machmode.h (decimal_float_mode_for_size): New function.
15898 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15899 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15900 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15901 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15902 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15903 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15905 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15907 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15908 (get_builtin_sync_mode): Likewise.
15909 (expand_ifn_atomic_compare_exchange): Likewise.
15910 (expand_builtin_atomic_clear): Likewise.
15911 (expand_builtin_atomic_test_and_set): Likewise.
15912 (fold_builtin_atomic_always_lock_free): Likewise.
15913 * calls.c (compute_argument_addresses): Likewise.
15914 (emit_library_call_value_1): Likewise.
15915 (store_one_arg): Likewise.
15916 * combine.c (combine_instructions): Likewise.
15917 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15918 * config/arm/arm.c (arm_function_value): Likewise.
15919 (aapcs_allocate_return_reg): Likewise.
15920 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15921 * config/i386/i386.c (construct_container): Likewise.
15922 (ix86_gimplify_va_arg): Likewise.
15923 (ix86_expand_sse_cmp): Likewise.
15924 (emit_memmov): Likewise.
15925 (emit_memset): Likewise.
15926 (expand_small_movmem_or_setmem): Likewise.
15927 (ix86_expand_pextr): Likewise.
15928 (ix86_expand_pinsr): Likewise.
15929 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15930 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15932 * config/mips/mips.c (mips_function_value_1) Likewise.
15933 (mips_block_move_straight): Likewise.
15934 (mips_expand_ins_as_unaligned_store): Likewise.
15935 * config/powerpcspe/powerpcspe.c
15936 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15937 (rs6000_darwin64_record_arg_flush): Likewise.
15938 * config/rs6000/rs6000.c
15939 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15940 (rs6000_darwin64_record_arg_flush): Likewise.
15941 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15942 (sparc_function_value_1): Likewise.
15943 * config/spu/spu.c (adjust_operand): Likewise.
15944 (spu_emit_branch_or_set): Likewise.
15945 (arith_immediate_p): Likewise.
15946 * emit-rtl.c (gen_lowpart_common): Likewise.
15947 * expr.c (expand_expr_real_1): Likewise.
15948 * function.c (assign_parm_setup_block): Likewise.
15949 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15950 * reload1.c (alter_reg): Likewise.
15951 * stor-layout.c (mode_for_vector): Likewise.
15952 (layout_type): Likewise.
15954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15956 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15957 (spu_convert_move): Likewise.
15958 * lower-subreg.c (resolve_simple_move): Likewise.
15960 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15963 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15964 define_insn to a define_expand.
15965 (altivec_vsum2sws_direct): New define_insn.
15966 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15968 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15970 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15973 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15978 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15979 realign stack if stack alignment needed is less than incoming
15982 2017-09-05 Marek Polacek <polacek@redhat.com>
15985 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15988 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15990 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15992 2017-09-05 Richard Biener <rguenther@suse.de>
15994 PR tree-optimization/82084
15995 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15997 2017-09-05 Richard Biener <rguenther@suse.de>
15999 PR tree-optimization/82102
16000 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16002 2017-09-05 Martin Liska <mliska@suse.cz>
16004 PR tree-optimization/82032
16005 * tree-cfg.c (generate_range_test): New function.
16006 * tree-cfg.h (generate_range_test): Declared here.
16007 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16008 (cleanup_control_expr_graph): Use it.
16009 * tree-switch-conversion.c (try_switch_expansion): Remove
16011 (emit_case_nodes): Use generate_range_test.
16013 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16016 * config/i386/i386.md (*<btsc><mode>_mask): Add
16017 TARGET_USE_BT to insn constraint.
16018 (*btr<mode>_mask): Ditto.
16020 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16022 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16023 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16025 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16028 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16029 TARGET_NEON and TARGET_IWMMXT.
16030 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16031 TARGET_NEON and TARGET_IWMMXT.
16032 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16034 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16036 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16037 (ix86_rewrite_tls_address): Ditto.
16038 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16039 (ix86_rewrite_tls_address_1): Ditto.
16040 (ix86_rewrite_tls_address): Ditto.
16041 * config/i386/predicates.md (tls_address_pattern): New predicate.
16042 * config/i386/i386.md (TLS address splitter): New splitter.
16044 2017-09-04 Richard Biener <rguenther@suse.de>
16046 PR tree-optimization/82084
16047 * fold-const.h (can_native_encode_string_p): Declare.
16048 * fold-const.c (can_native_encode_string_p): Factor out from ...
16049 (native_encode_string): ... here.
16050 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16051 vectorizing stores from constants we later cannot handle.
16053 2017-09-04 Marek Polacek <polacek@redhat.com>
16056 * doc/invoke.texi: Update -Wtautological-compare documentation.
16058 2017-09-04 Jeff Law <law@redhat.com>
16060 PR tree-optimization/64910
16061 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16062 swap the first and last operand if the last is a constant.
16064 2017-09-04 Marek Polacek <polacek@redhat.com>
16067 * convert.c (do_narrow): When sanitizing signed integer overflows,
16068 bail out for signed types.
16069 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16071 2017-09-04 Richard Biener <rguenther@suse.de>
16073 PR tree-optimization/82060
16074 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16075 Move devirtualization after stmt folding and before EH/AB/noreturn
16076 cleanup to get the stmt refs canonicalized. Use a bool instead
16077 of gimple_modified_p since that doesn't work for NOPs. Schedule
16078 NOPs generated by folding for removal.
16080 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16081 Alan Hayward <alan.hayward@arm.com>
16082 David Sherwood <david.sherwood@arm.com>
16084 * coretypes.h (pad_direction): New enum.
16085 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16086 (FUNCTION_ARG_PADDING): Likewise.
16087 * target.def (function_arg_padding): New hook.
16088 * targhooks.h (default_function_arg_padding): Declare.
16089 * targhooks.c (default_function_arg_padding): New function.
16090 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16091 (TARGET_FUNCTION_ARG_PADDING): ...this.
16092 * doc/tm.texi: Regenerate.
16093 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16094 instead of direction.
16095 (compute_argument_addresses): Likewise.
16096 (load_register_parameters): Likewise.
16097 (emit_library_call_value_1): Likewise.
16098 (store_one_arg): Use targetm.calls.function_arg_padding instead
16099 of FUNCTION_ARG_PADDING.
16100 (must_pass_in_stack_var_size_or_pad): Likewise.
16101 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16102 (emit_group_store): Likewise.
16103 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16104 instead of FUNCTION_ARG_PADDING.
16105 (emit_push_insn): Likewise, and propagate enum change throughout
16107 * function.h (direction): Delete.
16108 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16110 * function.c (assign_parm_find_stack_rtl): Likewise.
16111 (assign_parm_setup_block_p): Likewise.
16112 (assign_parm_setup_block): Likewise.
16113 (gimplify_parameters): Likewise.
16114 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16115 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16117 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16118 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16119 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16120 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16121 (aarch64_function_arg_padding): ...this new function.
16122 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16123 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16124 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16125 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16126 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16127 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16128 (arm_pad_arg_upward): Replace with...
16129 (arm_function_arg_padding): ...this new function.
16130 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16132 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16133 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16134 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16135 (ia64_hpux_function_arg_padding): Replace with...
16136 (ia64_function_arg_padding): ...this new function. Use pad_direction
16137 instead of direction. Check for TARGET_HPUX.
16138 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16139 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16140 (iq2000_function_arg_padding): New function.
16141 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16142 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16143 (mips_function_arg_padding): ...this new function.
16144 (mips_pad_reg_upward): Update accordingly.
16145 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16146 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16147 targetm.calls.function_arg_padding.
16148 (FUNCTION_ARG_PADDING): Delete.
16149 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16150 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16151 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16152 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16153 (nios2_block_reg_padding): Return pad_direction instead of direction.
16154 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16155 instead of direction.
16156 (nios2_function_arg_padding): Likewise. Make static.
16157 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16158 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16159 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16160 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16161 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16162 (pa_function_arg_padding): Make static. Return pad_direction instead
16164 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16165 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16166 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16167 instead of direction. Use targetm.calls.function_arg_padding.
16168 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16169 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16170 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16171 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16172 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16174 (function_arg_padding): Rename to...
16175 (rs6000_function_arg_padding): ...this. Make static. Return
16176 pad_direction instead of direction.
16177 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16178 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16179 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16180 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16181 instead of direction. Use targetm.calls.function_arg_padding.
16182 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16183 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16184 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16185 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16186 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16187 (function_arg_padding): Rename to...
16188 (rs6000_function_arg_padding): ...this. Make static. Return
16189 pad_direction instead of direction.
16190 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16191 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16192 * config/s390/s390.c (s390_function_arg_padding): New function.
16193 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16194 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16195 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16196 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16197 (function_arg_padding): Rename to...
16198 (sparc_function_arg_padding): ...this. Make static. Return
16199 pad_direction instead of direction.
16200 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16201 * config/spu/spu.c (spu_function_arg_padding): New function.
16202 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16203 * system.h (FUNCTION_ARG_PADDING): Poison.
16205 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16206 Alan Hayward <alan.hayward@arm.com>
16207 David Sherwood <david.sherwood@arm.com>
16209 * target.def (modes_tieable_p): New hook.
16210 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16211 (TARGET_MODES_TIEABLE_P): ...this.
16212 * doc/tm.texi.in: Regenerate.
16213 * hooks.h (hook_bool_mode_mode_true): Declare.
16214 * hooks.c (hook_bool_mode_mode_true): New function.
16215 * combine.c (subst): Use targetm.modes_tieable_p instead of
16217 * dse.c (find_shift_sequence): Likewise.
16218 * expmed.c (extract_low_bits): Likewise.
16219 * lower-subreg.c: Include target.h.
16220 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16222 * rtlanal.c (rtx_cost): Likewise.
16223 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16224 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16225 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16226 (TARGET_MODES_TIEABLE_P): Redefine.
16227 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16228 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16229 (TARGET_MODES_TIEABLE_P): Redefine.
16230 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16231 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16232 (arc_modes_tieable_p): New function.
16233 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16234 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16235 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16236 (arm_modes_tieable_p): Make static.
16237 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16238 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16239 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16240 (TARGET_MODES_TIEABLE_P): Redefine.
16241 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16242 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16243 (TARGET_MODES_TIEABLE_P): Redefine.
16244 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16245 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16246 (cr16_modes_tieable_p): New function.
16247 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16248 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16249 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16250 (TRULY_NOOP_TRUNCATION): Update comment.
16251 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16252 (TRULY_NOOP_TRUNCATION): Update comment.
16253 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16254 (frv_modes_tieable_p): New function.
16255 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16256 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16257 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16258 (TARGET_MODES_TIEABLE_P): Redefine.
16259 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16260 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16261 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16262 (TARGET_MODES_TIEABLE_P): Redefine.
16263 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16264 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16265 (ia64_modes_tieable_p): New function.
16266 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16267 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16268 (iq2000_modes_tieable_p): New function.
16269 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16270 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16271 (lm32_modes_tieable_p): New function.
16272 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16273 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16274 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16275 (TARGET_MODES_TIEABLE_P): Redefine.
16276 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16277 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16278 (m32r_modes_tieable_p): New function.
16279 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16280 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16281 (m68k_modes_tieable_p): New function.
16282 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16283 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16284 (mcore_modes_tieable_p): New function.
16285 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16286 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16288 (TARGET_MODES_TIEABLE_P): Redefine.
16289 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16290 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16291 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16292 (TARGET_MODES_TIEABLE_P): Redefine.
16293 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16294 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16295 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16296 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16297 (mn10300_modes_tieable_p): ...this and make static.
16298 (TARGET_MODES_TIEABLE_P): Redefine.
16299 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16300 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16301 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16302 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16303 (msp430_modes_tieable_p): Make static.
16304 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16305 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16306 (TARGET_MODES_TIEABLE_P): Redefine.
16307 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16308 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16309 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16310 (TARGET_MODES_TIEABLE_P): Redefine.
16311 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16312 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16313 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16314 (TARGET_MODES_TIEABLE_P): Redefine.
16315 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16316 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16317 (pdp11_modes_tieable_p): New function.
16318 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16319 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16320 (rs6000_modes_tieable_p): New function.
16321 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16322 * config/powerpcspe/powerpcspe.md: Update comment.
16323 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16324 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16325 (TARGET_MODES_TIEABLE_P): Redefine.
16326 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16327 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16328 (rl78_modes_tieable_p): New function.
16329 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16330 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16331 (rs6000_modes_tieable_p): New function.
16332 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16333 * config/rs6000/rs6000.md: Update comment.
16334 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16335 * config/rx/rx.c (rx_modes_tieable_p): New function.
16336 (TARGET_MODES_TIEABLE_P): Redefine.
16337 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16338 * config/s390/s390.c (s390_modes_tieable_p): New function.
16339 (TARGET_MODES_TIEABLE_P): Redefine.
16340 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16341 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16342 (sh_modes_tieable_p): New function.
16343 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16344 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16345 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16346 (sparc_modes_tieable_p): Make static.
16347 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16348 * config/spu/spu.c (spu_modes_tieable_p): New function.
16349 (TARGET_MODES_TIEABLE_P): Redefine.
16350 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16351 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16352 (TARGET_MODES_TIEABLE_P): Redefine.
16353 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16354 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16355 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16356 * config/v850/v850.c (v850_modes_tieable_p): New function.
16357 (TARGET_MODES_TIEABLE_P): Redefine.
16358 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16359 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16360 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16361 (visium_modes_tieable_p): New function.
16362 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16363 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16364 (xtensa_modes_tieable_p): New function.
16365 * system.h (MODES_TIEABLE_P): Poison.
16367 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16368 Alan Hayward <alan.hayward@arm.com>
16369 David Sherwood <david.sherwood@arm.com>
16371 * target.def (hard_regno_mode_ok): New hook.
16372 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16373 (TARGET_HARD_REGNO_MODE_OK): ...this.
16374 * doc/tm.texi.in: Regenerate.
16375 * hooks.h (hook_bool_uint_mode_true): Declare.
16376 * hooks.c (hook_bool_uint_mode_true): New function.
16377 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16378 HARD_REGNO_MODE_OK.
16379 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16380 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16381 instead of HARD_REGNO_MODE_OK.
16382 * caller-save.c: Include target.h.
16383 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16384 HARD_REGNO_MODE_OK.
16385 * combine.c (can_combine_p): Likewise.
16386 (combinable_i3pat): Likewise.
16387 (can_change_dest_mode): Likewise.
16388 * expr.c (init_expr_target): Likewise.
16389 (convert_move): Likewise.
16390 (convert_modes): Likewise.
16391 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16392 (setup_prohibited_mode_move_regs): Likewise.
16393 * ira.h (target_ira): Likewise.
16394 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16395 * lra-constraints.c (process_alt_operands): Likewise.
16396 (split_reg): Likewise.
16397 * recog.c (peep2_find_free_register): Likewise.
16398 * ree.c (combine_reaching_defs): Likewise.
16399 * regcprop.c (maybe_mode_change): Likewise.
16400 * reginfo.c (init_reg_sets_1): Likewise.
16401 (choose_hard_reg_mode): Likewise.
16402 (simplifiable_subregs): Likewise.
16403 * regrename.c (check_new_reg_p): Likewise.
16404 * reload.c (find_valid_class): Likewise.
16405 (find_valid_class_1): Likewise.
16406 (reload_inner_reg_of_subreg): Likewise.
16407 (push_reload): Likewise.
16408 (combine_reloads): Likewise.
16409 (find_dummy_reload): Likewise.
16410 (find_reloads): Likewise.
16411 * reload1.c (find_reg): Likewise.
16412 (set_reload_reg): Likewise.
16413 (allocate_reload_reg): Likewise.
16414 (choose_reload_regs): Likewise.
16415 (reload_adjust_reg_for_temp): Likewise.
16416 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16417 (simplify_subreg_regno): Likewise.
16418 * sel-sched.c (init_regs_for_mode): Likewise.
16419 * varasm.c (make_decl_rtl): Likewise.
16420 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16421 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16422 HARD_REGNO_MODE_OK.
16423 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16424 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16426 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16427 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16428 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16429 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16430 (arc_mode_class): Delete.
16431 (HARD_REGNO_MODE_OK): Delete.
16432 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16433 (arc_hard_regno_mode_ok): Rename old array to...
16434 (arc_hard_regno_mode_ok_modes): ...this.
16435 (arc_conditional_register_usage): Update accordingly.
16436 (arc_mode_class): Make static.
16437 (arc_hard_regno_mode_ok): New function.
16438 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16439 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16440 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16441 (arm_hard_regno_mode_ok): Make static.
16442 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16443 HARD_REGNO_MODE_OK.
16444 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16445 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16446 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16448 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16449 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16450 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16451 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16452 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16454 * config/bfin/predicates.md (valid_reg_operand): Use
16455 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16456 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16457 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16459 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16460 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16461 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16462 (cr16_hard_regno_mode_ok): Make static and return a bool.
16463 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16464 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16465 (cris_hard_regno_mode_ok): New function.
16466 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16467 (epiphany_mode_class): Delete.
16468 (HARD_REGNO_MODE_OK): Delete.
16469 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16470 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16471 (hard_regno_mode_ok): Rename to...
16472 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16473 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16474 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16475 HARD_REGNO_MODE_OK.
16476 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16477 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16478 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16479 (frv_hard_regno_mode_ok): Make static and return a bool.
16480 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16481 HARD_REGNO_MODE_OK.
16482 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16483 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16484 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16485 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16487 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16488 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16489 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16490 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16492 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16493 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16494 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16495 (ia64_hard_regno_mode_ok): New function.
16496 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16497 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16498 (iq2000_hard_regno_mode_ok): New function.
16499 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16500 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16501 (lm32_hard_regno_mode_ok): New function.
16502 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16503 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16504 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16505 instead of HARD_REGNO_MODE_OK.
16506 (m32c_hard_regno_ok): Rename to...
16507 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16508 (m32c_cannot_change_mode_class): Update accordingly.
16509 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16510 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16511 (m32r_mode_class): Delete.
16512 (HARD_REGNO_MODE_OK): Delete.
16513 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16514 (m32r_hard_regno_mode_ok): Rename to...
16515 (m32r_hard_regno_modes): ...this.
16516 (m32r_mode_class): Make static.
16517 (m32r_hard_regno_mode_ok): New function.
16518 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16519 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16520 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16521 (m68k_hard_regno_mode_ok): Make static.
16522 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16523 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16524 (mcore_hard_regno_mode_ok): New function.
16525 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16526 (HARD_REGNO_MODE_OK): Delete.
16527 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16529 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16530 (microblaze_hard_regno_mode_ok): New function.
16531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16532 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16533 (mips_hard_regno_mode_ok): Delete.
16534 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16535 (mips_hard_regno_mode_ok_p): ...this and make static.
16536 (mips_hard_regno_mode_ok_p): Rename to...
16537 (mips_hard_regno_mode_ok_uncached): ...this.
16538 (mips_hard_regno_mode_ok): New function.
16539 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16540 of HARD_REGNO_MODE_OK.
16541 (mips_option_override): Update after above name changes.
16542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16543 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16544 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16545 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16546 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16547 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16548 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16549 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16550 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16551 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16552 (msp430_hard_regno_mode_ok): Make static and return a bool.
16553 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16554 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16555 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16557 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16558 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16559 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16560 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16561 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16562 (PA_HARD_REGNO_MODE_OK): ...this
16563 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16564 (PA_HARD_REGNO_MODE_OK): ...this.
16565 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16566 (pa_hard_regno_mode_ok): New function.
16567 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16568 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16569 (pdp11_hard_regno_mode_ok): New function.
16570 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16571 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16573 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16576 (rs6000_hard_regno_mode_ok): Rename to...
16577 (rs6000_hard_regno_mode_ok_uncached): ...this.
16578 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16579 (rs6000_hard_regno_mode_ok): New function.
16580 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16581 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16582 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16583 (riscv_hard_regno_mode_ok): ...this and make static.
16584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16585 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16586 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16587 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16588 (rl78_hard_regno_mode_ok): Make static and return bool.
16589 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16590 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16592 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16594 (rs6000_hard_regno_mode_ok): Rename to...
16595 (rs6000_hard_regno_mode_ok_uncached): ...this.
16596 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16597 (rs6000_hard_regno_mode_ok): New function.
16598 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16599 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16600 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16601 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16602 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16603 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16605 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16606 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16607 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16608 (sh_hard_regno_mode_ok): Make static.
16609 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16610 instead of HARD_REGNO_MODE_OK.
16611 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16612 (sparc_mode_class): Delete.
16613 (HARD_REGNO_MODE_OK): Delete.
16614 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16615 (hard_regno_mode_classes): Make static.
16616 (sparc_mode_class): Likewise.
16617 (sparc_hard_regno_mode_ok): New function.
16618 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16619 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16620 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16622 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16623 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16624 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16625 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16626 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16628 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16629 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16630 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16631 (visium_hard_regno_mode_ok): New function.
16632 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16633 instead of HARD_REGNO_MODE_OK.
16634 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16635 (HARD_REGNO_MODE_OK): Delete.
16636 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16637 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16638 (xtensa_option_override): Update accordingly.
16639 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16640 (xtensa_hard_regno_mode_ok): New function.
16641 * system.h (HARD_REGNO_MODE_OK): Poison.
16643 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16644 Alan Hayward <alan.hayward@arm.com>
16645 David Sherwood <david.sherwood@arm.com>
16647 * target.def (hard_regno_call_part_clobbered): New hook.
16648 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16649 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16650 * doc/tm.texi: Regenerate.
16651 * hooks.h (hook_bool_uint_mode_false): Declare.
16652 * hooks.c (hook_bool_uint_mode_false): New function.
16653 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16654 * cselib.c (cselib_process_insn): Use
16655 targetm.hard_regno_call_part_clobbered instead of
16656 HARD_REGNO_CALL_PART_CLOBBERED.
16657 * ira-conflicts.c (ira_build_conflicts): Likewise.
16658 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16659 * lra-constraints.c (need_for_call_save_p): Likewise.
16660 * lra-lives.c: Include target.h.
16661 (check_pseudos_live_through_calls): Use
16662 targetm.hard_regno_call_part_clobbered instead of
16663 HARD_REGNO_CALL_PART_CLOBBERED.
16664 * regcprop.c: Include target.h.
16665 (copyprop_hardreg_forward_1): Use
16666 targetm.hard_regno_call_part_clobbered instead of
16667 HARD_REGNO_CALL_PART_CLOBBERED.
16668 * reginfo.c (choose_hard_reg_mode): Likewise.
16669 * regrename.c (check_new_reg_p): Likewise.
16670 * reload.c (find_equiv_reg): Likewise.
16671 * reload1.c (emit_reload_insns): Likewise.
16672 * sched-deps.c (deps_analyze_insn): Likewise.
16673 * sel-sched.c (init_regs_for_mode): Likewise.
16674 (mark_unavailable_hard_regs): Likewise.
16675 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16676 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16677 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16679 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16680 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16681 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16683 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16686 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16687 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16689 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16690 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16691 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16693 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16694 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16696 * config/powerpcspe/powerpcspe.c
16697 (rs6000_hard_regno_call_part_clobbered): New function.
16698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16699 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16700 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16702 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16703 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16704 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16706 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16707 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16708 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16710 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16711 Alan Hayward <alan.hayward@arm.com>
16712 David Sherwood <david.sherwood@arm.com>
16714 * rtl.h (subreg_memory_offset): Declare.
16715 * emit-rtl.c (subreg_memory_offset): New function.
16716 * expmed.c (store_bit_field_1): Use it.
16717 * expr.c (undefined_operand_subword_p): Likewise.
16718 * simplify-rtx.c (simplify_subreg): Likewise.
16720 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16722 PR rtl-optimization/57448
16725 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16726 using atomic_load pattern.
16727 (expand_atomic_store): Likewise.
16729 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16732 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16733 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16734 handling. Use replace_call_with_value with NULL instead of
16735 gsi_replace, unlink_stmt_vdef and release_defs.
16737 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16740 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16742 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16745 * rtl.h (emit_library_call_value_1): Declare.
16746 (emit_library_call): Replace declaration with a series of overloads.
16747 Remove the parameter count argument.
16748 (emit_library_call_value): Likewise.
16749 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16750 with an "rtx_mode_t *".
16751 (emit_library_call_value): Delete.
16752 (emit_library_call): Likewise.
16753 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16754 (asan_emit_allocas_unpoison): Likewise.
16755 * builtins.c (expand_builtin_powi): Likewise.
16756 (expand_asan_emit_allocas_unpoison): Likewise.
16757 * cfgexpand.c (expand_main_function): Likewise.
16758 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16759 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16760 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16761 * config/arm/arm.c (arm_trampoline_init): Likewise.
16762 (arm_call_tls_get_addr): Likewise.
16763 (arm_expand_divmod_libfunc): Likewise.
16764 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16765 (smulsi3_highpart): Likewise.
16766 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16767 (c6x_expand_compare): Likewise.
16768 (c6x_expand_movmem): Likewise.
16769 * config/frv/frv.c (frv_trampoline_init): Likewise.
16770 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16771 (ix86_expand_divmod_libfunc): Likewise.
16772 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16773 (ia64_expand_compare): Likewise.
16774 (ia64_profile_hook): Likewise.
16775 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16776 (nonlocal_goto): Likewise.
16777 (restore_stack_nonlocal): Likewise.
16778 * config/m32r/m32r.c (block_move_call): Likewise.
16779 (m32r_trampoline_init): Likewise.
16780 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16781 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16782 (m68k_call_m68k_read_tp): Likewise.
16783 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16784 (microblaze_expand_divide): Likewise.
16785 * config/mips/mips.h (mips_args): Likewise.
16786 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16787 (MIPS_ICACHE_SYNC): Likewise.
16788 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16789 (nios2_trampoline_init): Likewise.
16790 * config/pa/pa.c (hppa_tls_call): Likewise.
16791 (pa_trampoline_init): Likewise.
16792 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16793 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16794 (expand_strn_compare): Likewise.
16795 (rs6000_generate_compare): Likewise.
16796 (rs6000_expand_float128_convert): Likewise.
16797 (output_profile_hook): Likewise.
16798 (rs6000_trampoline_init): Likewise.
16799 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16800 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16801 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16802 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16803 (rs6000_generate_compare): Likewise.
16804 (rs6000_expand_float128_convert): Likewise.
16805 (output_profile_hook): Likewise.
16806 (rs6000_trampoline_init): Likewise.
16807 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16808 * config/sh/sh.c (sh_trampoline_init): Likewise.
16809 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16810 (sparc_emit_float_lib_cmp): Likewise.
16811 (sparc32_initialize_trampoline): Likewise.
16812 (sparc64_initialize_trampoline): Likewise.
16813 (sparc_profile_hook): Likewise.
16814 * config/spu/spu.c (ea_load_store): Likewise.
16815 * config/spu/spu.md (floatunssidf2): Likewise.
16816 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16817 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16818 * config/visium/visium.c (expand_block_move_4): Likewise.
16819 (expand_block_move_2): Likewise.
16820 (expand_block_move_1): Likewise.
16821 (expand_block_set_4): Likewise.
16822 (expand_block_set_2): Likewise.
16823 (expand_block_set_1): Likewise.
16824 (visium_trampoline_init): Likewise.
16825 (visium_profile_hook): Likewise.
16826 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16827 (xtensa_setup_frame_addresses): Likewise.
16828 (xtensa_trampoline_init): Likewise.
16829 * except.c (sjlj_emit_function_enter): Likewise.
16830 (sjlj_emit_function_exit): Likewise.
16831 * explow.c (allocate_dynamic_stack_space): Likewise.
16832 (probe_stack_range): Likewise.
16833 * expr.c (convert_mode_scalar): Likewise.
16834 * optabs.c (expand_binop): Likewise.
16835 (expand_twoval_binop_libfunc): Likewise.
16836 (expand_unop): Likewise.
16837 (prepare_cmp_insn): Likewise.
16838 (prepare_float_lib_cmp): Likewise.
16839 (expand_float): Likewise.
16840 (expand_fix): Likewise.
16841 (expand_fixed_convert): Likewise.
16842 (maybe_emit_sync_lock_test_and_set): Likewise.
16843 (expand_atomic_compare_and_swap): Likewise.
16844 (expand_mem_thread_fence): Likewise.
16845 (expand_atomic_fetch_op): Likewise.
16847 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16849 * doc/generic.texi (OpenACC): Adjust URL.
16850 * doc/invoke.texi (C Dialect Options): Ditto.
16852 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16854 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16855 predicate for operand 1. Add (m,<S>) constraint.
16856 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16857 Prevent memory operand 1 with register operand 2.
16859 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16861 PR rtl-optimization/82024
16862 * combine.c (try_combine): If the combination result is a PARALLEL,
16863 and we only need to retain the SET in there that would be placed
16864 at I2, check that we can place that at I3 instead, before doing so.
16866 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16869 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16871 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16872 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16875 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16876 Jeff Law <law@redhat.com>
16878 * varasm.c (bss_initializer_p): Do not put constants into .bss
16879 (categorize_decl_for_section): Handle bss_initializer_p returning
16880 false when DECL_INITIAL is NULL.
16882 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16885 * config/s390/s390.c (s390_can_inline_p): New function.
16887 2017-09-01 Jeff Law <law@redhat.com>
16889 PR tree-optimization/82052
16890 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16891 Always initialize the returned slot after a hash table miss
16892 when INSERT is true.
16894 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16896 * config/s390/s390.md (mem_signal_fence): Remove.
16897 * doc/md.texi (mem_signal_fence): Remove.
16898 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16900 * target-insns.def (mem_signal_fence): Remove.
16902 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16905 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16908 * asan.c (create_odr_indicator): Strip name encoding from assembler
16909 name before appending it after __odr_asan_.
16911 2017-09-01 Martin Liska <mliska@suse.cz>
16913 PR tree-optimization/82059
16914 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16915 frequency only when an edge is redirected.
16917 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16919 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16920 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16921 (arc_conditional_register_usage): Remove ARC600 lp_count
16923 (arc_file_start): Emit Tag_ARC_CPU_variation.
16924 (arc_can_use_doloop_p): New conditions to use ZOLs.
16925 (hwloop_fail): New function.
16926 (hwloop_optimize): Likewise.
16927 (hwloop_pattern_reg): Likewise.
16928 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16929 (arc_reorg_loops): New function, calls reorg_loops.
16930 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16931 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16933 (arc_loop_hazard): Remove function, functionality moved into
16935 (arc_hazard): Remove arc_loop_hazard call.
16936 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16937 into hwloop_optimize.
16938 (arc_label_align): Remove ZOL handling.
16939 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16940 * config/arc/arc.md (doloop_begin): Remove pattern.
16941 (doloop_begin_i): Likewise.
16942 (doloop_end_i): Likewise.
16943 (doloop_fallback): Likewise.
16944 (doloop_fallback_m): Likewise.
16945 (doloop_end): Reimplement expand.
16946 (arc_lp): New pattern for LP instruction.
16947 (loop_end): New pattern.
16948 (loop_fail): Likewise.
16949 (decrement_and_branch_until_zero): Likewise.
16950 * config/arc/arc.opt (mlpc-width): New option.
16951 * doc/invoke.texi (mlpc-width): Document option.
16953 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16955 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16956 (arc_ccfsm_advance): Fix checking for delay slots.
16957 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16959 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16961 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16963 (movhi_insn): Update store instruction constraint which are saving
16964 6-bit short immediates.
16965 (movsi_insn): Consider also short scaled load operations.
16966 (zero_extendhisi2_i): Use Usd constraint instead of T.
16967 (extendhisi2_i): Add q constraint.
16968 (arc_clzsi2): Add type and length attributes.
16969 (arc_ctzsi2): Likewise.
16970 * config/arc/constraints.md (Usc): Update constraint, the
16971 assembler can parse two relocations for a single instruction.
16973 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16975 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16976 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16978 2017-08-31 Olivier Hainque <hainque@adacore.com>
16980 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16981 match as powerpc-wrs-vxworks*.
16983 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16985 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16986 register constraint for by-element operand.
16987 (aarch64_mls_elt_merge<mode>): Likewise.
16989 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16991 * config/arc/arc.c (arc_can_follow_jump): Check for short
16994 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16996 * config.gcc: Use g.opt for arc.
16997 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16998 functionality moved to ...
16999 (legitimate_scaled_address_p): New function, ...here.
17000 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17001 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17002 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17004 (arc_override_options): Handle G option.
17005 (arc_output_pic_addr_const): Correct function definition.
17006 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17007 (arc_decl_anon_ns_mem_p): Delete.
17008 (arc_in_small_data_p): Overhaul this function to take into
17009 consideration the value given via G option.
17010 (arc_rewrite_small_data_1): Renamed and corrected old
17011 arc_rewrite_small_data function.
17012 (arc_rewrite_small_data): New function.
17013 (small_data_pattern): Don't use pic_offset_table_rtx.
17014 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17015 * config/arc/simdext.md (movmisalignv2hi): Use
17016 prepare_move_operands function.
17018 (movmisalign*): Likewise.
17019 * doc/invoke.texi (ARC options): Document -G option.
17021 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17023 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17025 * config/arc/arc.c (arc_print_operand): Output scalled address for
17026 sdata whenever is possible.
17027 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17028 load/stores are available.
17029 (compact_sda_memory_operand): Check for the alignment required by
17030 code density instructions.
17031 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17033 * config/arc/constraints.md (Usd): Update constraint.
17034 (Us0): New constraint.
17035 (Usc): Update constraint.
17037 2017-08-31 Richard Biener <rguenther@suse.de>
17039 PR middle-end/82054
17040 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17041 function only once.
17043 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17045 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17046 Resize type_signature.
17048 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17049 Alan Hayward <alan.hayward@arm.com>
17050 David Sherwood <david.sherwood@arm.com>
17052 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17053 subregs whose inner modes can be stored in GPRs.
17054 (aarch64_classify_index): Likewise.
17056 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17057 Alan Hayward <alan.hayward@arm.com>
17058 David Sherwood <david.sherwood@arm.com>
17060 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17061 (V_INT_EQUIV): ...this.
17062 (v_cmp_result): Rename to...
17063 (v_int_equiv): ...this.
17064 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17065 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17066 (copysign<mode>3): Likewise.
17067 (aarch64_simd_bsl<mode>_internal): Likewise.
17068 (aarch64_simd_bsl<mode>): Likewise.
17069 (vec_cmp<mode><mode>): Likewise.
17070 (vcond<mode><mode>): Likewise.
17071 (vcond<v_cmp_mixed><mode>): Likewise.
17072 (vcondu<mode><v_cmp_mixed>): Likewise.
17073 (aarch64_cm<optab><mode>): Likewise.
17074 (aarch64_cmtst<mode>): Likewise.
17075 (aarch64_fac<optab><mode>): Likewise.
17076 (vec_perm_const<mode>): Likewise.
17077 (vcond_mask_<mode><v_cmp_result>): Rename to...
17078 (vcond_mask_<mode><v_int_equiv>): ...this.
17079 (vec_cmp<mode><v_cmp_result>): Rename to...
17080 (vec_cmp<mode><v_int_equiv>): ...this.
17082 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17083 Alan Hayward <alan.hayward@arm.com>
17084 David Sherwood <david.sherwood@arm.com>
17086 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17088 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17089 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17090 (UNSPEC_LD4_DREG): New unspecs.
17091 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17092 (aarch64_ld2<mode>_dreg_be): Replace with...
17093 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17095 (aarch64_ld3<mode>_dreg_le)
17096 (aarch64_ld3<mode>_dreg_be): Replace with...
17097 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17099 (aarch64_ld4<mode>_dreg_le)
17100 (aarch64_ld4<mode>_dreg_be): Replace with...
17101 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17104 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17106 PR tree-optimization/81987
17107 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17108 insert an initializer in a location not dominated by the stride
17111 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17113 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17114 on the entire header of the finally block in the fallthru case.
17116 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17118 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17120 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17122 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17123 rs6000_emit_move_from_cr and call renamed function.
17124 (rs6000_emit_prologue): Call renamed functions.
17125 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17126 movesi_from_cr, remove volatile CRs.
17128 2017-08-30 Jon Beniston <jon@beniston.com>
17129 Richard Biener <rguenther@suse.de>
17131 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17132 of VECTOR_MODE_P check.
17133 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17134 element vector types.
17136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17138 * df.h (df_read_modify_subreg_p): Remove in favor of...
17139 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17140 const_rtx instead of an rtx.
17141 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17142 * df-problems.c (df_word_lr_mark_ref): Likewise.
17143 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17144 (mark_pseudo_reg_dead): Likewise.
17145 (mark_ref_dead): Likewise.
17146 * reginfo.c (init_subregs_of_mode): Likewise.
17147 * sched-deps.c (sched_analyze_1): Likewise.
17148 * df-scan.c (df_def_record_1): Likewise.
17149 (df_uses_record): Likewise.
17150 (df_read_modify_subreg_p): Remove in favor of...
17151 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17152 const_rtx instead of an rtx.
17154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17155 Alan Hayward <alan.hayward@arm.com>
17156 David Sherwood <david.sherwood@arm.com>
17158 * rtl.h (partial_subreg_p): New function.
17159 * caller-save.c (save_call_clobbered_regs): Use it.
17160 * calls.c (expand_call): Likewise.
17161 * combine.c (combinable_i3pat): Likewise.
17162 (simplify_set): Likewise.
17163 (make_extraction): Likewise.
17164 (make_compound_operation_int): Likewise.
17165 (gen_lowpart_or_truncate): Likewise.
17166 (force_to_mode): Likewise.
17167 (make_field_assignment): Likewise.
17168 (reg_truncated_to_mode): Likewise.
17169 (record_truncated_value): Likewise.
17170 (move_deaths): Likewise.
17171 * cse.c (record_jump_cond): Likewise.
17172 (cse_insn): Likewise.
17173 * cselib.c (cselib_lookup_1): Likewise.
17174 * expmed.c (extract_bit_field_using_extv): Likewise.
17175 * function.c (assign_parm_setup_reg): Likewise.
17176 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17177 * ira-build.c (create_insn_allocnos): Likewise.
17178 * lra-coalesce.c (merge_pseudos): Likewise.
17179 * lra-constraints.c (match_reload): Likewise.
17180 (simplify_operand_subreg): Likewise.
17181 (curr_insn_transform): Likewise.
17182 * lra-lives.c (process_bb_lives): Likewise.
17183 * lra.c (new_insn_reg): Likewise.
17184 (lra_substitute_pseudo): Likewise.
17185 * regcprop.c (mode_change_ok): Likewise.
17186 (maybe_mode_change): Likewise.
17187 (copyprop_hardreg_forward_1): Likewise.
17188 * reload.c (push_reload): Likewise.
17189 (find_reloads): Likewise.
17190 (find_reloads_subreg_address): Likewise.
17191 * reload1.c (alter_reg): Likewise.
17192 (eliminate_regs_1): Likewise.
17193 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17195 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17197 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17198 back to if statements, including unpack.
17200 2017-08-30 Martin Liska <mliska@suse.cz>
17202 PR inline-asm/82001
17203 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17205 (func_checker::compare_asm_inputs_outputs): ... this function.
17206 (func_checker::compare_gimple_asm): Use the function to compare
17207 also ASM constrains.
17208 * ipa-icf-gimple.h: Rename the function.
17210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17211 Alan Hayward <alan.hayward@arm.com>
17212 David Sherwood <david.sherwood@arm.com>
17214 * coretypes.h (complex_mode): New type.
17215 * gdbhooks.py (build_pretty_printer): Handle it.
17216 * machmode.h (complex_mode): New class.
17217 (complex_mode::includes_p): New function.
17218 (is_complex_int_mode): Likewise.
17219 (is_complex_float_mode): Likewise.
17220 * genmodes.c (get_mode_class): Handle complex mode classes.
17221 * function.c (expand_function_end): Use is_complex_int_mode.
17223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17224 Alan Hayward <alan.hayward@arm.com>
17225 David Sherwood <david.sherwood@arm.com>
17227 * coretypes.h (scalar_mode_pod): New typedef.
17228 * gdbhooks.py (build_pretty_printer): Handle it.
17229 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17230 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17231 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17232 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17233 as_a <scalar_mode>.
17235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17236 Alan Hayward <alan.hayward@arm.com>
17237 David Sherwood <david.sherwood@arm.com>
17239 * machmode.h (mode_for_vector): Take a scalar_mode instead
17241 * stor-layout.c (mode_for_vector): Likewise.
17242 * explow.c (promote_mode): Use as_a <scalar_mode>.
17243 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17246 Alan Hayward <alan.hayward@arm.com>
17247 David Sherwood <david.sherwood@arm.com>
17249 * target.def (preferred_simd_mode): Take a scalar_mode
17250 instead of a machine_mode.
17251 * targhooks.h (default_preferred_simd_mode): Likewise.
17252 * targhooks.c (default_preferred_simd_mode): Likewise.
17253 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17254 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17255 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17256 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17257 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17258 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17259 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17260 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17262 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17263 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17264 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17265 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17266 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17267 * doc/tm.texi: Regenerate.
17268 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17272 Alan Hayward <alan.hayward@arm.com>
17273 David Sherwood <david.sherwood@arm.com>
17275 * target.def (scalar_mode_supported_p): Take a scalar_mode
17276 instead of a machine_mode.
17277 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17278 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17279 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17280 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17281 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17282 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17283 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17284 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17285 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17286 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17287 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17288 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17289 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17290 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17292 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17293 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17294 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17295 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17296 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17298 * doc/tm.texi: Regenerate.
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 * coretypes.h (opt_scalar_mode): New typedef.
17305 * gdbhooks.py (build_pretty_printers): Handle it.
17306 * machmode.h (mode_iterator::get_2xwider): Add overload for
17308 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17310 * expr.c (convert_mode_scalar): Likewise.
17311 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17312 * optabs.c (expand_float): Likewise.
17313 (expand_fix): Likewise.
17314 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17317 Alan Hayward <alan.hayward@arm.com>
17318 David Sherwood <david.sherwood@arm.com>
17320 * optabs.c (expand_float): Explicitly check for scalars before
17321 using a branching expansion.
17322 (expand_fix): Likewise.
17324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17325 Alan Hayward <alan.hayward@arm.com>
17326 David Sherwood <david.sherwood@arm.com>
17328 * expr.c (convert_mode): Split scalar handling out into...
17329 (convert_mode_scalar): ...this new function. Treat the modes
17332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17333 Alan Hayward <alan.hayward@arm.com>
17334 David Sherwood <david.sherwood@arm.com>
17336 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17338 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17341 Alan Hayward <alan.hayward@arm.com>
17342 David Sherwood <david.sherwood@arm.com>
17344 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17345 rather than a machine_mode.
17346 (fixed_from_string): Likewise.
17347 (fixed_convert): Likewise.
17348 (fixed_convert_from_int): Likewise.
17349 (fixed_convert_from_real): Likewise.
17350 (real_convert_from_fixed): Likewise.
17351 * fixed-value.c (fixed_from_double_int): Likewise.
17352 (fixed_from_string): Likewise.
17353 (fixed_convert): Likewise.
17354 (fixed_convert_from_int): Likewise.
17355 (fixed_convert_from_real): Likewise.
17356 (real_convert_from_fixed): Likewise.
17357 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17360 Alan Hayward <alan.hayward@arm.com>
17361 David Sherwood <david.sherwood@arm.com>
17363 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17364 of separate mode class checks. Do not allow vector modes here.
17365 (immed_wide_int_const): Use as_a <scalar_mode>.
17366 * explow.c (trunc_int_for_mode): Likewise.
17367 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17368 (wi::shwi): Likewise.
17369 (wi::min_value): Likewise.
17370 (wi::max_value): Likewise.
17371 * dwarf2out.c (loc_descriptor): Likewise.
17372 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17373 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
17380 * expr.c (expand_expr_addr_expr_1): Use it.
17381 (expand_expr_real_2): Likewise.
17382 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17383 (fold_convert_const_fixed_from_int): Likewise.
17384 (fold_convert_const_fixed_from_real): Likewise.
17385 (native_encode_fixed): Likewise
17386 (native_encode_complex): Likewise
17387 (native_encode_vector): Likewise.
17388 (native_interpret_fixed): Likewise.
17389 (native_interpret_real): Likewise.
17390 (native_interpret_complex): Likewise.
17391 (native_interpret_vector): Likewise.
17392 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17393 (simd_clone_adjust_argument_types): Likewise.
17394 (simd_clone_init_simd_arrays): Likewise.
17395 (simd_clone_adjust): Likewise.
17396 * stor-layout.c (layout_type): Likewise.
17397 * tree.c (build_minus_one_cst): Likewise.
17398 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17399 * tree-inline.c (estimate_move_cost): Likewise.
17400 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17401 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17402 (vectorizable_reduction): Likewise.
17403 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17404 (vect_recog_mixed_size_cond_pattern): Likewise.
17405 (check_bool_pattern): Likewise.
17406 (adjust_bool_pattern): Likewise.
17407 (search_type_for_mask_1): Likewise.
17408 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17409 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17410 (vectorizable_load): Likewise.
17411 (vectorizable_store): Likewise.
17412 * ubsan.c (ubsan_encode_value): Likewise.
17413 * varasm.c (output_constant): Likewise.
17415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17416 Alan Hayward <alan.hayward@arm.com>
17417 David Sherwood <david.sherwood@arm.com>
17419 * coretypes.h (scalar_mode): New class.
17420 * machmode.h (scalar_mode): Likewise.
17421 (scalar_mode::includes_p): New function.
17422 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17423 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17424 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17425 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17426 * expmed.c (store_bit_field_1): Likewise.
17427 (extract_bit_field_1): Likewise.
17428 * expr.c (write_complex_part): Likewise.
17429 (read_complex_part): Likewise.
17430 (emit_move_complex_push): Likewise.
17431 (expand_expr_real_2): Likewise.
17432 * function.c (assign_parm_setup_reg): Likewise.
17433 (assign_parms_unsplit_complex): Likewise.
17434 * optabs.c (expand_binop): Likewise.
17435 * rtlanal.c (subreg_get_info): Likewise.
17436 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17437 * varasm.c (output_constant_pool_2): Likewise.
17439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17440 Alan Hayward <alan.hayward@arm.com>
17441 David Sherwood <david.sherwood@arm.com>
17443 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17445 (expmed_mult_highpart_optab): Likewise.
17446 (expmed_mult_highpart): Likewise.
17448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17449 Alan Hayward <alan.hayward@arm.com>
17450 David Sherwood <david.sherwood@arm.com>
17452 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17453 instead of a machine_mode.
17454 (builtin_memset_read_str): Likewise.
17455 * builtins.c (c_readstr): Likewise.
17456 (builtin_memcpy_read_str): Likewise.
17457 (builtin_strncpy_read_str): Likewise.
17458 (builtin_memset_read_str): Likewise.
17459 (builtin_memset_gen_str): Likewise.
17460 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17461 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17462 instead of a machine_mode.
17463 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17465 (make_extraction): Likewise.
17466 (try_widen_shift_mode): Take and return scalar_int_modes instead
17468 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17469 a scalar_int_mode instead of a machine_mode.
17470 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17471 (avr_addr_space_pointer_mode): Likewise.
17472 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17473 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17474 (msp430_unwind_word_mode): Likewise.
17475 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17476 (spu_addr_space_pointer_mode): Likewise.
17477 (spu_addr_space_address_mode): Likewise.
17478 (spu_libgcc_cmp_return_mode): Likewise.
17479 (spu_libgcc_shift_count_mode): Likewise.
17480 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17481 (rl78_addr_space_pointer_mode): Likewise.
17482 (fl78_unwind_word_mode): Likewise.
17483 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17485 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17486 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17487 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17488 (mips_valid_pointer_mode): Likewise.
17489 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17490 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17491 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17493 (ft32_addr_space_address_mode): Likewise.
17494 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17495 scalar_int_mode instead of a machine_mode.
17496 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17498 (m32c_addr_space_address_mode): Likewise.
17499 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17500 (rs6000_eh_return_filter_mode): Likewise.
17501 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17502 (rs6000_eh_return_filter_mode): Likewise.
17503 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17504 (s390_libgcc_shift_count_mode): Likewise.
17505 (s390_unwind_word_mode): Likewise.
17506 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17508 * target.def (mode_rep_extended): Likewise.
17509 (valid_pointer_mode): Likewise.
17510 (addr_space.valid_pointer_mode): Likewise.
17511 (eh_return_filter_mode): Return a scalar_int_mode rather than
17513 (libgcc_cmp_return_mode): Likewise.
17514 (libgcc_shift_count_mode): Likewise.
17515 (unwind_word_mode): Likewise.
17516 (addr_space.pointer_mode): Likewise.
17517 (addr_space.address_mode): Likewise.
17518 * doc/tm.texi: Regenerate.
17519 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17521 (do_jump): Use scalar_int_mode for local variables.
17522 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17523 rather than a machine_mode.
17524 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17525 (scompare_loc_descriptor_wide): Likewise.
17526 (scompare_loc_descriptor_narrow): Likewise.
17527 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17529 * except.c (sjlj_emit_dispatch_table): Likewise.
17530 (expand_builtin_eh_copy_values): Likewise.
17531 * explow.c (convert_memory_address_addr_space_1): Likewise.
17532 Take a scalar_int_mode rather than a machine_mode.
17533 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17534 than a machine_mode.
17535 (memory_address_addr_space): Use scalar_int_mode for local variables.
17536 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17537 rather than a machine_mode.
17538 * expmed.c (mask_rtx): Likewise.
17539 (init_expmed_one_conv): Likewise.
17540 (expand_mult_highpart_adjust): Likewise.
17541 (extract_high_half): Likewise.
17542 (expmed_mult_highpart_optab): Likewise.
17543 (expmed_mult_highpart): Likewise.
17544 (expand_smod_pow2): Likewise.
17545 (expand_sdiv_pow2): Likewise.
17546 (emit_store_flag_int): Likewise.
17547 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17549 (extract_low_bits): Likewise.
17550 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17552 * expr.c (pieces_addr::adjust): Likewise.
17553 (can_store_by_pieces): Likewise.
17554 (store_by_pieces): Likewise.
17555 (clear_by_pieces_1): Likewise.
17556 (expand_expr_addr_expr_1): Likewise.
17557 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17558 (expand_expr_real_1): Likewise.
17559 (try_casesi): Likewise.
17560 * final.c (shorten_branches): Likewise.
17561 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17562 type of "mode" to machine_mode.
17563 * internal-fn.c (expand_arith_overflow_result_store): Take a
17564 scalar_int_mode rather than a machine_mode.
17565 (expand_mul_overflow): Use scalar_int_mode for local variables.
17566 * loop-doloop.c (doloop_modify): Likewise.
17567 (doloop_optimize): Likewise.
17568 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17569 than a machine_mode.
17570 (expand_doubleword_shift_condmove): Likewise.
17571 (expand_doubleword_shift): Likewise.
17572 (expand_doubleword_clz): Likewise.
17573 (expand_doubleword_popcount): Likewise.
17574 (expand_doubleword_parity): Likewise.
17575 (expand_absneg_bit): Use scalar_int_mode for local variables.
17576 (prepare_float_lib_cmp): Likewise.
17577 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17578 rather than a machine_mode.
17579 (convert_memory_address_addr_space): Likewise.
17580 (get_mode_bounds): Likewise.
17581 (get_address_mode): Return a scalar_int_mode rather than a
17583 * rtlanal.c (get_address_mode): Likewise.
17584 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17585 than a machine_mode.
17586 * targhooks.c (default_mode_rep_extended): Likewise.
17587 (default_valid_pointer_mode): Likewise.
17588 (default_addr_space_valid_pointer_mode): Likewise.
17589 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17590 than a machine_mode.
17591 (default_libgcc_cmp_return_mode): Likewise.
17592 (default_libgcc_shift_count_mode): Likewise.
17593 (default_unwind_word_mode): Likewise.
17594 (default_addr_space_pointer_mode): Likewise.
17595 (default_addr_space_address_mode): Likewise.
17596 * targhooks.h (default_eh_return_filter_mode): Likewise.
17597 (default_libgcc_cmp_return_mode): Likewise.
17598 (default_libgcc_shift_count_mode): Likewise.
17599 (default_unwind_word_mode): Likewise.
17600 (default_addr_space_pointer_mode): Likewise.
17601 (default_addr_space_address_mode): Likewise.
17602 (default_mode_rep_extended): Take a scalar_int_mode rather than
17604 (default_valid_pointer_mode): Likewise.
17605 (default_addr_space_valid_pointer_mode): Likewise.
17606 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17608 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17609 rather than a machine_mode.
17610 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17611 for local variables.
17612 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17613 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17614 than a machine_mode.
17616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17617 Alan Hayward <alan.hayward@arm.com>
17618 David Sherwood <david.sherwood@arm.com>
17620 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17621 the mode argument to scalar_int_mode.
17622 (do_jump_by_parts_zero_rtx): Likewise.
17623 (do_jump_by_parts_equality_rtx): Likewise.
17624 (do_jump_by_parts_greater): Take a mode argument.
17625 (do_jump_by_parts_equality): Likewise.
17626 (do_jump_1): Update calls accordingly.
17628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17629 Alan Hayward <alan.hayward@arm.com>
17630 David Sherwood <david.sherwood@arm.com>
17632 * is-a.h (safe_dyn_cast): New function.
17633 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17634 (jump_table_for_label): Likewise.
17635 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17636 instead of an rtx_insn *.
17637 (shorten_branches): Use dyn_cast instead of LABEL_P and
17638 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17639 rtx_jump_table_data::get_data_mode.
17640 (final_scan_insn): Likewise.
17642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17643 Alan Hayward <alan.hayward@arm.com>
17644 David Sherwood <david.sherwood@arm.com>
17646 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17647 trying to combine a full-register integer set with a subreg
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 (expand_expr_addr_expr): Add a new_tmode local variable
17655 that is always either address_mode or pointer_mode.
17657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17658 Alan Hayward <alan.hayward@arm.com>
17659 David Sherwood <david.sherwood@arm.com>
17661 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17662 when the two are known to be equal.
17664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17665 Alan Hayward <alan.hayward@arm.com>
17666 David Sherwood <david.sherwood@arm.com>
17668 * simplify-rtx.c (simplify_const_unary_operation): Use
17669 is_a <scalar_int_mode> instead of checking for a nonzero
17670 precision. Forcibly convert op_mode to a scalar_int_mode
17671 in that case. More clearly differentiate the operand and
17672 result modes and use the former when deciding what the value
17673 of a count-bits operation should be. Use is_int_mode instead
17674 of checking for a MODE_INT. Remove redundant check for whether
17675 this mode has a zero precision.
17677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17678 Alan Hayward <alan.hayward@arm.com>
17679 David Sherwood <david.sherwood@arm.com>
17681 * optabs.c (widen_leading): Change the type of the mode argument
17682 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17683 (widen_bswap): Likewise.
17684 (expand_parity): Likewise.
17685 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17686 (expand_ffs): Likewise.
17687 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17696 (expand_expr_real_1): Update call accordingly.
17698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17699 Alan Hayward <alan.hayward@arm.com>
17700 David Sherwood <david.sherwood@arm.com>
17702 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17703 value_mode arguments. Use scalar_int_mode internally.
17704 (store_bit_field_1): Rename the new integer mode from imode
17705 to op0_mode and use it instead of GET_MODE (op0). Update calls
17706 to store_split_bit_field, store_bit_field_using_insv and
17707 store_fixed_bit_field.
17708 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17709 Use scalar_int_mode internally. Use a bit count rather than a mode
17710 when calculating the largest bit size for get_best_mode.
17711 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17712 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17713 Remove assertion that OP0 has a scalar integer mode.
17714 (store_split_bit_field): Add op0_mode and value_mode arguments.
17715 Update calls to extract_fixed_bit_field.
17716 (extract_bit_field_using_extv): Add an op0_mode argument.
17717 Use scalar_int_mode internally.
17718 (extract_bit_field_1): Rename the new integer mode from imode to
17719 op0_mode and use it instead of GET_MODE (op0). Update calls to
17720 extract_split_bit_field, extract_bit_field_using_extv and
17721 extract_fixed_bit_field.
17722 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17723 to extract_split_bit_field and extract_fixed_bit_field_1.
17724 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17725 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17726 on the target mode.
17727 (extract_split_bit_field): Add an op0_mode argument. Update call
17728 to extract_fixed_bit_field.
17730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17731 Alan Hayward <alan.hayward@arm.com>
17732 David Sherwood <david.sherwood@arm.com>
17734 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17735 * explow.c (hard_function_value): Likewise.
17736 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17737 convert_to_mode call outside the loop.
17738 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17739 for the mode iterator. Require the mode specified by max_pieces
17741 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17743 (copy_blkmode_to_reg): Likewise.
17744 (set_storage_via_setmem): Likewise.
17745 * optabs.c (prepare_cmp_insn): Likewise.
17746 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17747 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17754 * expr.c (convert_move): Use them.
17755 (convert_modes): Likewise.
17756 (store_expr_with_bounds): Likewise.
17758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17759 Alan Hayward <alan.hayward@arm.com>
17760 David Sherwood <david.sherwood@arm.com>
17762 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17763 parameter for the mode of "x". Remove the "known_x", "known_mode"
17764 and "known_ret" arguments. Change the type of the mode argument
17765 to scalar_int_mode.
17766 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17767 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17768 (reg_num_sign_bit_copies_for_combine): Likewise.
17769 * rtlanal.c (nonzero_bits1): Likewise.
17770 (num_sign_bit_copies1): Likewise.
17771 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17772 (reg_num_sign_bit_copies_general): Likewise.
17773 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17774 (reg_nonzero_bits_general): Likewise.
17776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17777 Alan Hayward <alan.hayward@arm.com>
17778 David Sherwood <david.sherwood@arm.com>
17780 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17781 than in subroutines. Return 1 for non-integer modes.
17782 (cached_num_sign_bit_copies): Change the type of the mode parameter
17783 to scalar_int_mode.
17784 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17785 classes. Handle CONST_INT_P first and then check whether X also
17786 has a scalar integer mode. Check the same thing for inner registers
17787 of a SUBREG and for values that are being extended or truncated.
17789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17790 Alan Hayward <alan.hayward@arm.com>
17791 David Sherwood <david.sherwood@arm.com>
17793 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17794 in subroutines. Return the mode mask for non-integer modes.
17795 (cached_nonzero_bits): Change the type of the mode parameter
17796 to scalar_int_mode.
17797 (nonzero_bits1): Likewise. Remove early exit for other mode
17798 classes. Handle CONST_INT_P first and then check whether X
17799 also has a scalar integer mode.
17801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17802 Alan Hayward <alan.hayward@arm.com>
17803 David Sherwood <david.sherwood@arm.com>
17805 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17806 Return a scalar_int_mode and assert that the size is greater than
17808 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17809 (op_by_pieces_d::op_by_pieces_d): Likewise.
17810 (op_by_pieces_d::run): Likewise.
17811 (can_store_by_pieces): Likewise.
17813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17814 Alan Hayward <alan.hayward@arm.com>
17815 David Sherwood <david.sherwood@arm.com>
17817 * combine.c (extract_left_shift): Add a mode argument and update
17819 (make_compound_operation_int): Change the type of the mode parameter
17820 to scalar_int_mode and update the call to extract_left_shift.
17822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17823 Alan Hayward <alan.hayward@arm.com>
17824 David Sherwood <david.sherwood@arm.com>
17826 * combine.c (simplify_and_const_int): Change the type of the mode
17827 parameter to scalar_int_mode.
17828 (simplify_and_const_int_1): Likewise. Update recursive call.
17830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17831 Alan Hayward <alan.hayward@arm.com>
17832 David Sherwood <david.sherwood@arm.com>
17834 * combine.c (simplify_compare_const): Check that the mode is a
17835 scalar_int_mode (rather than VOIDmode) before testing its
17837 (simplify_comparison): Move COMPARISON_P handling out of the
17838 loop and restrict the latter part of the loop to scalar_int_modes.
17839 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17840 and when considering SUBREG_REGs. Use is_int_mode instead of
17841 checking GET_MODE_CLASS against MODE_INT.
17843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17844 Alan Hayward <alan.hayward@arm.com>
17845 David Sherwood <david.sherwood@arm.com>
17847 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17848 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17849 shift_unit_mode and for modes involved in scalar shifts.
17851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17852 Alan Hayward <alan.hayward@arm.com>
17853 David Sherwood <david.sherwood@arm.com>
17855 * combine.c (force_int_to_mode): New function, split out from...
17856 (force_to_mode): ...here. Keep xmode up-to-date and use it
17857 instead of GET_MODE (x).
17859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17860 Alan Hayward <alan.hayward@arm.com>
17861 David Sherwood <david.sherwood@arm.com>
17863 * optabs-query.h (extraction_insn::struct_mode): Change type to
17864 opt_scalar_int_mode and update comment.
17865 (extraction_insn::field_mode): Change type to scalar_int_mode.
17866 (extraction_insn::pos_mode): Likewise.
17867 * combine.c (make_extraction): Update accordingly.
17868 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17869 (get_optab_extraction_insn): Likewise.
17870 * recog.c (simplify_while_replacing): Likewise.
17871 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17872 parameter to opt_scalar_int_mode.
17874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17875 Alan Hayward <alan.hayward@arm.com>
17876 David Sherwood <david.sherwood@arm.com>
17878 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17879 to a scalar_int_mode instead of a machine_mode.
17880 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17881 (get_best_mode): Return a boolean and use a pointer argument to store
17882 the selected mode. Replace the limit mode parameter with a bit limit.
17883 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17884 for the values returned by bit_field_mode_iterator::next_mode.
17885 (store_bit_field): Update call to get_best_mode.
17886 (store_fixed_bit_field): Likewise.
17887 (extract_fixed_bit_field): Likewise.
17888 * expr.c (optimize_bitfield_assignment_op): Likewise.
17889 * fold-const.c (optimize_bit_field_compare): Likewise.
17890 (fold_truth_andor_1): Likewise.
17891 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17892 Update for new type of m_mode.
17893 (get_best_mode): As above.
17895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17896 Alan Hayward <alan.hayward@arm.com>
17897 David Sherwood <david.sherwood@arm.com>
17899 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17900 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17901 (store_bit_field): Check is_a <scalar_int_mode> before calling
17902 strict_volatile_bitfield_p.
17903 (extract_bit_field): Likewise.
17905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17906 Alan Hayward <alan.hayward@arm.com>
17907 David Sherwood <david.sherwood@arm.com>
17909 * target.def (cstore_mode): Return a scalar_int_mode.
17910 * doc/tm.texi: Regenerate.
17911 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17912 * targhooks.h (default_cstore_mode): Likewise.
17913 * targhooks.c (default_cstore_mode): Likewise, using a forced
17915 * expmed.c (emit_cstore): Expect the target of the cstore to be
17918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17919 Alan Hayward <alan.hayward@arm.com>
17920 David Sherwood <david.sherwood@arm.com>
17922 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17924 (niter_desc): Likewise mode.
17925 (iv_analyze): Add a mode parameter.
17927 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17928 and change its type to scalar_int_mode.
17929 * loop-iv.c: Update commentary at head of file.
17930 (iv_constant): Pass the mode paraeter before the rtx it describes
17931 and change its type to scalar_int_mode. Remove VOIDmode handling.
17932 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17933 (iv_extend): Likewise.
17934 (shorten_into_mode): Likewise.
17935 (iv_add): Use scalar_int_mode.
17936 (iv_mult): Likewise.
17937 (iv_shift): Likewise.
17938 (canonicalize_iv_subregs): Likewise.
17939 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17940 it describes and change its mode to scalar_int_mode. Also change
17941 the type of the returned inner_mode to scalar_int_mode.
17942 (get_biv_step): Likewise, turning outer_mode from a pointer
17943 into a direct parameter. Update call to get_biv_step_1.
17944 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17945 iv_constant and get_biv_step.
17946 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17947 and change its type to scalar_int_mode. Don't initialise iv->mode
17948 to VOIDmode and remove later checks for its still being VOIDmode.
17949 Update calls to iv_analyze_op and iv_analyze_expr. Check
17950 is_a <scalar_int_mode> when changing the mode under consideration.
17951 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17952 Update call to iv_analyze_expr.
17953 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17954 inner register is not also a scalar_int_mode. Update call to
17956 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17957 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17958 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17959 separate mode class checks. Update calls to iv_analyze. Remove
17960 fix-up of VOIDmodes after iv_analyze_biv.
17961 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17962 don't have a scalar_int_mode. Update call to biv_p.
17964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17965 Alan Hayward <alan.hayward@arm.com>
17966 David Sherwood <david.sherwood@arm.com>
17968 * cfgexpand.c (convert_debug_memory_address): Use
17969 as_a <scalar_int_mode>.
17970 * combine.c (expand_compound_operation): Likewise.
17971 (make_extraction): Likewise.
17972 (change_zero_ext): Likewise.
17973 (simplify_comparison): Likewise.
17974 * cse.c (cse_insn): Likewise.
17975 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17976 (mem_loc_descriptor): Likewise.
17977 (loc_descriptor): Likewise.
17978 * expmed.c (init_expmed_one_mode): Likewise.
17979 (synth_mult): Likewise.
17980 (emit_store_flag_1): Likewise.
17981 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17982 of a comparison with size.
17983 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17984 (reduce_to_bit_field_precision): Likewise.
17985 * function.c (expand_function_end): Likewise.
17986 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17987 * loop-doloop.c (doloop_modify): Likewise.
17988 * optabs.c (expand_binop): Likewise.
17989 (expand_unop): Likewise.
17990 (expand_copysign_absneg): Likewise.
17991 (prepare_cmp_insn): Likewise.
17992 (maybe_legitimize_operand): Likewise.
17993 * recog.c (const_scalar_int_operand): Likewise.
17994 * rtlanal.c (get_address_mode): Likewise.
17995 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17996 (simplify_cond_clz_ctz): Likewise.
17997 * tree-nested.c (get_nl_goto_field): Likewise.
17998 * tree.c (build_vector_type_for_mode): Likewise.
17999 * var-tracking.c (use_narrower_mode): Likewise.
18001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18002 Alan Hayward <alan.hayward@arm.com>
18003 David Sherwood <david.sherwood@arm.com>
18005 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18006 * builtins.c (expand_builtin_signbit): Use it.
18007 * cfgexpand.c (expand_debug_expr): Likewise.
18008 * dojump.c (do_jump): Likewise.
18009 (do_compare_and_jump): Likewise.
18010 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18011 * expmed.c (make_tree): Likewise.
18012 * expr.c (expand_expr_real_2): Likewise.
18013 (expand_expr_real_1): Likewise.
18014 (try_casesi): Likewise.
18015 * fold-const-call.c (fold_const_call_ss): Likewise.
18016 * fold-const.c (unextend): Likewise.
18017 (extract_muldiv_1): Likewise.
18018 (fold_single_bit_test): Likewise.
18019 (native_encode_int): Likewise.
18020 (native_encode_string): Likewise.
18021 (native_interpret_int): Likewise.
18022 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18023 * internal-fn.c (expand_addsub_overflow): Likewise.
18024 (expand_neg_overflow): Likewise.
18025 (expand_mul_overflow): Likewise.
18026 (expand_arith_overflow): Likewise.
18027 * match.pd: Likewise.
18028 * stor-layout.c (layout_type): Likewise.
18029 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18030 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18031 * tree-ssanames.c (get_range_info): Likewise.
18032 * tree-switch-conversion.c (array_value_type) Likewise.
18033 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18034 (vect_recog_divmod_pattern): Likewise.
18035 (vect_recog_mixed_size_cond_pattern): Likewise.
18036 * tree-vrp.c (extract_range_basic): Likewise.
18037 (simplify_float_conversion_using_ranges): Likewise.
18038 * tree.c (int_fits_type_p): Likewise.
18039 * ubsan.c (instrument_bool_enum_load): Likewise.
18040 * varasm.c (mergeable_string_section): Likewise.
18041 (narrowing_initializer_constant_valid_p): Likewise.
18042 (output_constant): Likewise.
18044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18045 Alan Hayward <alan.hayward@arm.com>
18046 David Sherwood <david.sherwood@arm.com>
18048 * machmode.h (NARROWEST_INT_MODE): New macro.
18049 * expr.c (alignment_for_piecewise_move): Use it instead of
18050 GET_CLASS_NARROWEST_MODE (MODE_INT).
18051 (push_block): Likewise.
18052 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18054 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18057 Alan Hayward <alan.hayward@arm.com>
18058 David Sherwood <david.sherwood@arm.com>
18060 * postreload.c (move2add_valid_value_p): Change the type of the
18061 mode parameter to scalar_int_mode.
18062 (move2add_use_add2_insn): Add a mode parameter and use it instead
18064 (move2add_use_add3_insn): Likewise.
18065 (reload_cse_move2add): Update accordingly.
18067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18068 Alan Hayward <alan.hayward@arm.com>
18069 David Sherwood <david.sherwood@arm.com>
18071 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18073 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18074 * optabs.c (expand_unop): Likewise.
18076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18077 Alan Hayward <alan.hayward@arm.com>
18078 David Sherwood <david.sherwood@arm.com>
18080 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18081 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18082 (popcount_loc_descriptor): Likewise.
18083 (bswap_loc_descriptor): Likewise.
18084 (rotate_loc_descriptor): Likewise.
18085 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18086 calling the functions above.
18088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18089 Alan Hayward <alan.hayward@arm.com>
18090 David Sherwood <david.sherwood@arm.com>
18092 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18094 (try_combine): Likewise.
18095 (simplify_if_then_else): Likewise.
18096 * cse.c (cse_insn): Likewise.
18097 * dwarf2out.c (mem_loc_descriptor): Likewise.
18098 * emit-rtl.c (gen_lowpart_common): Likewise.
18099 * simplify-rtx.c (simplify_truncation): Likewise.
18100 (simplify_binary_operation_1): Likewise.
18101 (simplify_const_relational_operation): Likewise.
18102 (simplify_ternary_operation): Likewise.
18103 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18106 Alan Hayward <alan.hayward@arm.com>
18107 David Sherwood <david.sherwood@arm.com>
18109 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18110 * reload.c (push_reload): Likewise.
18111 (find_reloads): Likewise.
18113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18114 Alan Hayward <alan.hayward@arm.com>
18115 David Sherwood <david.sherwood@arm.com>
18117 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18118 (make_compound_operation_int): Likewise.
18119 (change_zero_ext): Likewise.
18120 * expr.c (convert_move): Likewise.
18121 (convert_modes): Likewise.
18122 * fwprop.c (forward_propagate_subreg): Likewise.
18123 * loop-iv.c (get_biv_step_1): Likewise.
18124 * optabs.c (widen_operand): Likewise.
18125 * postreload.c (move2add_valid_value_p): Likewise.
18126 * recog.c (simplify_while_replacing): Likewise.
18127 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18128 (simplify_binary_operation_1): Likewise. Remove redundant
18129 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
18136 is_a <scalar_int_mode>.
18137 (simplify_if_then_else): Likewise.
18138 (make_field_assignment): Likewise.
18139 (simplify_comparison): Likewise.
18140 * ifcvt.c (noce_try_bitop): Likewise.
18141 * loop-invariant.c (canonicalize_address_mult): Likewise.
18142 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18145 Alan Hayward <alan.hayward@arm.com>
18146 David Sherwood <david.sherwood@arm.com>
18148 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18149 is_a <scalar_int_mode> instead of != BLKmode.
18151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18152 Alan Hayward <alan.hayward@arm.com>
18153 David Sherwood <david.sherwood@arm.com>
18155 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18156 instead of != VOIDmode.
18157 * combine.c (if_then_else_cond): Likewise.
18158 (change_zero_ext): Likewise.
18159 * dwarf2out.c (mem_loc_descriptor): Likewise.
18160 (loc_descriptor): Likewise.
18161 * rtlanal.c (canonicalize_condition): Likewise.
18162 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18165 Alan Hayward <alan.hayward@arm.com>
18166 David Sherwood <david.sherwood@arm.com>
18168 * simplify-rtx.c (simplify_binary_operation_1): Use
18169 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18172 Alan Hayward <alan.hayward@arm.com>
18173 David Sherwood <david.sherwood@arm.com>
18175 * wide-int.h (int_traits<unsigned char>) New class.
18176 (int_traits<unsigned short>) Likewise.
18177 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18178 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18180 * combine.c (set_nonzero_bits_and_sign_copies): Use
18181 is_a <scalar_int_mode>.
18182 (find_split_point): Likewise.
18183 (combine_simplify_rtx): Likewise.
18184 (simplify_logical): Likewise.
18185 (expand_compound_operation): Likewise.
18186 (expand_field_assignment): Likewise.
18187 (make_compound_operation): Likewise.
18188 (extended_count): Likewise.
18189 (change_zero_ext): Likewise.
18190 (simplify_comparison): Likewise.
18191 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18192 (ucompare_loc_descriptor): Likewise.
18193 (minmax_loc_descriptor): Likewise.
18194 (mem_loc_descriptor): Likewise.
18195 (loc_descriptor): Likewise.
18196 * expmed.c (init_expmed_one_mode): Likewise.
18197 * lra-constraints.c (lra_constraint_offset): Likewise.
18198 * optabs.c (prepare_libcall_arg): Likewise.
18199 * postreload.c (move2add_note_store): Likewise.
18200 * reload.c (operands_match_p): Likewise.
18201 * rtl.h (load_extend_op): Likewise.
18202 * rtlhooks.c (gen_lowpart_general): Likewise.
18203 * simplify-rtx.c (simplify_truncation): Likewise.
18204 (simplify_unary_operation_1): Likewise.
18205 (simplify_binary_operation_1): Likewise.
18206 (simplify_const_binary_operation): Likewise.
18207 (simplify_const_relational_operation): Likewise.
18208 (simplify_subreg): Likewise.
18209 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18210 * var-tracking.c (adjust_mems): Likewise.
18211 (prepare_call_arguments): Likewise.
18213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18214 Alan Hayward <alan.hayward@arm.com>
18215 David Sherwood <david.sherwood@arm.com>
18217 * machmode.h (is_int_mode): New fuction.
18218 * combine.c (find_split_point): Use it.
18219 (combine_simplify_rtx): Likewise.
18220 (simplify_if_then_else): Likewise.
18221 (simplify_set): Likewise.
18222 (simplify_shift_const_1): Likewise.
18223 (simplify_comparison): Likewise.
18224 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18225 * cse.c (notreg_cost): Likewise.
18226 (cse_insn): Likewise.
18227 * cselib.c (cselib_lookup_1): Likewise.
18228 * dojump.c (do_jump_1): Likewise.
18229 (do_compare_rtx_and_jump): Likewise.
18230 * dse.c (get_call_args): Likewise.
18231 * dwarf2out.c (rtl_for_decl_init): Likewise.
18232 (native_encode_initializer): Likewise.
18233 * expmed.c (emit_store_flag_1): Likewise.
18234 (emit_store_flag): Likewise.
18235 * expr.c (convert_modes): Likewise.
18236 (store_field): Likewise.
18237 (expand_expr_real_1): Likewise.
18238 * fold-const.c (fold_read_from_constant_string): Likewise.
18239 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18240 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18241 * optabs.c (expand_binop): Likewise.
18242 (expand_unop): Likewise.
18243 (expand_abs_nojump): Likewise.
18244 (expand_one_cmpl_abs_nojump): Likewise.
18245 * simplify-rtx.c (mode_signbit_p): Likewise.
18246 (val_signbit_p): Likewise.
18247 (val_signbit_known_set_p): Likewise.
18248 (val_signbit_known_clear_p): Likewise.
18249 (simplify_relational_operation_1): Likewise.
18250 * tree.c (vector_type_mode): Likewise.
18252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18253 Alan Hayward <alan.hayward@arm.com>
18254 David Sherwood <david.sherwood@arm.com>
18256 * machmode.h (smallest_mode_for_size): Fix formatting.
18257 (smallest_int_mode_for_size): New function.
18258 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18259 instead of smallest_mode_for_size.
18260 * combine.c (make_extraction): Likewise.
18261 * config/arc/arc.c (arc_expand_movmem): Likewise.
18262 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18263 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18264 * config/s390/s390.c (s390_expand_insv): Likewise.
18265 * config/sparc/sparc.c (assign_int_registers): Likewise.
18266 * config/spu/spu.c (spu_function_value): Likewise.
18267 (spu_function_arg): Likewise.
18268 * coverage.c (get_gcov_type): Likewise.
18269 (get_gcov_unsigned_t): Likewise.
18270 * dse.c (find_shift_sequence): Likewise.
18271 * expmed.c (store_bit_field_1): Likewise.
18272 * expr.c (convert_move): Likewise.
18273 (store_field): Likewise.
18274 * internal-fn.c (expand_arith_overflow): Likewise.
18275 * optabs-query.c (get_best_extraction_insn): Likewise.
18276 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18277 * stor-layout.c (layout_type): Likewise.
18278 (initialize_sizetypes): Likewise.
18279 * targhooks.c (default_get_mask_mode): Likewise.
18280 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18283 Alan Hayward <alan.hayward@arm.com>
18284 David Sherwood <david.sherwood@arm.com>
18286 * machmode.h (opt_mode::else_blk): New function.
18287 (int_mode_for_mode): Declare.
18288 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18289 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18291 * cfgexpand.c (expand_debug_expr): Likewise.
18292 * combine.c (gen_lowpart_or_truncate): Likewise.
18293 (gen_lowpart_for_combine): Likewise.
18294 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18295 * config/avr/avr.c (avr_to_int_mode): Likewise.
18296 (avr_out_plus_1): Likewise.
18297 (avr_out_plus): Likewise.
18298 (avr_out_round): Likewise.
18299 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18300 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18301 (s390_expand_vcond): Likewise.
18302 * config/spu/spu.c (spu_split_immediate): Likewise.
18303 (spu_expand_mov): Likewise.
18304 * dse.c (get_stored_val): Likewise.
18305 * expmed.c (store_bit_field_1): Likewise.
18306 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18308 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18309 (extract_low_bits): Likewise.
18310 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18311 handling rather than repeating the check.
18312 (emit_group_store): Likewise.
18313 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18314 * optabs.c (expand_absneg_bit): Likewise.
18315 (expand_copysign_absneg): Likewise.
18316 (expand_copysign_bit): Likewise.
18317 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18318 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18319 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18320 * var-tracking.c (prepare_call_arguments): Likewise.
18321 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18322 int_mode_for_mode instead of mode_for_size.
18323 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18326 Alan Hayward <alan.hayward@arm.com>
18327 David Sherwood <david.sherwood@arm.com>
18329 * machmode.h (int_mode_for_size): New function.
18330 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18331 instead of mode_for_size.
18332 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18334 * combine.c (expand_field_assignment): Use int_mode_for_size
18335 instead of mode_for_size.
18336 (make_extraction): Likewise.
18337 (simplify_shift_const_1): Likewise.
18338 (simplify_comparison): Likewise.
18339 * dojump.c (do_jump): Likewise.
18340 * dwarf2out.c (mem_loc_descriptor): Likewise.
18341 * emit-rtl.c (init_derived_machine_modes): Likewise.
18342 * expmed.c (flip_storage_order): Likewise.
18343 (convert_extracted_bit_field): Likewise.
18344 * expr.c (copy_blkmode_from_reg): Likewise.
18345 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18346 * internal-fn.c (expand_mul_overflow): Likewise.
18347 * lower-subreg.c (simple_move): Likewise.
18348 * optabs-libfuncs.c (init_optabs): Likewise.
18349 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18350 * tree.c (vector_type_mode): Likewise.
18351 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18352 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18353 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18354 * tree-vect-stmts.c (vectorizable_load): Likewise.
18355 (vectorizable_store): Likewise.
18357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18358 Alan Hayward <alan.hayward@arm.com>
18359 David Sherwood <david.sherwood@arm.com>
18361 * coretypes.h (pod_mode): New type.
18362 (scalar_int_mode_pod): New typedef.
18363 * machmode.h (pod_mode): New class.
18364 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18365 * genmodes.c (emit_mode_int_n): Update accordingly.
18366 * lower-subreg.h (target_lower_subreg): Change type to
18367 scalar_int_mode_pod.
18368 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18369 scalar_int_mode_pod.
18371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18372 Alan Hayward <alan.hayward@arm.com>
18373 David Sherwood <david.sherwood@arm.com>
18375 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18376 machine_mode to scalar_int_mode.
18377 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18378 (rs6000_option_override_internal): Remove cast to int.
18379 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18380 machine_mode to scalar_int_mode.
18381 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18382 (rs6000_option_override_internal): Remove cast to int.
18383 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18384 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18386 * config/s390/s390.c (s390_expand_builtin): Likewise.
18387 * coretypes.h (scalar_int_mode): New type.
18388 (opt_scalar_int_mode): New typedef.
18389 * machmode.h (scalar_int_mode): New class.
18390 (scalar_int_mode::includes_p): New function.
18391 (byte_mode): Change type to scalar_int_mode.
18392 (word_mode): Likewise.
18393 (ptr_mode): Likewise.
18394 * emit-rtl.c (byte_mode): Likewise.
18395 (word_mode): Likewise.
18396 (ptr_mode): Likewise.
18397 (init_derived_machine_modes): Update accordingly.
18398 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18399 and MODE_PARTIAL_INT.
18400 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18401 opt_scalar_int_mode.
18403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18404 Alan Hayward <alan.hayward@arm.com>
18405 David Sherwood <david.sherwood@arm.com>
18407 * target.def (libgcc_floating_mode_supported_p): Take a
18409 * doc/tm.texi: Regenerate.
18410 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18412 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18413 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18417 Alan Hayward <alan.hayward@arm.com>
18418 David Sherwood <david.sherwood@arm.com>
18420 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18421 * doc/tm.texi: Regenerate.
18422 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18423 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18424 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18425 * targhooks.h (default_floatn_mode): Likewise.
18426 * targhooks.c (default_floatn_mode): Likewise.
18427 * tree.c (build_common_tree_nodes): Update accordingly.
18429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18430 Alan Hayward <alan.hayward@arm.com>
18431 David Sherwood <david.sherwood@arm.com>
18433 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18434 (mode_iterator::iterate_p): Likewise.
18435 (mode_iterator::get_wider): Likewise.
18436 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18439 Alan Hayward <alan.hayward@arm.com>
18440 David Sherwood <david.sherwood@arm.com>
18442 * coretypes.h (opt_scalar_float_mode): New typedef.
18443 * machmode.h (float_mode_for_size): New function.
18444 * emit-rtl.c (double_mode): Delete.
18445 (init_emit_once): Use float_mode_for_size.
18446 * stor-layout.c (layout_type): Likewise.
18447 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18450 Alan Hayward <alan.hayward@arm.com>
18451 David Sherwood <david.sherwood@arm.com>
18453 * output.h (assemble_real): Take a scalar_float_mode.
18454 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18455 * config/arm/arm.md (consttable_4): Likewise.
18456 (consttable_8): Likewise.
18457 (consttable_16): Likewise.
18458 * config/mips/mips.md (consttable_float): Likewise.
18459 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18460 * varasm.c (assemble_real): Take a scalar_float_mode.
18461 (output_constant_pool_2): Update accordingly.
18462 (output_constant): Likewise.
18464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18465 Alan Hayward <alan.hayward@arm.com>
18466 David Sherwood <david.sherwood@arm.com>
18468 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18469 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18470 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18471 (native_encode_real): Likewise.
18472 (native_interpret_real): Likewise.
18473 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18474 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18477 Alan Hayward <alan.hayward@arm.com>
18478 David Sherwood <david.sherwood@arm.com>
18480 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18481 <scalar_float_mode>. Simplify.
18482 (gen_extend_conv_libfunc): Likewise.
18484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18485 Alan Hayward <alan.hayward@arm.com>
18486 David Sherwood <david.sherwood@arm.com>
18488 * coretypes.h (scalar_float_mode): New type.
18489 * machmode.h (mode_traits::from_int): Use machine_mode if
18490 USE_ENUM_MODES is defined.
18491 (is_a): New function.
18493 (dyn_cast): Likewise.
18494 (scalar_float_mode): New class.
18495 (scalar_float_mode::includes_p): New function.
18496 (is_float_mode): Likewise.
18497 * gdbhooks.py (MachineModePrinter): New class.
18498 (build_pretty_printer): Use it for scalar_float_mode.
18499 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18500 (format_helper::format_helper): Turn into a template.
18501 * genmodes.c (get_mode_class): New function.
18502 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18503 or machine_mode if none.
18504 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18505 as_a <scalar_float_mode>.
18506 * dwarf2out.c (mem_loc_descriptor): Likewise.
18507 (insert_float): Likewise.
18508 (add_const_value_attribute): Likewise.
18509 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18510 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18511 (expand_unop): Update accordingly.
18512 (expand_abs_nojump): Likewise.
18513 (expand_copysign_absneg): Take a scalar_float_mode.
18514 (expand_copysign_bit): Likewise.
18515 (expand_copysign): Update accordingly.
18517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18518 Alan Hayward <alan.hayward@arm.com>
18519 David Sherwood <david.sherwood@arm.com>
18521 * coretypes.h (opt_mode): New class.
18522 * machmode.h (opt_mode): Likewise.
18523 (opt_mode::else_void): New function.
18524 (opt_mode::require): Likewise.
18525 (opt_mode::exists): Likewise.
18526 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18527 (GET_MODE_2XWIDER_MODE): Likewise.
18528 (mode_iterator::get_wider): Update accordingly.
18529 (mode_iterator::get_2xwider): Likewise.
18530 (mode_iterator::get_known_wider): Likewise, turning into a template.
18531 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18532 forcing a wider mode to exist.
18533 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18534 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18535 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18536 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18537 * lower-subreg.c (init_lower_subreg): Likewise.
18538 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18539 on the final iteration.
18540 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18541 a wider mode exists before asking for a move pattern.
18542 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18543 forcing a wider mode to exist.
18544 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18545 returning false if no such mode exists.
18546 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18547 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18548 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18549 Avoid checking for a MODE_INT if we already know the mode is not a
18551 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18552 forcing a wider mode to exist.
18553 (expmed_mult_highpart_optab): Likewise.
18554 (expmed_mult_highpart): Likewise.
18555 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18557 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18558 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18559 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18560 * internal-fn.c (expand_mul_overflow): Update use of
18561 GET_MODE_2XWIDER_MODE.
18562 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18563 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18564 GET_MODE_WIDER_MODE.
18565 (convert_plusminus_to_widen): Likewise.
18566 * tree-switch-conversion.c (array_value_type): Likewise.
18567 * var-tracking.c (emit_note_insn_var_location): Likewise.
18568 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18569 Return false inside rather than outside the loop if no wider mode
18571 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18572 and GET_MODE_2XWIDER_MODE
18573 (can_compare_p): Use else_void.
18574 * gdbhooks.py (OptMachineModePrinter): New class.
18575 (build_pretty_printer): Use it for opt_mode.
18577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18578 Alan Hayward <alan.hayward@arm.com>
18579 David Sherwood <david.sherwood@arm.com>
18581 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18582 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18585 Alan Hayward <alan.hayward@arm.com>
18586 David Sherwood <david.sherwood@arm.com>
18588 * machmode.h (mode_traits): New structure.
18589 (get_narrowest_mode): New function.
18590 (mode_iterator::start): Likewise.
18591 (mode_iterator::iterate_p): Likewise.
18592 (mode_iterator::get_wider): Likewise.
18593 (mode_iterator::get_known_wider): Likewise.
18594 (mode_iterator::get_2xwider): Likewise.
18595 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18596 (FOR_EACH_MODE): Likewise.
18597 (FOR_EACH_MODE_FROM): Likewise.
18598 (FOR_EACH_MODE_UNTIL): Likewise.
18599 (FOR_EACH_WIDER_MODE): Likewise.
18600 (FOR_EACH_2XWIDER_MODE): Likewise.
18601 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18602 * combine.c (simplify_comparison): Likewise
18603 * config/i386/i386.c (type_natural_mode): Likewise.
18604 * cse.c (cse_insn): Likewise.
18605 * dse.c (find_shift_sequence): Likewise.
18606 * emit-rtl.c (init_derived_machine_modes): Likewise.
18607 (init_emit_once): Likewise.
18608 * explow.c (hard_function_value): Likewise.
18609 * expmed.c (extract_fixed_bit_field_1): Likewise.
18610 (extract_bit_field_1): Likewise.
18611 (expand_divmod): Likewise.
18612 (emit_store_flag_1): Likewise.
18613 * expr.c (init_expr_target): Likewise.
18614 (convert_move): Likewise.
18615 (alignment_for_piecewise_move): Likewise.
18616 (widest_int_mode_for_size): Likewise.
18617 (emit_block_move_via_movmem): Likewise.
18618 (copy_blkmode_to_reg): Likewise.
18619 (set_storage_via_setmem): Likewise.
18620 (compress_float_constant): Likewise.
18621 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18622 * optabs-query.c (get_best_extraction_insn): Likewise.
18623 * optabs.c (expand_binop): Likewise.
18624 (expand_twoval_unop): Likewise.
18625 (expand_twoval_binop): Likewise.
18626 (widen_leading): Likewise.
18627 (widen_bswap): Likewise.
18628 (expand_parity): Likewise.
18629 (expand_unop): Likewise.
18630 (prepare_cmp_insn): Likewise.
18631 (prepare_float_lib_cmp): Likewise.
18632 (expand_float): Likewise.
18633 (expand_fix): Likewise.
18634 (expand_sfix_optab): Likewise.
18635 * postreload.c (move2add_use_add2_insn): Likewise.
18636 * reg-stack.c (reg_to_stack): Likewise.
18637 * reginfo.c (choose_hard_reg_mode): Likewise.
18638 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18639 * stor-layout.c (mode_for_size): Likewise.
18640 (smallest_mode_for_size): Likewise.
18641 (mode_for_vector): Likewise.
18642 (finish_bitfield_representative): Likewise.
18643 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18644 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18645 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18646 * var-tracking.c (prepare_call_arguments): Likewise.
18648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18649 Alan Hayward <alan.hayward@arm.com>
18650 David Sherwood <david.sherwood@arm.com>
18652 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18653 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18654 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18655 * machmode.h (mode_size): Move earlier in file.
18656 (mode_precision): Likewise.
18657 (mode_inner): Likewise.
18658 (mode_nunits): Likewise.
18659 (mode_unit_size): Likewise.
18660 (unit_unit_precision): Likewise.
18661 (mode_wider): Likewise.
18662 (mode_2xwider): Likewise.
18663 (machine_mode): New class.
18664 (mode_to_bytes): New function.
18665 (mode_to_bits): Likewise.
18666 (mode_to_precision): Likewise.
18667 (mode_to_inner): Likewise.
18668 (mode_to_unit_size): Likewise.
18669 (mode_to_unit_precision): Likewise.
18670 (mode_to_nunits): Likewise.
18671 (GET_MODE_SIZE): Use mode_to_bytes.
18672 (GET_MODE_BITSIZE): Use mode_to_bits.
18673 (GET_MODE_PRECISION): Use mode_to_precision.
18674 (GET_MODE_INNER): Use mode_to_inner.
18675 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18676 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18677 (GET_MODE_NUNITS): Use mode_to_nunits.
18678 * system.h (ALWAYS_INLINE): New macro.
18679 * config/powerpcspe/powerpcspe-c.c
18680 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18681 int for arg1_mode and arg2_mode.
18683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18684 Alan Hayward <alan.hayward@arm.com>
18685 David Sherwood <david.sherwood@arm.com>
18687 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18688 Prefix mode names with E_ in case statements.
18689 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18690 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18691 (aarch64_split_simd_move): Likewise.
18692 (aarch64_gen_storewb_pair): Likewise.
18693 (aarch64_gen_loadwb_pair): Likewise.
18694 (aarch64_gen_store_pair): Likewise.
18695 (aarch64_gen_load_pair): Likewise.
18696 (aarch64_get_condition_code_1): Likewise.
18697 (aarch64_constant_pool_reload_icode): Likewise.
18698 (get_rsqrte_type): Likewise.
18699 (get_rsqrts_type): Likewise.
18700 (get_recpe_type): Likewise.
18701 (get_recps_type): Likewise.
18702 (aarch64_gimplify_va_arg_expr): Likewise.
18703 (aarch64_simd_container_mode): Likewise.
18704 (aarch64_emit_load_exclusive): Likewise.
18705 (aarch64_emit_store_exclusive): Likewise.
18706 (aarch64_expand_compare_and_swap): Likewise.
18707 (aarch64_gen_atomic_cas): Likewise.
18708 (aarch64_emit_bic): Likewise.
18709 (aarch64_emit_atomic_swap): Likewise.
18710 (aarch64_emit_atomic_load_op): Likewise.
18711 (aarch64_evpc_trn): Likewise.
18712 (aarch64_evpc_uzp): Likewise.
18713 (aarch64_evpc_zip): Likewise.
18714 (aarch64_evpc_ext): Likewise.
18715 (aarch64_evpc_rev): Likewise.
18716 (aarch64_evpc_dup): Likewise.
18717 (aarch64_gen_ccmp_first): Likewise.
18718 (aarch64_gen_ccmp_next): Likewise.
18719 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18720 (alpha_emit_xfloating_libcall): Likewise.
18721 (emit_insxl): Likewise.
18722 (alpha_arg_type): Likewise.
18723 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18724 (arc_preferred_simd_mode): Likewise.
18725 (arc_secondary_reload): Likewise.
18726 (get_arc_condition_code): Likewise.
18727 (arc_print_operand): Likewise.
18728 (arc_legitimate_constant_p): Likewise.
18729 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18730 * config/arc/arc.md (casesi_load): Likewise.
18731 (casesi_compact_jump): Likewise.
18732 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18733 (cc_use_register): Likewise.
18734 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18735 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18736 (arm_init_iwmmxt_builtins): Likewise.
18737 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18738 (neon_expand_vector_init): Likewise.
18739 (arm_attr_length_move_neon): Likewise.
18740 (maybe_get_arm_condition_code): Likewise.
18741 (arm_emit_vector_const): Likewise.
18742 (arm_preferred_simd_mode): Likewise.
18743 (arm_output_iwmmxt_tinsr): Likewise.
18744 (thumb1_output_casesi): Likewise.
18745 (thumb2_output_casesi): Likewise.
18746 (arm_emit_load_exclusive): Likewise.
18747 (arm_emit_store_exclusive): Likewise.
18748 (arm_expand_compare_and_swap): Likewise.
18749 (arm_evpc_neon_vuzp): Likewise.
18750 (arm_evpc_neon_vzip): Likewise.
18751 (arm_evpc_neon_vrev): Likewise.
18752 (arm_evpc_neon_vtrn): Likewise.
18753 (arm_evpc_neon_vext): Likewise.
18754 (arm_validize_comparison): Likewise.
18755 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18756 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18757 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18758 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18759 (c6x_preferred_simd_mode): Likewise.
18760 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18761 (epiphany_rtx_costs): Likewise.
18762 * config/epiphany/predicates.md (proper_comparison_operator):
18764 * config/frv/frv.c (condexec_memory_operand): Likewise.
18765 (frv_emit_move): Likewise.
18766 (output_move_single): Likewise.
18767 (output_condmove_single): Likewise.
18768 (frv_hard_regno_mode_ok): Likewise.
18769 (frv_matching_accg_mode): Likewise.
18770 * config/h8300/h8300.c (split_adds_subs): Likewise.
18771 (h8300_rtx_costs): Likewise.
18772 (h8300_print_operand): Likewise.
18773 (compute_mov_length): Likewise.
18774 (output_logical_op): Likewise.
18775 (compute_logical_op_length): Likewise.
18776 (compute_logical_op_cc): Likewise.
18777 (h8300_shift_needs_scratch_p): Likewise.
18778 (output_a_shift): Likewise.
18779 (compute_a_shift_length): Likewise.
18780 (compute_a_shift_cc): Likewise.
18781 (expand_a_rotate): Likewise.
18782 (output_a_rotate): Likewise.
18783 * config/i386/i386.c (classify_argument): Likewise.
18784 (function_arg_advance_32): Likewise.
18785 (function_arg_32): Likewise.
18786 (function_arg_64): Likewise.
18787 (function_value_64): Likewise.
18788 (ix86_gimplify_va_arg): Likewise.
18789 (ix86_legitimate_constant_p): Likewise.
18790 (put_condition_code): Likewise.
18791 (split_double_mode): Likewise.
18792 (ix86_avx256_split_vector_move_misalign): Likewise.
18793 (ix86_expand_vector_logical_operator): Likewise.
18794 (ix86_split_idivmod): Likewise.
18795 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18796 (ix86_build_const_vector): Likewise.
18797 (ix86_build_signbit_mask): Likewise.
18798 (ix86_match_ccmode): Likewise.
18799 (ix86_cc_modes_compatible): Likewise.
18800 (ix86_expand_branch): Likewise.
18801 (ix86_expand_sse_cmp): Likewise.
18802 (ix86_expand_sse_movcc): Likewise.
18803 (ix86_expand_int_sse_cmp): Likewise.
18804 (ix86_expand_vec_perm_vpermi2): Likewise.
18805 (ix86_expand_vec_perm): Likewise.
18806 (ix86_expand_sse_unpack): Likewise.
18807 (ix86_expand_int_addcc): Likewise.
18808 (ix86_split_to_parts): Likewise.
18809 (ix86_vectorize_builtin_gather): Likewise.
18810 (ix86_vectorize_builtin_scatter): Likewise.
18811 (avx_vpermilp_parallel): Likewise.
18812 (inline_memory_move_cost): Likewise.
18813 (ix86_tieable_integer_mode_p): Likewise.
18814 (x86_maybe_negate_const_int): Likewise.
18815 (ix86_expand_vector_init_duplicate): Likewise.
18816 (ix86_expand_vector_init_one_nonzero): Likewise.
18817 (ix86_expand_vector_init_one_var): Likewise.
18818 (ix86_expand_vector_init_concat): Likewise.
18819 (ix86_expand_vector_init_interleave): Likewise.
18820 (ix86_expand_vector_init_general): Likewise.
18821 (ix86_expand_vector_set): Likewise.
18822 (ix86_expand_vector_extract): Likewise.
18823 (emit_reduc_half): Likewise.
18824 (ix86_emit_i387_round): Likewise.
18825 (ix86_mangle_type): Likewise.
18826 (ix86_expand_round_sse4): Likewise.
18827 (expand_vec_perm_blend): Likewise.
18828 (canonicalize_vector_int_perm): Likewise.
18829 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18830 (expand_vec_perm_1): Likewise.
18831 (expand_vec_perm_interleave3): Likewise.
18832 (expand_vec_perm_even_odd_pack): Likewise.
18833 (expand_vec_perm_even_odd_1): Likewise.
18834 (expand_vec_perm_broadcast_1): Likewise.
18835 (ix86_vectorize_vec_perm_const_ok): Likewise.
18836 (ix86_expand_vecop_qihi): Likewise.
18837 (ix86_expand_mul_widen_hilo): Likewise.
18838 (ix86_expand_sse2_abs): Likewise.
18839 (ix86_expand_pextr): Likewise.
18840 (ix86_expand_pinsr): Likewise.
18841 (ix86_preferred_simd_mode): Likewise.
18842 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18843 * config/i386/sse.md (*andnot<mode>3): Likewise.
18844 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18845 (*<code><mode>3): Likewise.
18846 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18847 (ia64_expand_atomic_op): Likewise.
18848 (ia64_arg_type): Likewise.
18849 (ia64_mode_to_int): Likewise.
18850 (ia64_scalar_mode_supported_p): Likewise.
18851 (ia64_vector_mode_supported_p): Likewise.
18852 (expand_vec_perm_broadcast): Likewise.
18853 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18854 (iq2000_function_arg_advance): Likewise.
18855 (iq2000_function_arg): Likewise.
18856 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18857 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18858 (m68k_libcall_value): Likewise.
18859 (m68k_function_value): Likewise.
18860 (sched_attr_op_type): Likewise.
18861 * config/mcore/mcore.c (mcore_output_move): Likewise.
18862 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18864 (microblaze_function_arg): Likewise.
18865 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18866 (mips_print_operand): Likewise.
18867 (mips_mode_ok_for_mov_fmt_p): Likewise.
18868 (mips_vector_mode_supported_p): Likewise.
18869 (mips_preferred_simd_mode): Likewise.
18870 (mips_expand_vpc_loongson_even_odd): Likewise.
18871 (mips_expand_vec_unpack): Likewise.
18872 (mips_expand_vi_broadcast): Likewise.
18873 (mips_expand_vector_init): Likewise.
18874 (mips_expand_vec_reduc): Likewise.
18875 (mips_expand_msa_cmp): Likewise.
18876 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18877 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18878 (cc_flags_for_mode): Likewise.
18879 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18880 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18881 (nds32_output_casesi_pc_relative): Likewise.
18882 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18883 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18884 (nvptx_gen_unpack): Likewise.
18885 (nvptx_gen_pack): Likewise.
18886 (nvptx_gen_shuffle): Likewise.
18887 (nvptx_gen_wcast): Likewise.
18888 (nvptx_preferred_simd_mode): Likewise.
18889 * config/pa/pa.c (pa_secondary_reload): Likewise.
18890 * config/pa/predicates.md (base14_operand): Likewise.
18891 * config/powerpcspe/powerpcspe-c.c
18892 (altivec_resolve_overloaded_builtin): Likewise.
18893 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18895 (rs6000_preferred_simd_mode): Likewise.
18896 (output_vec_const_move): Likewise.
18897 (rs6000_expand_vector_extract): Likewise.
18898 (rs6000_split_vec_extract_var): Likewise.
18899 (reg_offset_addressing_ok_p): Likewise.
18900 (rs6000_legitimate_offset_address_p): Likewise.
18901 (rs6000_legitimize_address): Likewise.
18902 (rs6000_emit_set_const): Likewise.
18903 (rs6000_const_vec): Likewise.
18904 (rs6000_emit_move): Likewise.
18905 (spe_build_register_parallel): Likewise.
18906 (rs6000_darwin64_record_arg_recurse): Likewise.
18907 (swap_selector_for_mode): Likewise.
18908 (spe_init_builtins): Likewise.
18909 (paired_init_builtins): Likewise.
18910 (altivec_init_builtins): Likewise.
18911 (do_load_for_compare): Likewise.
18912 (rs6000_generate_compare): Likewise.
18913 (rs6000_expand_float128_convert): Likewise.
18914 (emit_load_locked): Likewise.
18915 (emit_store_conditional): Likewise.
18916 (rs6000_output_function_epilogue): Likewise.
18917 (rs6000_handle_altivec_attribute): Likewise.
18918 (rs6000_function_value): Likewise.
18919 (emit_fusion_gpr_load): Likewise.
18920 (emit_fusion_p9_load): Likewise.
18921 (emit_fusion_p9_store): Likewise.
18922 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18923 (fusion_gpr_mem_load): Likewise.
18924 (fusion_addis_mem_combo_load): Likewise.
18925 (fusion_addis_mem_combo_store): Likewise.
18926 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18927 (fusion_gpr_mem_load): Likewise.
18928 (fusion_addis_mem_combo_load): Likewise.
18929 (fusion_addis_mem_combo_store): Likewise.
18930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18932 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18933 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18934 (rs6000_preferred_simd_mode): Likewise.
18935 (output_vec_const_move): Likewise.
18936 (rs6000_expand_vector_extract): Likewise.
18937 (rs6000_split_vec_extract_var): Likewise.
18938 (reg_offset_addressing_ok_p): Likewise.
18939 (rs6000_legitimate_offset_address_p): Likewise.
18940 (rs6000_legitimize_address): Likewise.
18941 (rs6000_emit_set_const): Likewise.
18942 (rs6000_const_vec): Likewise.
18943 (rs6000_emit_move): Likewise.
18944 (rs6000_darwin64_record_arg_recurse): Likewise.
18945 (swap_selector_for_mode): Likewise.
18946 (paired_init_builtins): Likewise.
18947 (altivec_init_builtins): Likewise.
18948 (rs6000_expand_float128_convert): Likewise.
18949 (emit_load_locked): Likewise.
18950 (emit_store_conditional): Likewise.
18951 (rs6000_output_function_epilogue): Likewise.
18952 (rs6000_handle_altivec_attribute): Likewise.
18953 (rs6000_function_value): Likewise.
18954 (emit_fusion_gpr_load): Likewise.
18955 (emit_fusion_p9_load): Likewise.
18956 (emit_fusion_p9_store): Likewise.
18957 * config/rx/rx.c (rx_gen_move_template): Likewise.
18958 (flags_from_mode): Likewise.
18959 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18960 (s390_slb_comparison): Likewise.
18961 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18962 (s390_vector_mode_supported_p): Likewise.
18963 (s390_cc_modes_compatible): Likewise.
18964 (s390_match_ccmode_set): Likewise.
18965 (s390_canonicalize_comparison): Likewise.
18966 (s390_emit_compare_and_swap): Likewise.
18967 (s390_branch_condition_mask): Likewise.
18968 (s390_rtx_costs): Likewise.
18969 (s390_secondary_reload): Likewise.
18970 (__SECONDARY_RELOAD_CASE): Likewise.
18971 (s390_expand_cs): Likewise.
18972 (s390_preferred_simd_mode): Likewise.
18973 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18974 * config/sh/sh.c (sh_print_operand): Likewise.
18975 (dump_table): Likewise.
18976 (sh_secondary_reload): Likewise.
18977 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18978 * config/sh/sh.md (casesi_worker_1): Likewise.
18979 (casesi_worker_2): Likewise.
18980 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18981 (fcc_comparison_operator): Likewise.
18982 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18983 (emit_soft_tfmode_cvt): Likewise.
18984 (sparc_preferred_simd_mode): Likewise.
18985 (output_cbranch): Likewise.
18986 (sparc_print_operand): Likewise.
18987 (sparc_expand_vec_perm_bmask): Likewise.
18988 (vector_init_bshuffle): Likewise.
18989 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18990 (spu_vector_mode_supported_p): Likewise.
18991 (spu_expand_insv): Likewise.
18992 (spu_emit_branch_or_set): Likewise.
18993 (spu_handle_vector_attribute): Likewise.
18994 (spu_builtin_splats): Likewise.
18995 (spu_builtin_extract): Likewise.
18996 (spu_builtin_promote): Likewise.
18997 (spu_expand_sign_extend): Likewise.
18998 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18999 (tilegx_simd_int): Likewise.
19000 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19001 (tilepro_simd_int): Likewise.
19002 * config/v850/v850.c (const_double_split): Likewise.
19003 (v850_print_operand): Likewise.
19004 (ep_memory_offset): Likewise.
19005 * config/vax/vax.c (vax_rtx_costs): Likewise.
19006 (vax_output_int_move): Likewise.
19007 (vax_output_int_add): Likewise.
19008 (vax_output_int_subtract): Likewise.
19009 * config/visium/predicates.md (visium_branch_operator): Likewise.
19010 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19011 (visium_print_operand_address): Likewise.
19012 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19013 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19014 (xtensa_expand_conditional_branch): Likewise.
19015 (xtensa_copy_incoming_a7): Likewise.
19016 (xtensa_output_literal): Likewise.
19017 * dfp.c (decimal_real_maxval): Likewise.
19018 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19021 Alan Hayward <alan.hayward@arm.com>
19022 David Sherwood <david.sherwood@arm.com>
19024 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19025 (mode_nunits_inline): Likewise.
19026 (mode_inner_inline): Likewise.
19027 (mode_unit_size_inline): Likewise.
19028 (mode_unit_precision_inline): Likewise.
19029 (emit_insn_modes_h): Likewise. Also emit a #define of the
19031 (emit_mode_wider): Add an E_ prefix to mode names.
19032 (emit_mode_complex): Likewise.
19033 (emit_mode_inner): Likewise.
19034 (emit_mode_adjustments): Likewise.
19035 (emit_mode_int_n): Likewise.
19036 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19037 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19038 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19039 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19040 (CRC32_BUILTIN, ENTRY): Likewise.
19041 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19042 (aarch64_pop_regs): Likewise.
19043 (aarch64_process_components): Likewise.
19044 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19045 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19046 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19047 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19048 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19049 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19051 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19052 (pcmpistr): Likewise.
19053 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19054 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19055 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19057 * config/rl78/rl78.c (mduc_regs): Likewise.
19058 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19059 (htm_expand_builtin): Likewise.
19060 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19061 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19062 * config/xtensa/xtensa.c (print_operand): Likewise.
19063 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19064 (NUM_MODE_VECTOR_INT): Likewise.
19065 * genoutput.c (null_operand): Likewise.
19066 (output_operand_data): Likewise.
19067 * genrecog.c (print_parameter_value): Likewise.
19068 * lra.c (debug_operand_data): Likewise.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19072 * dwarf2out.c (scompare_loc_descriptor_wide)
19073 (scompare_loc_descriptor_narrow): New functions, split out from...
19074 (scompare_loc_descriptor): ...here.
19075 * expmed.c (emit_store_flag_int): New function, split out from...
19076 (emit_store_flag): ...here.
19078 2017-08-30 Richard Biener <rguenther@suse.de>
19080 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19081 (dwarf2out_early_finish): Move setting of AT_pubnames from
19082 early debug output to early finish.
19084 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19086 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19087 and -mdata-region to the assembler.
19089 2017-08-30 Richard Biener <rguenther@suse.de>
19091 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19093 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19094 (dwarf2out_early_global_decl): Only generate a DIE for the
19095 abstract origin if it doesn't already exist or is a declaration DIE.
19096 (resolve_addr): Do not add the linkage name twice when
19097 generating a stub DIE for the DW_TAG_GNU_call_site target.
19099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19101 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19102 Use machine_mode rather than int for arg1_mode.
19104 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19107 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19108 that the second argument of the built-in functions to unpack
19109 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19110 switch statement instead a lot of if statements.
19111 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19112 Allow 64-bit values to be in Altivec registers as well as
19113 traditional floating point registers.
19114 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19116 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19118 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19119 MAX_REGS_PER_ADDRESS == 1.
19121 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19123 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19124 * config/i386/i386.c (ix86_option_override_internal): Simplify
19125 setting of opts->x_flag_entry.
19127 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19128 Jakub Jelinek <jakub@redhat.com>
19129 Richard Biener <rguenther@suse.de>
19131 PR tree-optimization/81503
19132 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19133 folded constant fits in the target type; reorder tests for clarity.
19135 2017-08-29 Martin Liska <mliska@suse.cz>
19137 * passes.def: Include pass_lower_switch.
19138 * stmt.c (dump_case_nodes): Remove and move to
19139 tree-switch-conversion.
19140 (case_values_threshold): Likewise.
19141 (expand_switch_as_decision_tree_p): Likewise.
19142 (emit_case_decision_tree): Likewise.
19143 (expand_case): Likewise.
19144 (balance_case_nodes): Likewise.
19145 (node_has_low_bound): Likewise.
19146 (node_has_high_bound): Likewise.
19147 (node_is_bounded): Likewise.
19148 (emit_case_nodes): Likewise.
19149 (struct simple_case_node): New struct.
19150 (add_case_node): Remove.
19151 (emit_case_dispatch_table): Use vector instead of case_list.
19152 (reset_out_edges_aux): Remove.
19153 (compute_cases_per_edge): Likewise.
19154 (expand_case): Build list of simple_case_node.
19155 (expand_sjlj_dispatch_table): Use it.
19156 * tree-switch-conversion.c (struct case_node): Moved from
19157 stmt.c and adjusted.
19158 (emit_case_nodes): Likewise.
19159 (node_has_low_bound): Likewise.
19160 (node_has_high_bound): Likewise.
19161 (node_is_bounded): Likewise.
19162 (case_values_threshold): Likewise.
19163 (reset_out_edges_aux): Likewise.
19164 (compute_cases_per_edge): Likewise.
19165 (add_case_node): Likewise.
19166 (dump_case_nodes): Likewise.
19167 (balance_case_nodes): Likewise.
19168 (expand_switch_as_decision_tree_p): Likewise.
19169 (emit_jump): Likewise.
19170 (emit_case_decision_tree): Likewise.
19171 (try_switch_expansion): Likewise.
19172 (do_jump_if_equal): Likewise.
19173 (emit_cmp_and_jump_insns): Likewise.
19174 (fix_phi_operands_for_edge): New function.
19175 (record_phi_operand_mapping): Likewise.
19176 (class pass_lower_switch): New pass.
19177 (pass_lower_switch::execute): New function.
19178 (make_pass_lower_switch): Likewise.
19179 (conditional_probability):
19180 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19181 * tree-pass.h: Add make_pass_lower_switch.
19183 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19186 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19189 2017-08-29 Richard Biener <rguenther@suse.de>
19191 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19192 we do not add a DW_AT_inline attribute twice.
19193 (gen_subprogram_die): Remove code setting DW_AT_inline on
19194 DECL_ABSTRACT_P nodes.
19196 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19198 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19199 calls to internal functions.
19200 (gimplify_modify_expr): Likewise.
19201 * tree-call-cdce.c (use_internal_fn): Likewise.
19202 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19203 (convert_to_divmod): Set the nothrow flag.
19204 * tree-if-conv.c (predicate_mem_writes): Likewise.
19205 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19206 (vectorizable_call): Likewise.
19207 (vectorizable_store): Likewise.
19208 (vectorizable_load): Likewise.
19209 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19210 (vect_recog_mask_conversion_pattern): Likewise.
19212 2017-08-29 Martin Liska <mliska@suse.cz>
19215 * gcc.c (driver_handle_option): Add new argument.
19216 * opts-common.c (handle_option): Pass
19217 target_option_override_hook.
19218 * opts-global.c (lang_handle_option): Add new option.
19219 (set_default_handlers): Add new argument.
19220 (decode_options): Likewise.
19221 * opts.c (target_handle_option): Likewise.
19222 (common_handle_option): Call target_option_override_hook.
19223 * opts.h (struct cl_option_handler_func): Add hook for
19224 target option override.
19225 (struct cl_option_handlers): Likewise.
19226 (set_default_handlers): Add new argument.
19227 (decode_options): Likewise.
19228 (common_handle_option): Likewise.
19229 (target_handle_option): Likewise.
19230 * toplev.c (toplev::main): Pass targetm.target_option.override
19233 2017-08-29 Richard Biener <rguenther@suse.de>
19234 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19236 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19237 life to the active subtree.
19239 2017-08-28 Jeff Law <law@redhat.com>
19241 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19242 derive_equivalences.
19243 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19244 Code moved into....
19245 (edge_info::derive_equivalences): New private member function
19247 * tree-ssa-dom.c (class edge_info): Changed from a struct
19248 to a class. Add ctor/dtor, methods and data members.
19249 (edge_info::edge_info): Renamed from allocate_edge_info.
19250 Initialize additional members.
19251 (edge_info::~edge_info): New.
19252 (free_dom_edge_info): Delete the edge info.
19253 (record_edge_info): Use new class & associated member functions.
19254 Tighten forms for testing for edge equivalences.
19255 (record_temporary_equivalences): Iterate over the simple
19256 equivalences rather than assuming there's only one per edge.
19257 (cprop_into_successor_phis): Iterate over the simple
19258 equivalences rather than assuming there's only one per edge.
19259 (optimize_stmt): Use operand_equal_p rather than pointer
19260 equality for mini-DSE code.
19262 2017-08-28 Nathan Sidwell <nathan@acm.org>
19264 * gcc.c (execute): Fold SIGPIPE handling into switch
19265 statement. Adjust internal error message.
19267 2017-08-28 Richard Biener <rguenther@suse.de>
19270 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19271 Do nothing for removed DIEs.
19273 2017-08-28 Richard Biener <rguenther@suse.de>
19275 PR tree-optimization/81977
19276 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19279 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19282 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19283 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19284 using targetm.gen_mem_thread_fence.
19286 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19289 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19290 predicate to register_operand. Reorder operands.
19291 (*btr<mode>): Ditto.
19292 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19293 (*btr<mode>_mask): Ditto.
19295 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19297 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19298 * config/rs6000/xmmintrin.h: New file.
19299 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19301 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19304 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19305 parameter and_insn and return it.
19306 (recombine_lvx_pattern): Insert a copy to ensure availability of
19307 the base register of the copied masking operation at the point of
19308 the instruction replacement.
19309 (recombine_stvx_pattern): Likewise.
19311 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19313 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19314 undocumented switches.
19315 (-mpower9-dform-vector): Likewise.
19316 (-mpower9-dform): Likewise.
19317 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19318 comments to delete references to -mpower9-dform* switches.
19319 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19320 Delete reference to -mpower9-dform* switches, test for
19321 -mpower9-vector instead.
19322 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19323 (OTHER_P9_VECTOR_MASKS): Likewise.
19324 (POWERPC_MASKS): Likewise.
19325 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19326 tests against -mpower9-dform* to -mpower9-vector. Delete code
19327 that checked for -mpower9-dform* consistancy with other options.
19328 Add test for -mpower9-misc to enable other power9 switches.
19329 (rs6000_init_hard_regno_mode_ok): Likewise.
19330 (rs6000_option_override_internal): Likewise.
19331 (rs6000_emit_prologue): Likewise.
19332 (rs6000_emit_epilogue): Likewise.
19333 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19334 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19335 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19337 (emit_fusion_p9_store): Likewise.
19338 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19339 resetting these macros if the assembler does not support ISA 3.0
19341 (TARGET_P9_DFORM_VECTOR): Likewise.
19342 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19343 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19345 2017-08-25 Alan Modra <amodra@gmail.com>
19347 PR rtl-optimization/81747
19348 * cse.c (cse_extended_basic_block): Don't attempt to record
19349 equivalences for degenerate conditional branches that branch
19350 to their fall-through.
19352 2017-08-24 Martin Sebor <msebor@redhat.com>
19354 PR middle-end/81908
19355 * gimple-fold.c (size_must_be_zero_p): New function.
19356 (gimple_fold_builtin_memory_op): Call it.
19358 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19360 * config/rs6000/mm_malloc.h: New file.
19362 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19364 PR tree-optimization/81913
19365 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19366 analysis when either IVs in condition can wrap.
19368 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19370 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19371 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19373 2017-08-24 Richard Biener <rguenther@suse.de>
19376 * targhooks.c (default_target_can_inline_p): Properly
19377 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19378 is present and always compare.
19379 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19380 infer -mfpmath=sse from TARGET_SSE_P.
19381 (ix86_can_inline_p): Properly use target_option_default_node when
19382 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19384 2017-08-24 Richard Biener <rguenther@suse.de>
19387 * dwarf2out.c (output_die): Handle flag_generate_offload like
19389 (output_comp_unit): Likewise.
19390 (gen_array_type_die): Likewise.
19391 (dwarf2out_early_finish): Likewise.
19392 (note_variable_value_in_expr): Likewise.
19393 (dwarf2out_finish): Likewise. Adjust assert.
19394 * cgraphunit.c (symbol_table::compile): Move setting of
19395 flag_generate_offload earlier ...
19396 (symbol_table::finalize_compilation_unit): ... here, before
19397 early debug finalization.
19399 2017-08-24 Richard Biener <rguenther@suse.de>
19401 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19402 and ipa-fnsummary.h.
19403 (ix86_can_inline_p): When ix86_fpmath flags do not match
19404 check whether the callee uses FP math at all.
19406 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19408 PR middle-end/81931
19409 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19410 instead of TYPE_PRECISION.
19412 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19413 Alan Hayward <alan.hayward@arm.com>
19414 David Sherwood <david.sherwood@arm.com>
19416 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19418 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19420 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19422 (ix86_frame::stack_realign_allocate): New field.
19423 (struct machine_frame_state): Modify comments.
19424 (machine_frame_state::sp_realigned_fp_end): New field.
19425 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19426 layout calculation.
19427 (sp_valid_at): Add assertion to assure no attempt to access invalid
19428 offset of a realigned stack.
19429 (fp_valid_at): Likewise.
19430 (choose_baseaddr): Modify comments.
19431 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19432 ix86_expand_prologue.
19433 (ix86_expand_prologue): Modify stack realignment and allocation.
19434 (ix86_expand_epilogue): Modify comments.
19435 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19436 avx2_runtime, avx512f, and avx512f_runtime.
19438 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19440 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19441 (mstackrealign): Do not init to -1.
19442 * config/i386/i386.c (ix86_option_override_internal):
19443 Check opts_set, not opts when setting default value of
19444 opts->x_ix86_force_align_arg_pointer.
19446 2017-08-23 Richard Biener <rguenther@suse.de>
19448 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19449 lang_hooks.decl_printable_name.
19450 * print-rtl-function.c (print_rtx_function): Likewise.
19451 * tree-pretty-print.c (dump_function_header): Likewise.
19453 2017-08-23 Richard Biener <rguenther@suse.de>
19456 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19457 -g0 at compile-time.
19459 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19461 PR middle-end/19706
19462 * doc/sourcebuild.texi (Other hardware attributes):
19465 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19467 PR middle-end/19706
19468 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19469 Fix single-use check.
19471 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19473 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19475 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19477 * doc/install.texi: Modify to add more details on running selected
19480 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19482 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19483 is combined with -mabi=ms.
19484 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19487 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19489 PR tree-optimization/81488
19490 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19491 and cached_basis fields.
19492 (MAX_SPREAD): New constant.
19493 (alloc_cand_and_find_basis): Initialize new fields.
19494 (clear_visited): New function.
19495 (create_phi_basis_1): Rename from create_phi_basis, set visited
19496 and cached_basis fields.
19497 (create_phi_basis): New wrapper function.
19498 (phi_add_costs_1): Rename from phi_add_costs, add spread
19499 parameter, set visited field, short-circuit when limits reached.
19500 (phi_add_costs): New wrapper function.
19501 (record_phi_increments_1): Rename from record_phi_increments, set
19503 (record_phi_increments): New wrapper function.
19504 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19505 (phi_incr_cost): New wrapper function.
19506 (all_phi_incrs_profitable_1): Rename from
19507 all_phi_incrs_profitable, set visited field.
19508 (all_phi_incrs_profitable): New wrapper function.
19510 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19511 Alan Hayward <alan.hayward@arm.com>
19512 David Sherwood <david.sherwood@arm.com>
19514 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19515 that takes the outer and inner modes.
19516 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19517 comparison as the canonical test for a paradoxical subreg.
19518 * combine.c (simplify_set): Use paradoxical_subreg_p.
19519 (make_extraction): Likewise.
19520 (force_to_mode): Likewise.
19521 (rtx_equal_for_field_assignment_p): Likewise.
19522 (gen_lowpart_for_combine): Likewise.
19523 (simplify_comparison): Likewise.
19524 * cse.c (equiv_constant): Likewise.
19525 * expmed.c (store_bit_field_1): Likewise.
19526 * final.c (alter_subreg): Likewise.
19527 * fwprop.c (propagate_rtx): Likewise.
19528 (forward_propagate_subreg): Likewise.
19529 * ira-conflicts.c (ira_build_conflicts): Likewise.
19530 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19531 * lra-constraints.c (curr_insn_transform): Likewise.
19532 (split_reg): Likewise.
19533 * lra-eliminations.c (move_plus_up): Likewise.
19534 (lra_eliminate_regs_1): Likewise.
19535 * recog.c (general_operand): Likewise.
19536 * ree.c (combine_reaching_defs): Likewise.
19537 * reload.c (push_reload): Likewise.
19538 (find_reloads): Likewise.
19539 * reload1.c (elimination_effects): Likewise.
19540 (compute_reload_subreg_offset): Likewise.
19541 (choose_reload_regs): Likewise.
19542 * rtlanal.c (subreg_lsb_1): Likewise.
19543 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19544 (simplify_subreg): Likewise.
19545 * var-tracking.c (track_loc_p): Likewise.
19546 * emit-rtl.c (byte_lowpart_offset): Likewise.
19547 (paradoxical_subreg_p): Delete out-of-line definition.
19549 2017-08-22 Jeff Law <law@redhat.com>
19551 PR tree-optimization/81741
19552 PR tree-optimization/71947
19553 * tree-ssa-dom.c: Include tree-inline.h.
19554 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19555 equivalences if one is more expensive to compute than the other.
19556 * tree-ssa-scopedtables.h (class const_or_copies): Make
19557 record_const_or_copy_raw method private.
19558 (class avail_exprs_stack): New method simplify_binary_operation.
19559 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19560 avail_exprs_stack::simplify_binary_operation as needed.
19561 (avail_exprs_stack::simplify_binary_operation): New function.
19563 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19565 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19566 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19567 (DOT_SYMBOLS): Likewise.
19568 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19569 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19570 (RS6000_ABI_NAME): Likewise.
19571 (TARGET_CMODEL): Likewise.
19572 (TOC_SECTION_ASM_OP): Likewise.
19573 (SET_CMODEL): New macro.
19574 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19576 2017-08-22 Richard Biener <rguenther@suse.de>
19578 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19579 to allow for free-lang-data replacements similar to verify_type_variant.
19581 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19582 Alan Hayward <alan.hayward@arm.com>
19583 David Sherwood <david.sherwood@arm.com>
19585 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19586 VOIDmode for the LABEL_REF.
19588 2017-08-22 Richard Biener <rguenther@suse.de>
19590 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19592 2017-08-22 Richard Biener <rguenther@suse.de>
19594 * common.opt (feliminate-dwarf2-dups): Ignore.
19595 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19596 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19597 same_die_p_wrap, compute_section_prefix,
19598 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19599 (comdat_symbol_id, comdat_symbol_number): Likewise.
19600 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19602 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19603 (output_die): Mark unreachable path unreachable.
19604 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19605 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19606 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19607 (dwarf2out_early_finish): Likewise.
19609 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19611 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19613 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19616 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19617 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19618 (avr_attribute_table) <io, io_low, address>: Initialize
19619 .decl_required with true.
19621 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19623 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19624 undocumented debugging options.
19625 (-mvsx-scalar-double): Likewise.
19626 (-mallow-df-permute): Likewise.
19627 (-mvectorize-builtins): Likewise.
19628 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19629 (rs6000_builtin_vectorized_function): Likewise.
19630 (rs6000_builtin_md_vectorized_function): Likewise.
19631 (rs6000_opt_vars): Likewise.
19633 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19636 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19637 (*btrq_imm): Rename from *btrq.
19638 (*btcq_imm): Rename from *btcq.
19639 (btsc): New code attribute.
19640 (*<btsc><mode>): New insn pattern.
19641 (*btr<mode>): Ditto.
19642 (*<btsc><mode>_mask): New insn_and_split pattern.
19643 (*btr<mode>_mask): Ditto.
19645 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19646 Alan Hayward <alan.hayward@arm.com>
19647 David Sherwood <david.sherwood@arm.com>
19649 * function.c (pad_below): Simplify padding calculation.
19651 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19652 Alan Hayward <alan.hayward@arm.com>
19653 David Sherwood <david.sherwood@arm.com>
19655 * target.def (function_prologue): Remove frame size argument.
19656 (function_epilogue): Likewise.
19657 * doc/tm.texi: Regenerate.
19658 * final.c (final_start_function): Update call to function_prologue.
19659 (final_end_function): Update call to function_epilogue.
19660 (default_function_pro_epilogue): Remove frame size argument.
19661 * output.h (default_function_pro_epilogue): Likewise.
19662 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19663 (arm_output_function_prologue): Likewise.
19664 * config/frv/frv.c (frv_function_prologue): Likewise.
19665 (frv_function_epilogue): Likewise.
19666 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19667 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19668 (ia64_output_function_epilogue): Likewise.
19669 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19670 (m32r_output_function_epilogue): Likewise.
19671 * config/microblaze/microblaze.c (microblaze_function_prologue)
19672 (microblaze_function_epilogue): Likewise.
19673 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19674 (mips_output_function_epilogue): Likewise.
19675 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19676 (mmix_target_asm_function_epilogue): Likewise.
19677 * config/msp430/msp430.c (msp430_start_function): Likewise.
19678 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19679 (nds32_asm_function_epilogue): Likewise.
19680 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19681 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19682 (pa_output_function_epilogue): Likewise.
19683 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19684 (rs6000_output_function_epilogue): Likewise.
19685 * config/rl78/rl78.c (rl78_start_function): Likewise.
19686 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19687 (rs6000_output_function_epilogue): Likewise.
19688 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19689 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19690 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19691 (sparc_asm_function_epilogue): Likewise.
19693 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19695 * tree.h (type_has_mode_precision_p): New function.
19696 * convert.c (convert_to_integer_1): Use it.
19697 * expr.c (expand_expr_real_2): Likewise.
19698 (expand_expr_real_1): Likewise.
19699 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19700 * match.pd: Likewise.
19701 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19702 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19703 * tree-tailcall.c (process_assignment): Likewise.
19704 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19705 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19706 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19707 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19708 (vectorizable_assignment): Likewise.
19709 (vectorizable_shift): Likewise.
19710 (vectorizable_operation): Likewise.
19711 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19713 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19715 * match.pd: Add pow (C, x) simplification.
19717 2017-08-21 Richard Biener <rguenther@suse.de>
19719 PR tree-optimization/81900
19720 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19721 for blocks with abnormal predecessors.
19722 (compute_antic): Do not set visited flag prematurely.
19724 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19727 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19729 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19731 * stor-layout.h (vector_type_mode): Move to...
19732 * tree.h (vector_type_mode): ...here.
19733 * stor-layout.c (vector_type_mode): Move to...
19734 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19736 2017-08-21 Richard Biener <rguenther@suse.de>
19738 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19739 register_external_die hooks.
19740 (debug_false_tree_charstarstar_uhwistar): Declare.
19741 (debug_nothing_tree_charstar_uhwi): Likewise.
19742 * debug.c (do_nothing_debug_hooks): Adjust.
19743 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19744 (debug_nothing_tree_charstar_uhwi): Likewise.
19745 * dbxout.c (dbx_debug_hooks): Adjust.
19746 (xcoff_debug_hooks): Likewise.
19747 * sdbout.c (sdb_debug_hooks): Likewise.
19748 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19749 * dwarf2out.c (macinfo_label_base): New global.
19750 (dwarf2out_register_external_die): New function for the
19751 register_external_die hook.
19752 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19753 (dwarf2_debug_hooks): Use them.
19754 (dwarf2_lineno_debug_hooks): Adjust.
19755 (struct die_struct): Add with_offset flag.
19756 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19757 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19758 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19759 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19760 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19761 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19762 defining section names for the early LTO debug variants.
19763 (reset_indirect_string): New helper.
19764 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19765 (print_dw_val): Add support for offsetted symbol references.
19766 (get_ultimate_context): Split out from is_cxx.
19767 (is_cxx): Use get_ultimate_context.
19768 (is_fortran): Add decl overload.
19769 (compute_comp_unit_symbol): Split out worker from
19770 compute_section_prefix.
19771 (compute_section_prefix): Call compute_comp_unit_symbol and
19772 set comdat_type_p here.
19773 (output_die): Skip DIE symbol output for the LTO added one.
19774 Handle DIE symbol references with offset.
19775 (output_comp_unit): Guard section name mangling properly.
19776 For LTO debug sections emit a symbol at the section beginning
19777 which we use to refer to its DIEs.
19778 (add_abstract_origin_attribute): For DIEs registered via
19779 dwarf2out_register_external_die directly refer to the early
19780 DIE rather than indirectly through the shadow one we created.
19781 Remove obsolete call to dwarf2out_abstract_function for
19782 non-function/block origins.
19783 (gen_array_type_die): When generating early LTO debug do
19784 not emit DW_AT_string_length.
19785 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19786 late when in LTO. As suggested place a gcc_unreachable for
19787 the DECL_ABSTRACT_P case.
19788 (gen_subprogram_die): Avoid another specification DIE
19789 for early built declarations/definitions for the late LTO case.
19790 (gen_variable_die): Add type references for late duplicated VLA dies
19792 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19793 we have the abstract instance already.
19794 (process_scope_var): Adjust decl DIE contexts in LTO which
19795 first puts them in limbo.
19796 (gen_decl_die): Do not generate type DIEs late apart from
19797 types for VLAs or for decls we do not yet have a DIE. Do not
19798 call dwarf2out_abstract_function late.
19799 (dwarf2out_early_global_decl): Make sure to create DIEs
19800 for abstract instances of a decl first.
19801 (dwarf2out_late_global_decl): Adjust comment.
19802 (output_macinfo_op): With multiple macro sections use
19803 macinfo_label_base to distinguish labels.
19804 (output_macinfo): Likewise. Update macinfo_label_base.
19805 Pass in the line info label.
19806 (note_variable_value_in_expr): When generating LTO resolve
19807 all variable values here by generating DIEs as needed.
19808 (init_sections_and_labels): Add early LTO debug flag parameter
19809 and generate different sections and names if set. Add generation
19810 counter for the labels so we can have multiple of them.
19811 (reset_dies): Helper to allow DIEs to be output multiple times.
19812 (dwarf2out_finish): When outputting DIEs to the fat part of an
19813 LTO object first reset DIEs.
19814 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19815 (modified_type_die): Check for decl_ultimate_origin being self
19817 (gen_type_die_with_usage): Likewise.
19818 (gen_typedef_die): Allow decl_ultimate_origin being self.
19819 (set_decl_abstract_flags): Remove.
19820 (set_block_abstract_flags): Likewise.
19821 (dwarf2out_abstract_function): Treat the early generated DIEs
19822 as the abstract copy and only add DW_AT_inline and
19823 DW_AT_artificial here and call set_decl_origin_self.
19824 If the DIE has an abstract origin don't do anything.
19825 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19826 if we have none yet (Go fails to build one, PR78628).
19827 (variably_modified_type_p): Prevent endless recursion for Ada
19828 cyclic pointer types.
19829 * lto-streamer-in.c: Include debug.h.
19830 (dref_queue): New global.
19831 (lto_read_tree_1): Stream in DIE references.
19832 (lto_input_tree): Register DIE references.
19833 (input_function): Stream DECL_DEBUG_ARGS.
19834 * lto-streamer-out.c: Include debug.h.
19835 (lto_write_tree_1): Output DIE references.
19836 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19837 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19838 (output_function): Stream DECL_DEBUG_ARGS.
19839 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19840 Stream DECL_ABSTRACT_ORIGIN.
19841 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19842 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19843 DECL_CONTEXT for file-scope decls.
19844 * lto-streamer.h (struct dref_entry): Declare.
19845 (dref_queue): Likewise.
19846 * cfgexpand.c (pass_expand::execute): Do not call the
19847 outlining_inline_function hook here.
19848 * lto-wrapper.c (debug_obj): New global.
19849 (tool_cleanup): Unlink it if required.
19850 (debug_objcopy): New function.
19851 (run_gcc): Handle early debug sections in the IL files by
19852 extracting them to separate files, partially linkin them and
19853 feeding the result back as result to the linker.
19854 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19855 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19856 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19857 sections into a separate segment.
19858 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19860 (darwin_asm_dwarf_section): Likewise.
19861 (darwin_asm_output_dwarf_offset): Likewise.
19862 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19864 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19865 Alan Hayward <alan.hayward@arm.com>
19866 David Sherwood <david.sherwood@arm.com>
19868 * read-md.h (md_reader::record_potential_iterator_use): Replace
19869 pointer argument with an rtx and an index.
19870 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19871 (apply_mode_iterator): Likewise.
19872 (apply_code_iterator): Likewise.
19873 (apply_int_iterator): Likewise.
19874 (apply_subst_iterator): Likewise.
19875 (record_iterator_use): Likewise.
19876 (record_attribute_use): Likewise.
19877 (md_reader::record_potential_iterator_use): Likewise. Update calls
19878 to record_iterator_use and apply_iterator.
19879 (iterator_use): Replace ptr with x and index.
19880 (attribute_use): Likewise.
19881 (apply_attribute_uses): Update calls to apply_iterator.
19882 (apply_iterators): Likewise. Update initialization of iterator_use.
19883 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19884 and record_potential_iterator_use.
19885 (rtx_reader::read_rtx_operand): Likewise.
19887 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19888 Alan Hayward <alan.hayward@arm.com>
19889 David Sherwood <david.sherwood@arm.com>
19891 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19894 2017-08-21 Richard Biener <rguenther@suse.de>
19896 PR middle-end/81884
19897 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19898 at struct end conservatively when comparing common bases.
19900 2017-08-21 Richard Biener <rguenther@suse.de>
19902 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19903 (mem_ref_in_stmt): Remove.
19904 (determine_max_movement): Use ref index to get at the reference.
19905 (invariantness_dom_walker::before_dom_children): Deal with
19906 lim data already initialized.
19907 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19909 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19911 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19912 (TARGET_ISA_ROUND): Ditto.
19913 (TARGET_ROUND): Ditto.
19914 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19915 * config/i386/i386.md: Ditto.
19916 * config/i386/sse.md: Ditto.
19917 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19918 with OPTION_MASK_ISA_SSE4_1.
19920 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19923 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19924 __builtin_ia32_lzcnt_u16.
19926 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19929 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19930 (rs6000_set_current_function): Rewrite function to use it.
19932 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19935 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19936 and TYPE_WARN_IF_NOT_ALIGN.
19937 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19938 (handle_warn_if_not_align): New.
19939 (place_union_field): Call handle_warn_if_not_align.
19940 (place_field): Call handle_warn_if_not_align.
19941 Copy TYPE_WARN_IF_NOT_ALIGN.
19942 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19943 (layout_type): Likewise.
19944 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19946 (tree_decl_common): Add warn_if_not_align.
19947 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19948 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19949 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19950 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19951 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19952 * doc/extend.texi: Document warn_if_not_aligned attribute.
19953 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19955 2017-08-17 Martin Liska <mliska@suse.cz>
19958 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19959 (get_data_dependence): Use it as pointer type.
19960 (distribute_loop): Likewise.
19962 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19964 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19965 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19966 (p8_vmrgow_<mode>_direct): New define_insn.
19967 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19968 handle endianness for vmrgew and vmrgow permute patterns.
19970 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19972 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19973 * config/rs6000/rs6000-cpus.def: Remove comment.
19974 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19975 (POWERPC_MASKS): Likewise.
19976 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19977 use of TARGET_VSX_TIMODE.
19978 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19979 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19980 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19981 (rs6000_option_override_internal): Remove dead code.
19982 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19983 (rs6000_legitimize_reload_address): Likewise.
19984 (rs6000_legitimate_address_p): Likewise.
19985 (rs6000_opt_masks): Delete "vsx-timode".
19986 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19987 from function comment.
19988 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19989 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19990 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19992 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19993 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19994 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19995 (VSX_TI): Likewise.
19997 (define_peephole2): Likewise.
19999 2017-08-17 Martin Sebor <msebor@redhat.com>
20002 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20003 past the end of an array.
20004 (test_pp_format): Add test cases.
20006 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20008 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20009 missing ECF_NOTHROW flags.
20011 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20014 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20015 operands residing in integer registers.
20016 (*vsx_le_perm_load_<mode>): Likewise.
20017 (*vsx_le_perm_store_<mode>): Likewise.
20018 (define_peephole2): Add peepholes to optimize the above.
20020 2017-08-17 Marek Polacek <polacek@redhat.com>
20022 PR middle-end/81814
20023 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20024 to mimic what shorten_compare did. Change the return type to bool.
20025 (fold_cond_expr_with_comparison): Update call to
20026 operand_equal_for_comparison_p.
20027 (fold_ternary_loc): Likewise.
20029 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20031 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20033 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20035 2017-08-17 Richard Biener <rguenther@suse.de>
20037 * tree-ssa-structalias.c (solve_graph): When propagating
20038 to successors update the graphs succ edges and avoid duplicate work.
20040 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20043 * config/i386/i386.c (ix86_option_override_internal): Save target
20044 specific options after ix86_stack_protector_guard_reg was changed.
20046 2017-08-17 Richard Biener <rguenther@suse.de>
20048 PR tree-optimization/81827
20049 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20050 (new_var_info): Initialize it conservatively.
20051 (get_call_vi): Mark register vars.
20052 (new_scalar_tmp_constraint_exp): Likewise.
20053 (handle_rhs_call): Likewise.
20054 (handle_const_call): Likewise.
20055 (create_function_info_for): Likewise.
20056 (solve_constraints): Sort varinfos to separate register from
20057 non-register vars to pack points-to solution bitmaps during
20060 2017-08-17 Marek Polacek <polacek@redhat.com>
20062 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20064 2017-08-17 Richard Biener <rguenther@suse.de>
20066 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20067 to true when overflow is undefined and we saturated the result.
20069 2017-08-17 Alan Modra <amodra@gmail.com>
20072 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20073 Don't use store multiple if only one reg needs saving.
20074 (interesting_frame_related_regno): New function.
20075 (rs6000_frame_related): Don't emit frame info for regs that
20077 (rs6000_emit_epilogue): Likewise.
20079 2017-08-16 Nathan Sidwell <nathan@acm.org>
20081 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20082 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20083 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20084 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20085 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20086 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20087 (verify_type): Adjust for TYPE_BINFO move.
20088 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20089 process TYPE_BINFO directly.
20090 (hash_tree): Likewise.
20091 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20093 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20096 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20098 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20100 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20103 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20104 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20105 Add ix86_binary_operator_ok to insn constraint.
20106 (*iordi_1_bts): Ditto.
20107 (*xordi_1_btc): Ditto.
20108 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20109 Update corresponding peephole2 pattern.
20113 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20115 PR tree-optimization/81832
20116 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20117 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20119 2017-08-16 Marek Polacek <polacek@redhat.com>
20122 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20123 regarding size_zero_node.
20125 2017-08-16 Martin Liska <mliska@suse.cz>
20128 * config.gcc: Respect previously set extra_objs in case
20131 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20133 PR tree-optimization/81835
20134 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20135 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20136 not depend on the phi.
20138 2017-08-16 Alan Modra <amodra@gmail.com>
20140 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20143 2017-08-16 Alan Modra <amodra@gmail.com>
20145 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20146 (save_reg_p): ..into this. Update all callers.
20147 (first_reg_to_save): Simplify.
20149 2017-08-16 Alan Modra <amodra@gmail.com>
20151 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20154 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20158 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20159 constant count if that count is less than 32.
20161 2017-08-15 Nathan Sidwell <nathan@acm.org>
20163 * gcc.c (execute): Emit friendlier message if inferior is killed
20164 by an external cause.
20166 2017-08-15 Richard Biener <rguenther@suse.de>
20168 PR tree-optimization/81790
20169 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20170 CONSTRUCTORs from simplifying and VN.
20172 2017-08-14 Martin Sebor <msebor@redhat.com>
20174 * builtin-attrs.def: Add comments.
20176 2017-08-14 Martin Sebor <msebor@redhat.com>
20179 * doc/extend.texi (attribute nonstring): Document new attribute.
20181 2017-08-14 Martin Sebor <msebor@redhat.com>
20184 * tree-diagnostic.c (default_tree_printer): Handle %G.
20185 * gimple-pretty-print.h (percent_G_format): Declare new function.
20186 * gimple-pretty-print.c (percent_G_format): Define.
20187 * tree-pretty-print.c (percent_K_format): Add argument.
20189 2017-08-14 Martin Sebor <msebor@redhat.com>
20191 PR translation/79998
20192 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20193 Remove a stray space.
20195 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20198 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20199 (*iordi_1_bts): Ditto.
20200 (*xordi_1_btc): Ditto.
20202 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20205 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20206 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20208 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20209 quoted strings, and make more translator-friendly.
20210 (darwin_rs6000_override_options): Likewise.
20211 (rs6000_option_override_internal): Likewise.
20212 (rs6000_return_in_memory): Fix overlong line.
20213 (init_cmulative_args): Use quoted strings, and make more
20214 translator-friendly.
20215 (rs6000_pass_by_reference): Fix overlong line.
20216 (def_builtin): Use quoted strings.
20217 (altivec_expand_predicate_builtin): Use quoted strings, and make
20218 more translator-friendly.
20219 (htm_expand_builtin): Use quoted strings.
20220 (cpu_expand_builtin): Use quoted strings, and make more
20221 translator-friendly.
20222 (altivec_expand_builtin): Likewise.
20223 (paired_expand_predicate_builtin): Likewise.
20224 (rs6000_invalid_builtin): Likewise.
20225 (builtin_function_type): Use quoted strings.
20226 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20227 more translator-friendly.
20228 (rs6000_trampoline_init): Likewise.
20229 (rs6000_handle_altivec_attribute): Likewise.
20230 (rs6000_inner_target_options): Use quoted strings.
20231 (rs6000_disable_incompatible_switches): Likewise.
20232 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20233 strings, and make more translator-friendly.
20234 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20236 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20238 PR tree-optimization/81799
20239 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20240 cond_expr to simple gimple operand.
20242 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20244 PR middle-end/46932
20245 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20247 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20251 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20252 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20253 TARGET_GASISR_PROLOGUES.
20254 * config/avr/avr.c (avr_option_override): Same.
20255 (avr_pass_pre_proep::execute): Same.
20257 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20260 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20261 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20263 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20265 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20266 define_insn_and_split. Split to a memory load from 0 in
20267 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20268 using PTR mode iterator.
20269 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20270 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20271 (*add_tp_<mode>): Redefine as define_insn_and_split.
20272 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20273 address space. Merge with *add_tp_x32 using PTR mode iterator.
20274 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20275 Split to an add with a memory load from 0 in
20276 DEFAULT_TLS_SEG_REG address space.
20278 2017-08-12 Andrew Pinski <apinski@cavium.com>
20280 * config/aarch64/aarch64-option-extensions.def (rdma):
20281 Fix feature string to what Linux prints out in /proc/cpuinfo.
20283 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20286 * dwarf2out.c (modified_type_die): For C typedef types that have
20287 an ultimate origin, process the ultimate origin instead of the
20289 (gen_typedef_die): Assert that input DECLs have no ultimate
20291 (gen_type_die_with_usage): For typedef variants that have an
20292 ultimate origin, just call gen_decl_die on the original DECL.
20293 (process_scope_var): Avoid creating DIEs for local typedefs and
20294 concrete static variables.
20296 2017-08-12 Alan Modra <amodra@gmail.com>
20300 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20301 match gnu-user.h startfile.
20302 (ENDFILE_LINUX_SPEC): Similarly.
20304 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20307 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20309 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20311 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20312 * config/aarch64/aarch64.md (mov<mode>): Change.
20313 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20314 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20315 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20317 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20319 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20320 for storage order barriers.
20322 2017-08-11 Martin Liska <mliska@suse.cz>
20325 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20326 variables of void type.
20328 2017-08-11 Martin Liska <mliska@suse.cz>
20330 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20331 TARGET_SUPPORTS_ALIASES.
20332 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20333 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20334 (optimize_weakref): Likewise.
20335 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20336 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20337 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20339 2017-08-11 Martin Liska <mliska@suse.cz>
20342 * config/i386/i386.c (make_resolver_func): Do complete
20343 refactoring of the function.
20345 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20348 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20349 * config/i386/i386.c (ix86_stack_protect_guard): Use
20350 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20351 * doc/invoke.texi (x86 Options): Document
20352 -mstack-protector-guard-symbol= option.
20354 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20356 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20357 * config/i386/i386.c (ix86_split_stack_guard): New function.
20358 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20359 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20360 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20361 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20362 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20363 (split_stack_space_check): Call ix86_split_stack_guard.
20365 2017-08-10 Martin Sebor <msebor@redhat.com>
20367 * print-tree.c (print_node): Print location using the established
20369 Replace spaces with colons.
20370 (debug_raw, debug): Ditto.
20372 2017-08-10 Martin Sebor <msebor@redhat.com>
20375 * pretty-print.c (pp_format): Correct the handling of %s precision.
20377 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20380 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20382 (ix86_finalize_stack_frame_flags): This. Also clear
20383 frame_pointer_needed if -fno-omit-frame-pointer is used without
20385 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20386 with ix86_finalize_stack_frame_flags.
20387 (ix86_expand_epilogue): Likewise.
20388 (ix86_expand_split_stack_prologue): Likewise.
20389 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20391 2017-08-10 Martin Liska <mliska@suse.cz>
20394 * c-attribs.c (handle_target_attribute):
20395 Report warning for an empty string argument of target attribute.
20397 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20400 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20402 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20403 or DECL_NONLOCAL labels.
20404 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20405 or DECL_NONLOCAL labels here.
20407 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20409 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20410 to indicate when early gimple folding has been disabled.
20411 (rs6000_gimple_fold_builtin): Add debug content.
20412 (rs6000_invalid_builtin): Fix whitespace.
20413 (rs6000_expand_builtin): Fix whitespace.
20414 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20416 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20419 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20420 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20422 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20424 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20425 (qdf24xx): Likewise.
20426 * config/aarch64/aarch64.md: Include falkor.md.
20427 * config/aarch64/falkor.md: New.
20429 2017-08-09 Marek Polacek <polacek@redhat.com>
20432 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20433 * diagnostic.c (emit_diagnostic): Add a comment.
20434 (emit_diagnostic_valist): New function.
20436 2017-08-09 Marek Polacek <polacek@redhat.com>
20439 * input.c (make_location): New overload.
20440 * input.h (make_location): Declare.
20442 2017-08-08 Alan Modra <amodra@gmail.com>
20443 H.J. Lu <hongjiu.lu@intel.com>
20446 * gcc.c (NO_PIE_SPEC): Delete.
20447 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20449 (LINK_PIE_SPEC): ..to here.
20450 (LINK_COMMAND_SPEC): Support -no-pie.
20451 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20452 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20454 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20455 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20456 (ENDFILE_CRTEND_SPEC): Similarly.
20458 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20461 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20462 (mstack-protector-guard-offset=): Ditto.
20463 * config/i386/i386.c (ix86_option_override): Handle
20464 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20466 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20467 ix86_stack_protect_guard_offset variables.
20468 (TARGET_STACK_PROTECT_GUARD): Always define.
20469 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20470 and -mstack-protector-guard-offset= options.
20472 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20474 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20475 boundary case for the last candidate.
20477 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20479 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20480 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20482 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20484 PR middle-end/19706
20485 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20486 * config/aarch64/aarch64-builtins.c
20487 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20488 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20489 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20491 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20492 Andrew Pinski <pinskia@gmail.com>
20494 PR middle-end/19706
20495 * internal-fn.def (XORSIGN): New.
20496 * optabs.def (xorsign_optab): New.
20497 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20498 (convert_expand_mult_copysign): New.
20499 (pass_optimize_widening_mul::execute): Call
20500 convert_expand_mult_copysign.
20502 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20504 PR tree-optimization/81354
20505 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20506 Insert on edges rather than explicitly creating landing pads.
20507 (analyze_candidates_and_replace): Commit edge inserts.
20509 2017-08-08 Richard Biener <rguenther@suse.de>
20511 PR middle-end/81719
20512 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20513 (expand_simple_operations): Also look through ADDR_EXPRs with
20514 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20516 2017-08-08 Richard Biener <rguenther@suse.de>
20518 PR tree-optimization/81723
20519 * tree-vect-slp.c (struct bst_traits): New hash traits.
20520 (bst_fail): New global.
20521 (vect_build_slp_tree_2): New worker, split out from ...
20522 (vect_build_slp_tree): ... this now wrapping it with using
20523 bst_fail set to cache SLP tree build fails. Properly handle
20525 (vect_analyze_slp_instance): Allocate and free bst_fail.
20527 2017-08-08 Martin Liska <mliska@suse.cz>
20530 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20531 LABEL_DECLs that can be from a different function.
20533 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20535 PR tree-optimization/81744
20536 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20537 loop's number of iterations.
20539 2017-08-08 Martin Liska <mliska@suse.cz>
20541 * asan.c: Include header files.
20542 * attribs.c (build_decl_attribute_variant): New function moved
20544 (build_type_attribute_qual_variant): Likewise.
20545 (cmp_attrib_identifiers): Likewise.
20546 (simple_cst_list_equal): Likewise.
20547 (omp_declare_simd_clauses_equal): Likewise.
20548 (attribute_value_equal): Likewise.
20549 (comp_type_attributes): Likewise.
20550 (build_type_attribute_variant): Likewise.
20551 (lookup_ident_attribute): Likewise.
20552 (remove_attribute): Likewise.
20553 (merge_attributes): Likewise.
20554 (merge_type_attributes): Likewise.
20555 (merge_decl_attributes): Likewise.
20556 (merge_dllimport_decl_attributes): Likewise.
20557 (handle_dll_attribute): Likewise.
20558 (attribute_list_equal): Likewise.
20559 (attribute_list_contained): Likewise.
20560 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20561 (lookup_attribute_by_prefix): Likewise.
20562 * bb-reorder.c: Include header files.
20563 * builtins.c: Likewise.
20564 * calls.c: Likewise.
20565 * cfgexpand.c: Likewise.
20566 * cgraph.c: Likewise.
20567 * cgraphunit.c: Likewise.
20568 * convert.c: Likewise.
20569 * dwarf2out.c: Likewise.
20570 * final.c: Likewise.
20571 * fold-const.c: Likewise.
20572 * function.c: Likewise.
20573 * gimple-expr.c: Likewise.
20574 * gimple-fold.c: Likewise.
20575 * gimple-pretty-print.c: Likewise.
20576 * gimple.c: Likewise.
20577 * gimplify.c: Likewise.
20578 * hsa-common.c: Likewise.
20579 * hsa-gen.c: Likewise.
20580 * internal-fn.c: Likewise.
20581 * ipa-chkp.c: Likewise.
20582 * ipa-cp.c: Likewise.
20583 * ipa-devirt.c: Likewise.
20584 * ipa-fnsummary.c: Likewise.
20585 * ipa-inline.c: Likewise.
20586 * ipa-visibility.c: Likewise.
20588 * lto-cgraph.c: Likewise.
20589 * omp-expand.c: Likewise.
20590 * omp-general.c: Likewise.
20591 * omp-low.c: Likewise.
20592 * omp-offload.c: Likewise.
20593 * omp-simd-clone.c: Likewise.
20594 * opts-global.c: Likewise.
20595 * passes.c: Likewise.
20596 * predict.c: Likewise.
20597 * sancov.c: Likewise.
20598 * sanopt.c: Likewise.
20599 * symtab.c: Likewise.
20600 * toplev.c: Likewise.
20601 * trans-mem.c: Likewise.
20602 * tree-chkp.c: Likewise.
20603 * tree-eh.c: Likewise.
20604 * tree-into-ssa.c: Likewise.
20605 * tree-object-size.c: Likewise.
20606 * tree-parloops.c: Likewise.
20607 * tree-profile.c: Likewise.
20608 * tree-ssa-ccp.c: Likewise.
20609 * tree-ssa-live.c: Likewise.
20610 * tree-ssa-loop.c: Likewise.
20611 * tree-ssa-sccvn.c: Likewise.
20612 * tree-ssa-structalias.c: Likewise.
20613 * tree-ssa.c: Likewise.
20614 * tree-streamer-in.c: Likewise.
20615 * tree-vectorizer.c: Likewise.
20616 * tree-vrp.c: Likewise.
20617 * tsan.c: Likewise.
20618 * ubsan.c: Likewise.
20619 * varasm.c: Likewise.
20620 * varpool.c: Likewise.
20621 * tree.c: Remove functions moved to attribs.[ch].
20622 * tree.h: Likewise.
20623 * config/aarch64/aarch64.c: Add attrs.h header file.
20624 * config/alpha/alpha.c: Likewise.
20625 * config/arc/arc.c: Likewise.
20626 * config/arm/arm.c: Likewise.
20627 * config/avr/avr.c: Likewise.
20628 * config/bfin/bfin.c: Likewise.
20629 * config/c6x/c6x.c: Likewise.
20630 * config/cr16/cr16.c: Likewise.
20631 * config/cris/cris.c: Likewise.
20632 * config/darwin.c: Likewise.
20633 * config/epiphany/epiphany.c: Likewise.
20634 * config/fr30/fr30.c: Likewise.
20635 * config/frv/frv.c: Likewise.
20636 * config/ft32/ft32.c: Likewise.
20637 * config/h8300/h8300.c: Likewise.
20638 * config/i386/winnt.c: Likewise.
20639 * config/ia64/ia64.c: Likewise.
20640 * config/iq2000/iq2000.c: Likewise.
20641 * config/lm32/lm32.c: Likewise.
20642 * config/m32c/m32c.c: Likewise.
20643 * config/m32r/m32r.c: Likewise.
20644 * config/m68k/m68k.c: Likewise.
20645 * config/mcore/mcore.c: Likewise.
20646 * config/microblaze/microblaze.c: Likewise.
20647 * config/mips/mips.c: Likewise.
20648 * config/mmix/mmix.c: Likewise.
20649 * config/mn10300/mn10300.c: Likewise.
20650 * config/moxie/moxie.c: Likewise.
20651 * config/msp430/msp430.c: Likewise.
20652 * config/nds32/nds32-isr.c: Likewise.
20653 * config/nds32/nds32.c: Likewise.
20654 * config/nios2/nios2.c: Likewise.
20655 * config/nvptx/nvptx.c: Likewise.
20656 * config/pa/pa.c: Likewise.
20657 * config/pdp11/pdp11.c: Likewise.
20658 * config/powerpcspe/powerpcspe.c: Likewise.
20659 * config/riscv/riscv.c: Likewise.
20660 * config/rl78/rl78.c: Likewise.
20661 * config/rx/rx.c: Likewise.
20662 * config/s390/s390.c: Likewise.
20663 * config/sh/sh.c: Likewise.
20664 * config/sol2.c: Likewise.
20665 * config/sparc/sparc.c: Likewise.
20666 * config/spu/spu.c: Likewise.
20667 * config/stormy16/stormy16.c: Likewise.
20668 * config/tilegx/tilegx.c: Likewise.
20669 * config/tilepro/tilepro.c: Likewise.
20670 * config/v850/v850.c: Likewise.
20671 * config/vax/vax.c: Likewise.
20672 * config/visium/visium.c: Likewise.
20673 * config/xtensa/xtensa.c: Likewise.
20675 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20678 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20679 constraints since the -mupper-regs-* switches have been
20681 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20682 into a vector from a double word element that was extracted from
20683 another vector, and eliminate extra XXPERMDI instructions.
20684 (vsx_concat_<mode>_2): Likewise.
20685 (vsx_concat_<mode>_3): Likewise.
20686 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20687 concat to allow optimizing inserts from previous extracts.
20689 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20691 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20692 memory reference to a SSP offset in TLS address space.
20693 (ix86_print_operand) <case '@'>: Remove.
20694 (ix86_print_operand_punct_valid_p): Remove '@' code.
20695 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20696 UNSPEC_SP_TLS_TEST.
20697 (stack_tls_protect_set_<mode>): Remove.
20698 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20699 (stack_tls_protect_test_<mode>): Remove.
20700 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20702 2017-08-07 Olivier Hainque <hainque@adacore.com>
20705 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20707 2017-08-07 Douglas Rupp <rupp@adacore.com>
20709 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20710 variable was referenced as multidir in command.
20712 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20715 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20718 2017-08-07 Martin Liska <mliska@suse.cz>
20720 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20721 * config/rl78/rl78.c: Add include of attribs.h.
20722 * config/sh/sh.c: Likewise.
20723 * config/v850/v850.c: Likewise.
20725 2017-08-07 Tom de Vries <tom@codesourcery.com>
20727 PR middle-end/78266
20728 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20730 2017-08-07 Martin Liska <mliska@suse.cz>
20732 * config/mips/mips.c: Include attribs.h.
20734 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20737 * doc/invoke.texi: Document change in behvaior for -Ofast for
20740 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20742 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20744 (aarch64_pop_regs): Likewise.
20745 (aarch64_gen_load_pair): Likewise.
20746 (aarch64_save_callee_saves): Likewise.
20747 (aarch64_restore_callee_saves): Likewise.
20749 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20751 * config/i386/i386.c: Revert the last change.
20753 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20756 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20758 (ix86_finalize_stack_frame_flags): This. Also clear
20759 frame_pointer_needed if -fno-omit-frame-pointer is used without
20761 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20762 with ix86_finalize_stack_frame_flags.
20763 (ix86_expand_epilogue): Likewise.
20764 (ix86_expand_split_stack_prologue): Likewise.
20766 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20769 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20770 to P_AES for Westmere.
20772 2017-08-07 Jonathan Yong <10walls@gmail.com>
20774 * config/i386/mingw.opt (fset-stack-executable): Removed.
20775 * config/i386/cygming.opt (fset-stack-executable): Moved
20777 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20779 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20781 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20783 2017-08-07 Marek Polacek <polacek@redhat.com>
20785 PR middle-end/81737
20786 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20788 2017-08-07 Martin Liska <mliska@suse.cz>
20790 * attribs.h (canonicalize_attr_name): New function.
20791 (cmp_attribs): Move from c-format.c and adjusted.
20792 (is_attribute_p): Moved from tree.h.
20793 * tree-inline.c: Add new includes.
20794 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20795 (private_is_attribute_p): Remove.
20796 (private_lookup_attribute): Likewise.
20797 (private_lookup_attribute_by_prefix): Simplify.
20798 (remove_attribute): Use is_attribute_p.
20799 * tree.h: Remove removed declarations.
20801 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20803 PR middle-end/81698
20804 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20805 instead of computing it in the function. Formatting fix.
20806 (expand_case): Don't rely on default_edge being the first edge,
20807 clear it if removing it, pass default_edge to
20808 emit_case_dispatch_table.
20809 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20812 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20814 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20815 insn in the function, emit NOP after the insn.
20817 2017-08-06 Tom de Vries <tom@codesourcery.com>
20819 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20822 2017-08-06 Tom de Vries <tom@codesourcery.com>
20824 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20827 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20829 PR tree-optimization/57371
20830 * match.pd: New pattern.
20832 2017-08-04 Marek Polacek <polacek@redhat.com>
20834 PR middle-end/81695
20835 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20836 perform the computation in offset_int.
20838 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20840 PR tree-optimization/81136
20841 * tree-vectorizer.h: Include tree-hash-traits.h.
20842 (vec_base_alignments): New typedef.
20843 (vec_info): Add a base_alignments field.
20844 (vect_record_base_alignments): Declare.
20845 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20847 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20848 (create_data_ref): Add an is_conditional_in_stmt argument.
20849 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20850 the is_conditional_in_stmt field.
20851 (data_ref_loc): Add an is_conditional_in_stmt field.
20852 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20853 (find_data_references_in_stmt): Update call to create_data_ref.
20854 (graphite_find_data_references_in_stmt): Likewise.
20855 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20856 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20857 (vect_record_base_alignment): New function.
20858 (vect_record_base_alignments): Likewise.
20859 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20860 for nested statements even if we fail to compute a misalignment.
20861 Use pooled base alignments for unconditional references.
20862 (vect_find_same_alignment_drs): Compare base addresses instead
20864 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20865 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20867 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20869 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20870 Add an explicit name for the enum. Use auto_vec for slp_instances
20871 and grouped_stores.
20872 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20874 (_bb_vec_info): Add a constructor and destructor.
20875 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20876 (destroy_loop_vec_info): Delete.
20877 (vect_destroy_datarefs): Likewise.
20878 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20879 (vec_info::vec_info): New function.
20880 (vec_info::~vec_info): Likewise.
20881 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20882 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20883 destroy_loop_vec_info.
20884 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20885 (_loop_vec_info::_loop_vec_info): ...this.
20886 (destroy_loop_vec_info): Replace with...
20887 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20888 the stmt_vec_infos. Leave handling of vec_info information to its
20889 destructor. Remove explicit vector releases.
20890 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20891 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20892 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20893 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20894 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20895 (destroy_bb_vec_info): Replace with...
20896 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20897 information to its destructor.
20898 (vect_slp_analyze_bb_1): Use new and delete instead of
20899 new_bb_vec_info and destroy_bb_vec_info.
20900 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20903 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20905 * tree-data-ref.h (subscript): Add access_fn field.
20906 (data_dependence_relation): Add could_be_independent_p.
20907 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20908 (same_access_functions): Move to tree-data-ref.c.
20909 * tree-data-ref.c (ref_contains_union_access_p): New function.
20910 (access_fn_component_p): Likewise.
20911 (access_fn_components_comparable_p): Likewise.
20912 (dr_analyze_indices): Add a reference to access_fn_component_p.
20913 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20915 (constant_access_functions): Likewise.
20916 (add_other_self_distances): Likewise.
20917 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20918 (initialize_data_dependence_relation): Use XCNEW and remove
20919 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20920 of access functions that have the same type. Allow the
20921 subsequence to end with different bases in some circumstances.
20922 Record the chosen access functions in SUB_ACCESS_FN.
20923 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20924 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20925 (subscript_dependence_tester_1): Likewise dra and drb.
20926 (build_classic_dist_vector): Update calls accordingly.
20927 (subscript_dependence_tester): Likewise.
20928 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20929 DDR_COULD_BE_INDEPENDENT_P.
20930 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20931 comp_alias_ddrs instead of may_alias_ddrs.
20932 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20934 (vect_analyze_data_ref_dependence): Use it if
20935 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20936 distance vectors if that fails.
20937 (dependence_distance_ge_vf): New function.
20938 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20939 LOOP_VINFO_MAY_ALIAS_DDRS.
20941 2017-08-04 Richard Biener <rguenther@suse.de>
20943 PR middle-end/81705
20944 * fold-const.c (fold_binary_loc): Properly restrict
20945 minus_var0 && minus_var1 case when associating undefined overflow
20948 2017-08-04 Tom de Vries <tom@codesourcery.com>
20950 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20952 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20955 Don't start diagnostic messages with a capital letter.
20956 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20958 (rs6000_invalid_builtin): Likewise.
20959 (rs6000_trampoline_init): Likewise.
20961 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20964 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20965 after setting changeable df flags.
20967 2017-08-03 Richard Biener <rguenther@suse.de>
20969 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20970 up if the use is in USE - X.
20972 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20974 * toplev.c (dumpfile.h): New include.
20975 (internal_error_reentered): New static function. Use it...
20976 (internal_error_function): ...here to handle reentered internal_error.
20978 2017-08-03 Richard Biener <rguenther@suse.de>
20980 PR middle-end/81148
20981 * fold-const.c (split_tree): Add minus_var and minus_con
20982 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20983 here but always use minus_*.
20984 (associate_trees): Assert we never associate with MINUS_EXPR
20985 and NULL first operand. Do not recurse for PLUS_EXPR operands
20986 when associating as MINUS_EXPR either.
20987 (fold_binary_loc): Track minus_var and minus_con.
20989 2017-08-03 Tom de Vries <tom@codesourcery.com>
20992 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20993 ACCEL_COMPILER, apply finish_options on
20994 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20996 2017-08-03 Tom de Vries <tom@codesourcery.com>
20999 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21000 function_entry_patch_area_size > 0.
21002 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21005 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21006 instead of 10??LU, perform unit multiplication in wide_int,
21007 don't change alloc_object_size_limit if the limit is larger
21010 PR tree-optimization/81655
21011 PR tree-optimization/81588
21012 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21013 the case when ranges[i].low and high are 1 for unsigned type with
21016 PR middle-end/81052
21017 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21018 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21020 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21022 * tree-vrp.h: Add include guard.
21024 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21027 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21028 (ud2): New insn pattern.
21029 * config/i386/i386.c (ix86_expand_epilogue):
21030 For naked functions, generate ud2 instead of trap insn.
21032 2017-08-02 Marek Polacek <polacek@redhat.com>
21035 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21037 2017-08-02 Tom de Vries <tom@codesourcery.com>
21038 Cesar Philippidis <cesar@codesourcery.com>
21040 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21041 Add missing edge probabilities.
21043 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21045 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21046 Correct endianness.
21048 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21050 PR middle-end/79499
21051 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21052 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21053 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21054 prologue_seq sequences - if any.
21056 2017-08-02 Richard Biener <rguenther@suse.de>
21058 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21059 via vectors if supported, integer extracts via punning if supported
21060 or otherwise vector extracts.
21062 2017-08-02 Richard Biener <rguenther@suse.de>
21064 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21066 (bitmap_insert_into_set): ... this.
21068 2017-08-02 Richard Biener <rguenther@suse.de>
21070 PR tree-optimization/81633
21072 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21074 PR tree-optimization/71752
21075 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21077 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21079 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21080 (machine_function::call_ms2sysv_pad_out): Remove field.
21081 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21082 (ix86_compute_frame_layout): Likewise.
21084 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21087 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21088 attribute with interrupt attribute.
21090 2017-08-01 Andrew Pinski <apinski@cavium.com>
21092 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21093 BIT_INSERT_EXPR's operand 1
21094 to see if the types precision matches.
21096 2017-08-01 Martin Liska <mliska@suse.cz>
21098 PR middle-end/70140
21099 * builtins.c (expand_builtin_memcpy_args): Remove.
21100 (expand_builtin_memcpy): Call newly added function
21101 expand_builtin_memory_copy_args.
21102 (expand_builtin_memcpy_with_bounds): Likewise.
21103 (expand_builtin_mempcpy): Remove last argument.
21104 (expand_builtin_mempcpy_with_bounds): Likewise.
21105 (expand_builtin_memory_copy_args): New function created from
21106 expand_builtin_mempcpy_args with small modifications.
21107 (expand_builtin_mempcpy_args): Remove.
21108 (expand_builtin_stpcpy): Remove unused argument.
21109 (expand_builtin): Likewise.
21110 (expand_builtin_with_bounds): Likewise.
21112 2017-08-01 Martin Liska <mliska@suse.cz>
21115 Make mempcpy more optimal (PR middle-end/70140).
21117 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21120 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21121 __builtin_vec_cmpne verify both arguments are compatible vectors
21122 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21123 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21124 move computation of aligned to after checking the argument types.
21128 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21129 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21132 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21133 Doug Rupp <rupp@adacore.com>
21134 Olivier Hainque <hainque@adacore.com>
21136 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21137 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21139 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21140 for TARGET_OS_CPP_BUILTIN.
21141 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21142 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21144 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21145 passing required abi options to the assembler for EABI configurations.
21146 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21147 of .text.hot and .text.unlikely sections for kernel modules when
21148 using ARM style exceptions.
21149 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21150 options. Add EXTRA_CC1_SPEC.
21151 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21153 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21155 (ARM_TARGET2_DWARF_FORMAT): Define.
21156 * config/arm/t-vxworks: Adjust multilib control to removal of the
21157 Diab command line options.
21159 2017-08-01 Martin Liska <mliska@suse.cz>
21161 PR gcov-profile/81561
21162 * gcov.c (unblock): Make unblocking safe as we need to preserve
21163 index correspondence of blocks and block_lists.
21165 2017-08-01 Richard Biener <rguenther@suse.de>
21167 PR tree-optimization/81181
21168 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21169 (compute_antic): ... end of iteration here.
21171 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21173 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21174 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21175 (ftree-slp-vectorize): Likewise.
21176 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21177 can no longer be set independent of flag_tree_loop_vectorize.
21178 * omp-general.c (emp_max_vf): Likewise.
21179 * opts.c (enable_fdo_optimizations): Remove references to
21180 flag_tree_vectorize, these are now implicit.
21181 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21182 and leave it for the options machinery.
21184 2017-08-01 Martin Liska <mliska@suse.cz>
21186 PR middle-end/70140
21187 * builtins.c (expand_builtin_memcpy_args): Remove.
21188 (expand_builtin_memcpy): Call newly added function
21189 expand_builtin_memory_copy_args.
21190 (expand_builtin_memcpy_with_bounds): Likewise.
21191 (expand_builtin_mempcpy): Remove last argument.
21192 (expand_builtin_mempcpy_with_bounds): Likewise.
21193 (expand_builtin_memory_copy_args): New function created from
21194 expand_builtin_mempcpy_args with small modifications.
21195 (expand_builtin_mempcpy_args): Remove.
21196 (expand_builtin_stpcpy): Remove unused argument.
21197 (expand_builtin): Likewise.
21198 (expand_builtin_with_bounds): Likewise.
21200 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21203 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21204 print "ds:" only for immediates in generic address space.
21206 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21209 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21210 (ix86_function_ok_for_sibcall): Return false for naked functions.
21212 2017-08-01 Richard Biener <rguenther@suse.de>
21214 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21215 (compute_antic): Seed worklist with exit block predecessors.
21216 * cfganal.c (dfs_find_deadend): For a cycle return the source
21217 of the edge closing it.
21219 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21221 * config/aarch64/aarch64.c
21222 (aarch64_can_const_movi_rtx_p): Move 0 check.
21224 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21226 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21227 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21229 * match.pd: Ditto in address comparison pattern.
21231 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21233 PR tree-optimization/81627
21234 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21235 closed ssa form for store-store chain.
21237 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21239 PR tree-optimization/81620
21240 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21241 for store-store chain.
21243 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21245 PR tree-optimization/81588
21246 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21247 ranges[i].in_p, invert comparison code ccode. For >/>=,
21248 swap rhs1 and rhs2 and comparison code unconditionally,
21249 for </<= don't do that. Don't swap rhs1/rhs2 again if
21250 ranges[i].in_p, instead invert comparison code ccode if
21251 opcode or oe->rank is BIT_IOR_EXPR.
21254 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21255 a direct optab to conversion optab.
21256 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21257 with GET_MODE_INNER as last argument instead of optab_handler.
21258 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21259 vector extraction if possible and optab is available.
21260 * expr.c (store_constructor): Use convert_optab_handler instead
21261 of optab_handler. Use vector initialization from smaller
21262 vectors if possible and optab is available.
21263 * tree-vect-stmts.c (vectorizable_load): Likewise.
21264 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21265 now have two modes.
21266 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21267 of vec_init from half-sized vectors with the same element mode.
21268 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21269 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21270 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21271 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21272 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21273 after mode in gen_vec_extract* calls.
21274 (vec_extract<mode>): Renamed to ...
21275 (vec_extract<mode><ssescalarmodelower>): ... this.
21276 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21277 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21278 element mode after mode in gen_vec_init* calls.
21279 (VEC_INIT_HALF_MODE): New mode iterator.
21280 (vec_init<mode>): Renamed to ...
21281 (vec_init<mode><ssescalarmodelower>): ... this.
21282 (vec_init<mode><ssehalfvecmodelower>): New expander.
21283 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21284 (vec_extractv2sfsf): ... this.
21285 (vec_initv2sf): Renamed to ...
21286 (vec_initv2sfsf): ... this.
21287 (vec_extractv2si): Renamed to ...
21288 (vec_extractv2sisi): ... this.
21289 (vec_initv2si): Renamed to ...
21290 (vec_initv2sisi): ... this.
21291 (vec_extractv4hi): Renamed to ...
21292 (vec_extractv4hihi): ... this.
21293 (vec_initv4hi): Renamed to ...
21294 (vec_initv4hihi): ... this.
21295 (vec_extractv8qi): Renamed to ...
21296 (vec_extractv8qiqi): ... this.
21297 (vec_initv8qi): Renamed to ...
21298 (vec_initv8qiqi): ... this.
21299 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21300 (vec_init<mode>): Renamed to ...
21301 (vec_init<mode><VEC_base_l>): ... this.
21302 (vec_extract<mode>): Renamed to ...
21303 (vec_extract<mode><VEC_base_l>): ... this.
21304 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21305 (vec_initv2sfsf): ... this.
21306 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21307 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21308 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21309 element mode after mode in gen_vec_init* calls.
21310 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21311 (vec_init<mode><Vel>): ... this.
21312 (vec_extract<mode>): Renamed to ...
21313 (vec_extract<mode><Vel>): ... this.
21314 * config/aarch64/iterators.md (Vel): New mode attribute.
21315 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21316 Add element mode after mode in gen_vec_extract* calls.
21317 * config/s390/vector.md (non_vec_l): New mode attribute.
21318 (vec_extract<mode>): Renamed to ...
21319 (vec_extract<mode><non_vec_l>): ... this.
21320 (vec_init<mode>): Renamed to ...
21321 (vec_init<mode><non_vec_l>): ... this.
21322 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21323 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21325 * config/arm/iterators.md (V_elem_l): New mode attribute.
21326 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21327 (vec_extract<mode><V_elem_l>): ... this.
21328 (vec_extractv2di): Renamed to ...
21329 (vec_extractv2didi): ... this.
21330 (vec_init<mode>): Renamed to ...
21331 (vec_init<mode><V_elem_l>): ... this.
21332 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21333 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21334 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21335 Add element mode after gen_vec_extract* calls.
21336 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21337 (vec_init<mode><unitmode>): ... this.
21338 (vec_extract<mode>): Renamed to ...
21339 (vec_extract<mode><unitmode>): ... this.
21340 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21341 (vec_init<mode><unitmode>): ... this.
21342 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21343 (vec_initv2sfsf): ... this.
21344 (vec_extractv2sf): Renamed to ...
21345 (vec_extractv2sfsf): ... this.
21346 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21347 Add element mode after gen_vec_extract* calls.
21348 * config/mips/mips.md (unitmode): New mode iterator.
21349 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21350 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21351 * config/spu/spu.md (inner_l): New mode attribute.
21352 (vec_init<mode>): Renamed to ...
21353 (vec_init<mode><inner_l>): ... this.
21354 (vec_extract<mode>): Renamed to ...
21355 (vec_extract<mode><inner_l>): ... this.
21356 * config/sparc/sparc.md (veltmode): New mode iterator.
21357 (vec_init<VMALL:mode>): Renamed to ...
21358 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21359 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21360 (vec_initv2sisi): ... this.
21361 (vec_initv2sf): Renamed to ...
21362 (vec_initv2sfsf): ... this.
21363 (vec_extractv2sf): Renamed to ...
21364 (vec_extractv2sfsf): ... this.
21365 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21366 (vec_init<mode>): Renamed to ...
21367 (vec_init<mode><VEC_base_l>): ... this.
21368 (vec_extract<mode>): Renamed to ...
21369 (vec_extract<mode><VEC_base_l>): ... this.
21370 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21371 (vec_initv2sfsf): ... this.
21372 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21373 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21374 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21375 gen_vec_init* calls.
21377 2017-08-01 Richard Biener <rguenther@suse.de>
21379 PR tree-optimization/81297
21380 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21381 TREE_OVERFLOW from INTEGER_CSTs.
21383 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21385 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21387 2017-07-31 Carl Love <cel@us.ibm.com>
21389 * config/rs6000/rs6000-c: Add support for built-in functions
21390 vector signed char vec_xl_be (signed long long, signed char *);
21391 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21392 vector signed int vec_xl_be (signed long long, signed int *);
21393 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21394 vector signed long long vec_xl_be (signed long long, signed long long *);
21395 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21396 vector signed short vec_xl_be (signed long long, signed short *);
21397 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21398 vector double vec_xl_be (signed long long, double *);
21399 vector float vec_xl_be (signed long long, float *);
21400 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21401 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21402 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21404 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21405 (altivec_expand_builtin): Add switch statement to call
21406 altivec_expand_xl_be for each builtin.
21407 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21408 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21409 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21410 __builtin_vsx_le_be_v16qi.
21411 * doc/extend.texi: Update the built-in documentation file for the
21412 new built-in functions.
21414 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21417 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21419 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21421 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21423 * config.gcc: Add z14.
21424 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21425 CPU model numbers for z13s and z14.
21426 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21428 * config/s390/s390-opts.h (enum processor_type): Rename
21429 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21430 * config/s390/s390.c (processor_table): Add field for CPU name to
21431 be passed to Binutils.
21432 (s390_asm_output_machine_for_arch): Use the new field in
21433 processor_table for Binutils.
21434 (s390_expand_builtin): Replace arch12 with z14.
21435 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21436 (s390_get_sched_attrmask): Likewise.
21437 (s390_get_unit_mask): Likewise.
21438 * config/s390/s390.opt: Add z14 to processor_type enum.
21440 2017-07-31 Martin Jambor <mjambor@suse.cz>
21443 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21444 regardless of optimization level.
21446 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21447 Martin Liska <mliska@suse.cz>
21449 * predict.def: Remove old comment and adjust probability.
21450 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21451 PREDICT statements.
21453 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21456 * config/i386/i386.c (ix86_function_naked): New function.
21457 (ix86_can_use_return_insn_p): Return false for naked functions.
21458 (ix86_expand_prologue): Skip prologue for naked functions.
21459 (ix86_expand_epilogue): Skip epilogue for naked functions
21460 and emit trap instruction.
21461 (ix86_warn_func_return): New function.
21462 (ix86_attribute_table): Add "naked" attribute specification.
21463 (TARGET_WARN_FUNC_RETURN): Define.
21464 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21466 2017-07-31 Martin Liska <mliska@suse.cz>
21468 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21469 (dump_gimple_bb_header): Always dump BB info.
21470 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21472 2017-07-31 Martin Liska <mliska@suse.cz>
21475 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21476 also with current_function_decl non-null equality.
21478 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21481 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21482 change type to the element type, instead add eltype variable and
21483 use it where we are interested in the element type.
21485 PR tree-optimization/81603
21486 * ipa-polymorphic-call.c
21487 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21488 offset arithmetic in offset_int, bail out if the resulting bit offset
21489 doesn't fit into shwi.
21491 2017-07-31 Martin Liska <mliska@suse.cz>
21493 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21494 (gimplify_save_expr): Fix comment.
21496 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21499 * config/i386/i386.c (ix86_function_arg): Update arguments for
21501 (ix86_compute_frame_layout): Set the initial stack offset to
21502 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21503 INCOMING_FRAME_SP_OFFSET.
21504 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21505 stack before exception handler returns.
21506 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21507 the 'ERROR_CODE' for exception handler.
21509 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21511 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21512 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21513 (ASM_OUTPUT_REG_POP): Ditto.
21514 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21515 instead of asm_fprintf to output pure string.
21517 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21519 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21520 to imported_module_or_decl hook.
21521 (debug_nothing_tree_tree_tree_bool): Remove.
21522 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21523 * debug.c (do_nothing_debug_hooks): Use
21524 debug_nothing_tree_tree_tree_bool_bool instead of
21525 debug_nothing_tree_tree_tree_bool.
21526 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21527 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21528 * sdbout.c (sdb_debug_hooks): Likewise.
21529 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21530 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21532 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21533 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21534 attribute, don't add anything.
21536 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21538 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21539 (fold_build2_stat_loc): Likewise.
21540 (fold_build3_stat_loc): Likewise.
21541 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21542 (fold_build1_loc): Remove macro.
21543 (fold_build2_loc): Likewise.
21544 (fold_build3_loc): Likewise.
21546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21548 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21549 (gimple_build_debug_bind_source_stat): Likewise.
21550 * gimple.h (gimple_build_debug_bind): Remove macro.
21551 (gimple_build_debug_bind_source): Likewise.
21553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21555 * bitmap.c (bitmap_alloc): Adjust.
21556 (bitmap_gc_alloc): Likewise.
21557 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21561 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21562 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21563 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21564 (bitmap_gc_alloc_stat): Likewise.
21565 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21569 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21570 * rtl.h (shallow_copy_rtx): Remove macro.
21572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21574 * emit-rtl.c (gen_raw_REG): Adjust.
21575 * gengenrtl.c (gendef): Likewise.
21576 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21577 * rtl.h (rtx_alloc): Remove macro.
21579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21581 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21582 (build_tree_list_stat): Likewise.
21583 * tree.h (build_tree_list): Remove macro.
21584 (build_tree_list_vec): Likewise.
21586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21588 * tree.c (make_vector_stat): Remove _stat from name.
21589 (build_vector_stat): Likewise.
21590 * tree.h (make_vector_stat): Remove macro.
21591 (build_vector_stat): Likewise.
21593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21595 * tree.h (build_var_debug_value): Remove prototype.
21597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21599 * tree.c (tree_cons_stat): Remove _stat from name.
21600 * tree.h (tree_cons): Remove macro.
21602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21604 * tree.c (build_vl_exp_stat): Remove _stat from name.
21605 * tree.h (build_vl_exp): Remove macro.
21607 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21609 * tree.c (build_decl_stat): Remove _stat from name.
21610 * tree.h (build_decl): Remove macro.
21612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21614 * gimple.c (gimple_build_with_ops_stat): Adjust.
21615 (gimple_alloc_stat): Remove _stat from name.
21616 * gimple.h (gimple_alloc): Remove macro.
21618 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21620 * tree.c (make_tree_vec_stat): Remove _stat from name.
21621 (grow_tree_vec_stat): Likewise.
21622 * tree.h (make_tree_vec_stat): Adjust prototype.
21623 (grow_tree_vec_stat): Likewise.
21624 (make_tree_vec): Remove macro.
21625 (grow_tree_vec): Likewise.
21627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21629 * fold-const.c (fold_build1_stat_loc): Adjust.
21630 (fold_build2_stat_loc): Likewise.
21631 (fold_build3_stat_loc): Likewise.
21632 * tree.c (build0_stat): Remove _stat from name.
21633 (build1_stat): Likewise.
21634 (build2_stat): Likewise.
21635 (build3_stat): Likewise.
21636 (build4_stat): Likewise.
21637 (build5_stat): Likewise.
21638 * tree.h (build1_loc): Remove macro, and rename _stat function
21640 (build2_loc): Likewise.
21641 (build3_loc): Likewise.
21642 (build4_loc): Likewise.
21643 (build5_loc): Likewise.
21645 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21647 * tree.c (make_int_cst_stat): Remove _stat from name.
21648 * tree.h (make_int_cst_stat): Adjust prototype.
21649 (make_int_cst): Remove macro.
21651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21653 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21654 * tree.h (make_tree_binfo_stat): Adjust prototype.
21655 (make_tree_binfo): Remove.
21657 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21659 * tree.c (copy_node_stat): Rename to copy_node.
21660 (build_distinct_type_copy): Adjust.
21661 * tree.h (copy_node_stat): Adjust prototype.
21662 (copy_node): Remove macro.
21664 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21666 * tree.c (make_node_stat): rename to make_node.
21667 (build_tree_list_stat): Adjust.
21668 (build0_stat): Likewise.
21669 (build2_stat): Likewise.
21670 (build3_stat): Likewise.
21671 (build4_stat): Likewise.
21672 (build5_stat): Likewise.
21673 (build_decl_stat): Likewise.
21674 * tree.h (make_node_stat): Adjust prototype.
21675 (make_node): remove macro.
21677 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21679 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21680 (PPC_FEATURE2_SCV): Likewise.
21681 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21683 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21685 * config/aarch64/aarch64.c
21686 (aarch64_internal_mov_immediate): Add new special pattern.
21687 * config/aarch64/aarch64.md (*movdi_aarch64):
21688 Add reg/32bit const mov case.
21690 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21691 Richard Sandiford <richard.sandiford@linaro.org>
21693 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21694 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21695 Add integer and movi cases.
21696 (movi-split-hf-df-sf split, fp16): New.
21697 (enabled): Added TARGET_FP_F16INST.
21698 * config/aarch64/iterators.md (GPF_HF): New.
21699 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21701 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21703 * config/aarch64/aarch64.c
21704 (aarch64_simd_container_mode): Add prototype.
21705 (aarch64_expand_mov_immediate): Add HI support.
21706 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21707 (aarch64_can_const_movi_rtx_p): New.
21708 (aarch64_preferred_reload_class):
21709 Remove restrictions of using FP registers for certain SIMD operations.
21710 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21711 (aarch64_valid_floating_const): Add integer move validation.
21712 (aarch64_simd_imm_scalar_p): Remove.
21713 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21714 (aarch64_legitimate_constant_p): Expand list of supported cases.
21715 * config/aarch64/aarch64-protos.h
21716 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21717 (aarch64_reinterpret_float_as_int): New.
21718 (aarch64_simd_imm_scalar_p): Remove.
21719 * config/aarch64/constraints.md (Uvi): New.
21720 (Dd): Split into Ds and new Dd.
21721 * config/aarch64/aarch64.md (*movsi_aarch64):
21723 (*movdi_aarch64): Add SIMD mov case.
21725 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21727 * tree-predcom.c: (struct chain): Handle store-store chain in which
21728 stores for elimination only store loop invariant values.
21729 (execute_pred_commoning_chain): Ditto.
21730 (prepare_initializers_chain_store_elim): Ditto.
21731 (prepare_finalizers): Ditto.
21732 (is_inv_store_elimination_chain): New function.
21733 (initialize_root_vars_store_elim_1): New function.
21735 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21737 * tree-predcom.c: Revise general description of the pass.
21738 (enum chain_type): New enum type for store elimination.
21739 (struct chain): New field supporting store elimination.
21740 (struct component): Ditto.
21741 (dump_chain): Dump store-stores chain.
21742 (release_chain): Release resources.
21743 (split_data_refs_to_components): Compute and create component
21744 contains only stores for elimination.
21745 (get_chain_last_ref_at): New function.
21746 (make_invariant_chain): Initialization.
21747 (make_rooted_chain): Specify chain type in parameter and record it.
21748 (add_looparound_copies): Skip for store-stores chain.
21749 (determine_roots_comp): Compute type of chain and pass it to
21751 (initialize_root_vars_store_elim_2): New function.
21752 (finalize_eliminated_stores): New function.
21753 (remove_stmt): Handle store for elimination.
21754 (execute_pred_commoning_chain): Execute predictive commoning on
21755 store-store chains.
21756 (determine_unroll_factor): Skip unroll for store-stores chain.
21757 (prepare_initializers_chain_store_elim): New function.
21758 (prepare_initializers_chain): Hanlde store-store chain.
21759 (prepare_finalizers_chain, prepare_finalizers): New function.
21760 (tree_predictive_commoning_loop): Return integer value indicating
21761 if loop is unrolled or lcssa form is corrupted.
21762 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21764 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21766 * tree-predcom.c (initialize_root): Delete.
21767 (execute_pred_commoning_chain): Initialize root vars and replace
21768 reference of non-combined chain directly, rather than call above
21771 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21773 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21774 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21776 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21778 * tree-predcom.c (struct chain): New field init_seq.
21779 (release_chain): Release init_seq.
21780 (prepare_initializers_chain): Record intialization stmts in above
21782 (insert_init_seqs): New function.
21783 (tree_predictive_commoning_loop): Call insert_init_seqs.
21785 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21787 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21789 2017-07-28 Richard Biener <rguenther@suse.de>
21791 * match.pd: Remove superfluous :c.
21792 * genmatch.c (simplify::id): Add member.
21793 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21795 (current_id): New global.
21796 (dt_node::parent): Move from ...
21797 (dt_operand::parent): ... here. Add for_id member.
21798 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21799 (decision_tree::find_node): Relax order requirement when
21800 merging DT_TRUE nodes to ones inbetween the current simplify
21801 and the one we try to merge with. Add diagnostic whenever
21802 we need to enforce pattern order by not merging.
21803 (decision_tree::insert): Set current_id.
21804 (decision_tree::print_node): Dump parent node and for_id.
21805 (parser::last_id): Add member.
21806 (parser::push_simplify): Assign unique id.
21807 (parser::parser): Initialize last_id.
21809 2017-07-28 Martin Liska <mliska@suse.cz>
21812 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21813 gimple_build_debug_bind.
21815 2017-07-28 Richard Biener <rguenther@suse.de>
21817 PR tree-optimization/81502
21818 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21820 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21822 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21823 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21824 for BIT_FIELD_REF args.
21825 * fold-const.c (make_bit_field_ref): Likewise.
21826 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21828 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21831 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21832 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21833 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21834 Or it into SANITIZER_UNDEFINED.
21835 * ubsan.c: Include gimple-fold.h and varasm.h.
21836 (ubsan_expand_ptr_ifn): New function.
21837 (instrument_pointer_overflow): New function.
21838 (maybe_instrument_pointer_overflow): New function.
21839 (instrument_object_size): Formatting fix.
21840 (pass_ubsan::execute): Call instrument_pointer_overflow
21841 and maybe_instrument_pointer_overflow.
21842 * internal-fn.c (expand_UBSAN_PTR): New function.
21843 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21844 * sanitizer.def (__ubsan_handle_pointer_overflow,
21845 __ubsan_handle_pointer_overflow_abort): New builtins.
21846 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21847 * internal-fn.def (UBSAN_PTR): New internal function.
21848 * opts.c (sanitizer_opts): Add pointer-overflow.
21849 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21850 * fold-const.c (build_range_check): Compute pointer range check in
21851 integral type if pointer arithmetics would be needed. Formatting
21854 2017-07-28 Martin Liska <mliska@suse.cz>
21857 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21858 parameters that are of a variable-length.
21860 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21862 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21864 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21865 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21866 (CRT_CALL_STATIC_FUNCTION): Likewise.
21867 (ASM_DEFAULT_SPEC): New define.
21868 (ASM_SPEC32): Likewise.
21869 (ASM_SPEC64): Likewise.
21870 (ASM_SPEC_COMMON): Likewise.
21871 (ASM_SPEC): Likewise.
21872 (INVALID_64BIT): Likewise.
21873 (LINK_OS_DEFAULT_SPEC): Likewise.
21874 (LINK_OS_SPEC32): Likewise.
21875 (LINK_OS_SPEC64): Likewise.
21876 (POWERPC_LINUX): Likewise.
21877 (PTRDIFF_TYPE): Likewise.
21878 (RESTORE_FP_PREFIX): Likewise.
21879 (RESTORE_FP_SUFFIX): Likewise.
21880 (SAVE_FP_PREFIX): Likewise.
21881 (SAVE_FP_SUFFIX): Likewise.
21882 (SIZE_TYPE): Likewise.
21883 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21884 (TARGET_64BIT): Likewise.
21885 (TARGET_64BIT): Likewise.
21886 (TARGET_AIX): Likewise.
21887 (WCHAR_TYPE_SIZE): Likewise.
21888 (WCHAR_TYPE): Undefine.
21889 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21890 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21891 (CPP_OS_RTEMS_SPEC): Delete.
21892 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21893 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21895 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21897 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21899 PR tree-optimization/81578
21900 * tree-parloops.c (build_new_reduction): Bail out if
21901 reduction_code isn't one of the standard OpenMP reductions.
21902 Move the details printing after that decision.
21904 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21906 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21907 related to reload_in_progress.
21908 (splat_input_operand): Likewise.
21909 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21911 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21913 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21914 (TARGET_INSTANTIATE_DECLS): Likewise.
21915 (legitimate_indexed_address_p): Delete reload_in_progress code.
21916 (rs6000_debug_legitimate_address_p): Likewise.
21917 (rs6000_eliminate_indexed_memrefs): Likewise.
21918 (rs6000_emit_le_vsx_store): Likewise.
21919 (rs6000_emit_move_si_sf_subreg): Likewise.
21920 (rs6000_emit_move): Likewise.
21921 (register_to_reg_type): Likewise.
21922 (rs6000_pre_atomic_barrier): Likewise.
21923 (rs6000_machopic_legitimize_pic_address): Likewise.
21924 (rs6000_allocate_stack_temp): Likewise.
21925 (rs6000_address_for_fpconvert): Likewise.
21926 (rs6000_address_for_altivec): Likewise.
21927 (rs6000_secondary_memory_needed_rtx): Delete function.
21928 (rs6000_check_sdmode): Likewise.
21929 (rs6000_alloc_sdmode_stack_slot): Likewise.
21930 (rs6000_instantiate_decls): Likewise.
21931 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21932 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21933 Delete reload_in_progress.
21934 (*vec_reload_and_plus_<mptrsize>): Likewise.
21935 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21936 (vsx_div_v2di): Likewise.
21937 (vsx_udiv_v2di): Likewise.
21939 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21941 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21942 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21943 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21944 (rs6000_debug_reg_global): Delete print of LRA status.
21945 (rs6000_option_override_internal): Delete dead LRA related code.
21946 (rs6000_lra_p): Delete function.
21947 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21949 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21951 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21952 * config/riscv/rtems.h: New file.
21954 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21955 Sudakshina Das <sudi.das@arm.com>
21957 * config/aarch64/aarch64.md
21958 (define_split for and<mode>3nr_compare): Move
21959 non aarch64_logical_operand to a register.
21960 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21961 register immediate operand to a register.
21962 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21964 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21966 PR middle-end/81564
21967 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21969 2017-07-27 Richard Biener <rguenther@suse.de>
21971 PR tree-optimization/81573
21972 PR tree-optimization/81494
21973 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21974 multi defuse cycle case.
21976 2017-07-27 Richard Biener <rguenther@suse.de>
21978 PR tree-optimization/81571
21979 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21982 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21984 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21985 earlier and only if MASK_FPU is set. Adjust formatting.
21987 2017-07-27 Martin Liska <mliska@suse.cz>
21989 * opt-functions.awk: Add validation of value of Init.
21990 * optc-gen.awk: Pass new argument.
21992 2017-07-27 Martin Liska <mliska@suse.cz>
21994 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21995 Fix wrong condition.
21997 2017-07-27 Martin Liska <mliska@suse.cz>
21999 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22000 BBs and edges seen by autoFDO.
22002 2017-07-27 Richard Biener <rguenther@suse.de>
22004 PR tree-optimization/81502
22005 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22006 with incompatible but same sized type.
22007 (execute_update_addresses_taken): Likewise.
22009 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22011 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22012 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22014 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22017 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22018 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22019 Change s_operand to memory_operand.
22021 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22023 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22024 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22025 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22026 Emit instructions rather than returning an expression. Handle TFmode
22027 and KFmode by casting to TImode.
22028 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22029 (rs6000_emit_le_vsx_store): Likewise.
22030 * config/rs6000/vsx.md (VSX_TI): New iterator.
22031 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22032 (*vsx_le_undo_permute_<mode>): Likewise.
22033 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22034 emit the split sequence.
22035 (*vsx_le_perm_store_<mode>): Likewise.
22037 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22039 PR tree-optimization/81555
22040 PR tree-optimization/81556
22041 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22042 if true, force CHANGED for the recursive invocation.
22043 (reassociate_bb): Remember original length of ops array, pass
22044 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22046 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22047 attributes for noipa attribute. For naked attribute use
22048 lookup_attribute first before lookup_attribute_spec.
22049 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22051 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22052 for functions with noipa attribute.
22053 (cgraph_externally_visible_p): Return true for functions with noipa
22055 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22056 for functions with noipa attribute.
22057 * doc/extend.texi: Document noipa function attribute.
22058 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22059 also for functions with noipa attribute.
22060 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22062 2017-07-26 Andrew Pinski <apinski@cavium.com>
22064 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22065 vec_unalign_load_cost and vec_unalign_store_cost.
22067 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22069 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22070 -mvsx-small-integer option.
22071 (ISA_3_0_MASKS_IEEE): Likewise.
22072 (OTHER_VSX_VECTOR_MASKS): Likewise.
22073 (POWERPC_MASKS): Likewise.
22074 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22075 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22076 code, only testing for DImode being allowed in non-VSX floating
22078 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22079 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22081 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22082 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22083 TARGET_P8_VECTOR test.
22084 (rs6000_secondary_reload_simple_move): Likewise.
22085 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22086 since TARGET_P9_VECTOR was already tested.
22087 (rs6000_opt_masks): Remove -mvsx-small-integer.
22088 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22089 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22091 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22092 test for TARGET_VEXTRACTUB was used, and that uses
22094 (p9 extract splitter): Likewise.
22095 (vsx_extract_<mode>_di_p9): Likewise.
22096 (vsx_extract_<mode>_store_p9): Likewise.
22097 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22098 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22099 the elimination of TARGET_VSX_SMALL_INTEGER.
22100 (vsx_extract_<mode>_p8): Likewise.
22101 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22102 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22103 (vsx_set_<mode>_p9): Likewise.
22104 (vsx_set_v4sf_p9): Likewise.
22105 (vsx_set_v4sf_p9_zero): Likewise.
22106 (vsx_insert_extract_v4sf_p9): Likewise.
22107 (vsx_insert_extract_v4sf_p9_2): Likewise.
22108 * config/rs6000/rs6000.md (sign extend splitter): Change
22109 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22110 (floatsi<mode>2_lfiwax_mem): Likewise.
22111 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22112 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22113 since a test for TARGET_P9_VECTOR was used.
22114 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22115 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22116 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22117 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22118 TARGET_P8_VECTOR test.
22119 (fix_trunc<mode>si2_stfiwx): Likewise.
22120 (fix_trunc<mode>si2_internal): Likewise.
22121 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22122 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22124 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22125 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22126 TARGET_P8_VECTOR test.
22127 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22128 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22129 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22131 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22132 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22133 since a test for TARGET_P9_VECTOR was used.
22134 (splitter for loading small constants): Likewise.
22136 2017-07-26 Andrew Pinski <apinski@cavium.com>
22138 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22141 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22144 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22145 (fp_valid_at): Likewise.
22147 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22149 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22150 (qdf24xx_addrcost_table): Likewise.
22151 (cortexa57_tunings): Update to use generic_branch_cost.
22152 (cortexa72_tunings): Likewise.
22153 (cortexa73_tunings): Likewise.
22154 (qdf24xx_tunings): Likewise.
22156 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22158 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22159 (thunderx2t99_branch_cost): Likewise.
22160 (cortexa35_tunings): Update to use generic_branch_cost.
22161 (cortexa53_tunings): Likewise.
22162 (cortexa57_tunings): Likewise.
22163 (cortexa72_tunings): Likewise.
22164 (cortexa73_tunings): Likewise.
22165 (thunderx2t99_tunings): Likewise.
22167 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22169 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22170 (sparc_option_override): Honour MASK_FSMULD.
22171 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22172 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22173 * config/sparc/sparc.opt (mfsmuld): New option.
22174 * doc/invoke.texi (mfsmuld): Document option.
22176 2017-07-26 Marek Polacek <polacek@redhat.com>
22178 PR middle-end/70992
22179 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22181 2017-07-26 Richard Biener <rguenther@suse.de>
22183 * gimple-match-head.c (do_valueize): Return OP if valueize
22185 (get_def): New helper to get at the def stmt of a SSA name
22186 if valueize allows.
22187 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22188 do_valueize to get at the def stmt.
22189 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22191 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22193 PR middle-end/46932
22194 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22196 2017-07-26 Martin Liska <mliska@suse.cz>
22199 * function.c (expand_function_start): Make expansion of
22200 nonlocal_goto_save_area after parm_birth_insn.
22202 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22204 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22205 from all CPU target flags enable members.
22207 2017-07-26 Richard Biener <rguenther@suse.de>
22209 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22210 (decision_tree::gen): Make 'type' const.
22211 (write_predicate): Likewise.
22213 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22216 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22217 (rs6000_option_override_internal): Likewise.
22218 (rs6000_expand_vector_set): Likewise.
22219 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22220 (TARGET_UPPER_REGS_SF): Likewise.
22221 (TARGET_UPPER_REGS_DI): Likewise.
22222 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22223 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22224 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22225 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22226 (Splitters for DI constants in Altivec registers): Likewise.
22227 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22228 (vsx_set_v4sf_p9): Likewise.
22229 (vsx_set_v4sf_p9_zero): Likewise.
22230 (vsx_insert_extract_v4sf_p9): Likewise.
22231 (vsx_insert_extract_v4sf_p9_2): Likewise.
22233 2017-07-25 Carl Love <cel@us.ibm.com>
22235 * doc/extend.texi: Update the built-in documentation file for the
22236 existing built-in functions
22237 vector signed char vec_cnttz (vector signed char);
22238 vector unsigned char vec_cnttz (vector unsigned char);
22239 vector signed short vec_cnttz (vector signed short);
22240 vector unsigned short vec_cnttz (vector unsigned short);
22241 vector signed int vec_cnttz (vector signed int);
22242 vector unsigned int vec_cnttz (vector unsigned int);
22243 vector signed long long vec_cnttz (vector signed long long);
22244 vector unsigned long long vec_cnttz (vector unsigned long long);
22246 2017-07-25 Andrew Pinski <apinski@cavium.com>
22248 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22249 accesses where the use is for the first operand of a BIT_INSERT.
22251 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22254 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22255 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22257 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22259 * config/i386/gstabs.h: Delete.
22260 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22262 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22264 * config/i386/i386.c (ix86_decompose_address): Do not check for
22265 register RTX when looking at index_reg or base_reg.
22266 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22268 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22270 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22271 to update EH info here.
22273 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22275 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22277 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22279 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22281 2017-07-25 Torsten Duwe <duwe@suse.de>
22283 * common.opt: Introduce -fpatchable-function-entry
22284 command line option, and its variables function_entry_patch_area_size
22285 and function_entry_patch_area_start.
22286 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22287 including a two-value parser.
22288 * target.def (print_patchable_function_entry): New target hook.
22289 * targhooks.h (default_print_patchable_function_entry): New function.
22290 * targhooks.c (default_print_patchable_function_entry): Likewise.
22291 * toplev.c (process_options): Switch off IPA-RA if
22292 patchable function entries are being generated.
22293 * varasm.c (assemble_start_function): Look at the
22294 patchable-function-entry command line switch and current
22295 function attributes and maybe generate NOP instructions by
22296 calling the print_patchable_function_entry hook.
22297 * doc/extend.texi: Document patchable_function_entry attribute.
22298 * doc/invoke.texi: Document -fpatchable_function_entry
22299 command line option.
22300 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22302 * doc/tm.texi: Re-generate.
22304 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22307 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22308 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22310 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22312 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22314 2017-07-25 Richard Biener <rguenther@suse.de>
22316 PR tree-optimization/81455
22317 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22318 not walk in cycles when looking for guards.
22320 2017-07-25 Richard Biener <rguenther@suse.de>
22322 PR tree-optimization/81529
22323 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22324 when optimizing backedge uses.
22326 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22328 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22330 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22331 to dl_section_ref. On AIX, append an expression to subtract
22332 the size of the section length to dl_section_ref.
22334 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22336 * configure.ac: If any of the config.* scripts fail, exit 1.
22337 * configure: Regenerate.
22339 2017-07-25 Richard Biener <rguenther@suse.de>
22341 PR middle-end/81546
22342 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22343 of immediate uses, be more verbose on errors.
22345 2017-07-25 Richard Biener <rguenther@suse.de>
22347 PR tree-optimization/81510
22348 * tree-vect-loop.c (vect_is_simple_reduction): When the
22349 reduction stmt is not inside the loop bail out.
22351 2017-07-25 Richard Biener <rguenther@suse.de>
22353 PR tree-optimization/81303
22354 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22355 profitability check against LOOP_VINFO_NITERSM1.
22357 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22359 * domwalk.c (cmp_bb_postorder): Simplify.
22360 (sort_bbs_postorder): New function. Use it...
22361 (dom_walker::walk): ...here to optimize common cases.
22363 2017-07-25 Martin Liska <mliska@suse.cz>
22366 * ipa-visibility.c (function_and_variable_visibility): Make the
22367 redirection just on target that supports aliasing.
22368 Fix GNU coding style.
22370 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22373 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22375 * config/arm/rtems.h: Likewise.
22376 * config/bfin/rtems.h: Likewise.
22377 * config/i386/rtemself.h: Likewise.
22378 * config/lm32/rtems.h: Likewise.
22379 * config/m32c/rtems.h: Likewise.
22380 * config/m68k/rtemself.h: Likewise.
22381 * config/microblaze/rtems.h: Likewise.
22382 * config/mips/rtems.h: Likewise.
22383 * config/moxie/rtems.h: Likewise.
22384 * config/nios2/rtems.h: Likewise.
22385 * config/powerpcspe/rtems.h: Likewise.
22386 * config/rs6000/rtems.h: Likewise.
22387 * config/rtems.h: Likewise.
22388 * config/sh/rtems.h: Likewise.
22389 * config/sh/rtemself.h: Likewise.
22390 * config/sparc/rtemself.h: Likewise.
22392 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22395 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22396 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22397 * tree-ssa-structalias.c (alias_get_name): Same.
22399 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22402 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22403 instructions if no du chain is found.
22405 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22407 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22409 2017-07-25 Richard Biener <rguenther@suse.de>
22411 PR middle-end/81505
22412 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22415 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22417 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22418 upper-regs options.
22419 (ISA_2_7_MASKS_SERVER): Likewise.
22420 (ISA_3_0_MASKS_IEEE): Likewise.
22421 (OTHER_P8_VECTOR_MASKS): Likewise.
22422 (OTHER_VSX_VECTOR_MASKS): Likewise.
22423 (POWERPC_MASKS): Likewise.
22424 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22425 duplicate list of options.
22426 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22427 explicit -mupper-regs options.
22428 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22429 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22430 alias for -mupper-regs-df.
22431 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22432 (rs6000_init_hard_regno_mode_ok): Likewise.
22433 (rs6000_option_override_internal): Likewise.
22434 (rs6000_opt_masks): Likewise.
22435 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22436 options in terms of whether -mvsx or -mpower8-vector was used.
22437 (TARGET_UPPER_REGS_DI): Likewise.
22438 (TARGET_UPPER_REGS_SF): Likewise.
22439 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22440 -mupper-regs-* options.
22442 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22444 * passes.c (emergency_dump_function): Print some empty lines and a
22445 header before the RTL dump.
22447 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22449 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22451 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22454 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22455 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22457 2017-07-24 Carl Love <cel@us.ibm.com>
22459 * config/rs6000/rs6000-c.c: Add support for built-in functions
22460 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22461 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22462 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22463 vec_extract_fp_from_shortl): Add defines for the two builtins.
22464 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22465 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22467 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22468 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22469 * doc/extend.texi: Update the built-in documentation file for the
22470 new built-in function.
22472 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22475 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22477 * doc/generic.texi: Likewise.
22478 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22479 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22481 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22483 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22484 (aarch64_mls_elt_merge<mode>): Likewise.
22486 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22488 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22489 having __cxa_atexit.
22491 2017-07-23 Michael Collison <michael.collison@arm.com>
22493 * config/arm/arm.c (arm_option_override): Deprecate
22494 use of -mstructure-size-boundary.
22495 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22496 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22498 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22501 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22502 Reduce cost estimate for direct moves.
22504 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22507 * config/i386/i386.c (ix86_option_override_internal): Disable
22508 BMI, BMI2 and TBM instructions for -m16.
22510 2017-07-21 Carl Love <cel@us.ibm.com>
22512 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22513 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22514 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22515 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22516 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22517 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22518 VMULOSW): New enum "unspec" values.
22519 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22520 altivec_vmulosw): New patterns.
22521 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22522 VMULOSW): Add definitions.
22524 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22526 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22527 (qdf24xx): Likewise.
22528 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22529 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22530 (AARCH64_FL_V8_1): Renumber.
22531 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22532 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22533 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22534 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22535 rdma to feature modifiers list.
22537 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22539 PR middle-end/56727
22540 * ipa-visibility (function_and_variable_visibility): Convert
22541 recursive PLT call to direct call if appropriate.
22543 2017-07-21 Andrew Pinski <apinski@cavium.com>
22545 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22546 operand 1 to see if the types precision matches.
22547 * fold-const.c (operand_equal_p): Likewise.
22549 2017-07-21 Richard Biener <rguenther@suse.de>
22551 PR tree-optimization/81303
22552 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22553 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22554 (vect_peeling_hash_get_lowest_cost): Adjust.
22555 (vect_enhance_data_refs_alignment): Likewise. Use
22556 vect_get_peeling_costs_all_drs to compute the penalty for no
22557 peeling to match up costs.
22559 2017-07-21 Richard Biener <rguenther@suse.de>
22561 PR tree-optimization/81500
22562 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22563 we didn't identify a reduction path.
22565 2017-07-21 Tom de Vries <tom@codesourcery.com>
22566 Cesar Philippidis <cesar@codesourcery.com>
22568 PR gcov-profile/81442
22569 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22572 2017-07-21 Tom de Vries <tom@codesourcery.com>
22575 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22577 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22578 nvptx_override_options_after_change.
22580 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22582 * dwarf2out.c (output_file_names): Avoid double testing for
22583 dwarf_version >= 5.
22585 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22587 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22589 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22591 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22593 (try_crossjump_to_edge): Do not punt on partitioned functions.
22595 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22597 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22598 Put all BBs reachable only via paths crossing cold region to cold
22600 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22602 2016-07-21 Richard Biener <rguenther@suse.de>
22604 PR tree-optimization/81303
22605 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22606 into account prologue and epilogue iterations when raising
22607 min_profitable_iters to sth at least covering one vector iteration.
22609 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22611 * config/arm/arm.c (arm_test_cpu_arch_dat):
22614 2017-07-20 Nathan Sidwell <nathan@acm.org>
22616 Remove TYPE_METHODS.
22617 * tree.h (TYPE_METHODS): Delete.
22618 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22619 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22620 (dbxout_type_methods): Scan TYPE_FIELDS.
22621 (dbxout_type): Don't check TYPE_METHODS here.
22622 * function.c (use_register_for_decl): Always ignore register for
22623 class types when not optimizing.
22624 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22625 * tree.c (free_lang_data_in_type): Stitch out member functions and
22626 templates from TYPE_FIELDS.
22627 (build_distinct_type_copy, verify_type_variant,
22628 verify_type): Member fns are on TYPE_FIELDS.
22629 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22630 * tree-pretty-print.c (dump_generic_node): Likewise.
22632 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22635 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22636 V2TImode and V4TImode.
22637 (ix86_expand_vector_extract): Likewise.
22638 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22639 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22640 (ssescalarmode): Handle V4TImode and V2TImode.
22641 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22642 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22643 (VEXTRACTI128_MODE): New mode iterator.
22644 (splitter for *vec_extractv?ti first element): New.
22645 (VEC_INIT_MODE): New mode iterator.
22646 (vec_init<mode>): Consolidate 3 expanders into one using
22647 VEC_INIT_MODE mode iterator.
22649 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22651 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22652 non_spilled_static_chain_regno_p.
22654 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22656 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22658 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22660 * bb-reorder.c (connect_traces): Allow copying of blocks within
22663 2017-07-20 Richard Biener <rguenther@suse.de>
22665 * gimple.h (gimple_phi_result): Add gphi * overload.
22666 (gimple_phi_result_ptr): Likewise.
22667 (gimple_phi_arg): Likewise. Adjust index assert to only
22668 allow actual argument accesses rather than all slots available
22670 (gimple_phi_arg_def): Add gphi * overload.
22671 * tree-phinodes.c (make_phi_node): Initialize only actual
22673 (resize_phi_node): Clear memory not covered by old node,
22674 do not initialize excess argument slots.
22675 (reserve_phi_args_for_new_edge): Initialize new argument slot
22678 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22680 PR tree-optimization/81388
22682 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22684 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22685 by removing computation of may_be_zero.
22687 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22688 Tom de Vries <tom@codesourcery.com>
22690 PR middle-end/81030
22691 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22692 when gimple level profile disagrees with what RTL expander did.
22694 2017-07-20 Richard Biener <rguenther@suse.de>
22696 PR tree-optimization/61171
22697 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22698 (vect_analyze_stmt): Add slp instance parameter.
22699 (vectorizable_reduction): Likewise.
22700 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22701 (vect_is_simple_reduction): Deal with chains not detected
22702 as SLP reduction chain, specifically not properly associated
22703 chains containing a mix of plus/minus.
22704 (get_reduction_op): Remove.
22705 (get_initial_defs_for_reduction): Simplify, pass in whether
22706 this is a reduction chain, pass in the SLP node for the PHIs.
22707 (vect_create_epilog_for_reduction): Get the SLP instance as
22709 (vectorizable_reduction): Get the SLP instance as arg.
22710 During analysis remember the SLP node with the PHIs in the
22711 instance. Simplify getting at the vectorized reduction PHIs.
22712 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22713 through SLP instance.
22714 (vect_slp_analyze_operations): Likewise.
22715 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22716 (vect_transform_stmt): Likewise.
22718 2017-07-20 Tom de Vries <tom@codesourcery.com>
22720 PR tree-optimization/81489
22721 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22722 read of phi arg location to before loop that modifies phi.
22724 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22726 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22729 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22731 PR middle-end/81331
22732 * except.c (execute): Fix ordering issue.
22734 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22736 PR rtl-optimization/81423
22737 * combine.c (make_compound_operation_int): Don't try to optimize
22738 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22740 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22742 PR rtl-optimization/81423
22743 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22744 with a constant that is -1 in the truncated to mode.
22746 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22748 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22749 (determine_unlikely_bbs): ... here.
22750 * predict.h (propagate_unlikely_bbs_forward): Declare.
22751 * cfgexpand.c (pass_expand::execute): Use it.
22752 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22754 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22755 propagate_unlikely_bbs_forward.
22757 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22759 PR middle-end/81331
22760 * except.c (maybe_add_nop_after_section_switch): New function.
22763 2017-07-19 Tom de Vries <tom@codesourcery.com>
22765 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22767 2017-07-19 Tom de Vries <tom@codesourcery.com>
22769 * gimple.h (gimple_phi_arg): Make assert more strict.
22771 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22773 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22774 * config/rs6000/mmintrin.h: New file.
22775 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22777 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22779 PR tree-optimization/81346
22780 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22782 2017-07-19 Tom de Vries <tom@codesourcery.com>
22784 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22786 2017-07-19 Tom de Vries <tom@codesourcery.com>
22788 * config/nvptx/nvptx-modes.def: Add V2DImode.
22789 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22790 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22791 (nvptx_output_mov_insn): Handle lack of mov.b128.
22792 (nvptx_print_operand): Handle 'H' and 'L' codes.
22793 (nvptx_vector_mode_supported): Allow V2DImode.
22794 (nvptx_preferred_simd_mode): New function.
22795 (nvptx_data_alignment): New function.
22796 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22797 nvptx_preferred_simd_mode.
22798 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22800 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22802 2017-07-19 Tom de Vries <tom@codesourcery.com>
22804 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22805 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22806 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22807 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22808 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22809 (mov<VECIM>_insn): New define_insn.
22810 (define_expand "mov<VECIM>): New define_expand.
22812 2017-07-19 Tom de Vries <tom@codesourcery.com>
22814 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22816 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22818 PR tree-optimization/81346
22819 * fold-const.h (fold_div_compare, range_check_type): Declare.
22820 * fold-const.c (range_check_type): New function.
22821 (build_range_check): Use range_check_type.
22822 (fold_div_compare): No longer static, rewritten into
22823 a match.pd helper function.
22824 (fold_comparison): Don't call fold_div_compare here.
22825 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22826 as helper function.
22828 2017-07-19 Nathan Sidwell <nathan@acm.org>
22830 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22831 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22832 * tree.c (find_decls_types_r, verify_type): Use
22833 TYPE_{MIN,MAX}_VALUE_RAW.
22834 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22835 (hash_tree): Likewise.
22836 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22838 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22841 2017-07-18 Tom de Vries <tom@codesourcery.com>
22843 PR middle-end/81464
22844 * omp-expand.c (expand_omp_for_static_chunk): Handle
22845 equal-argument loop exit phi.
22847 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22850 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22851 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22852 operand 2 predicate.
22853 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22854 operand 2 predicate.
22855 (ror,rol -> rorx splitters): Use const_int_operand as
22856 operand 2 predicate.
22858 2017-06-18 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/81410
22861 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22862 the gap in the ! slp_perm SLP case after each group.
22864 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22866 PR middle-end/81463
22867 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22870 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22872 PR middle-end/81462
22873 * predict.c (set_even_probabilities): Cleanup; do not affect
22874 probabilities that are already known.
22875 (combine_predictions_for_bb): Call even when count is set.
22877 2017-07-18 Nathan Sidwell <nathan@acm.org>
22879 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22882 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22885 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22886 optimization for loop niter analysis.
22888 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22891 * config/avr/avr.c (avr_optimize_casesi): Don't use
22892 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22894 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22896 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22897 body_cost_vec from _vect_peel_extended_info.
22898 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22899 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22902 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22904 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22906 2017-07-18 Richard Biener <rguenther@suse.de>
22908 PR tree-optimization/80620
22909 PR tree-optimization/81403
22910 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22911 info when re-using a VN table entry.
22913 2017-07-18 Richard Biener <rguenther@suse.de>
22915 PR tree-optimization/81418
22916 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22917 vectype_in. Verify that with lane-reducing reduction operations
22918 we have a single def-use cycle.
22920 2017-07-17 Carl Love <cel@us.ibm.com>
22922 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22924 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22925 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22926 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22927 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22928 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22929 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22930 VMULOSW): New enum "unspec" values.
22931 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22932 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22933 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22934 altivec_vmulosw): New patterns.
22935 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22936 VMULOSW): Add definitions.
22938 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22940 * config/alpha/alpha.c: Include predict.h.
22942 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22944 * tree-vrp.c (compare_assert_loc): Fix comparison function
22945 to return predictable results.
22947 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22949 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22951 (subdi3): Likewise.
22952 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22953 * doc/invoke.texi (mexpand-adddi): Update text.
22955 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22957 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22958 that also clobbers the CC register. The old expand code is moved
22960 (*arc_clzsi2): ... here.
22961 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22962 the CC register. The old expand code is moved to ...
22963 (arc_ctzsi2): ... here.
22965 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22967 * config/arc/arc.opt (mindexed-loads): Use initial value
22968 TARGET_INDEXED_LOADS_DEFAULT.
22969 (mauto-modify-reg): Use initial value
22970 TARGET_AUTO_MODIFY_REG_DEFAULT.
22971 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22972 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22973 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22974 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22976 2017-07-17 Martin Liska <mliska@suse.cz>
22979 * opts.c (finish_options): Do not allow -fgnu-tm
22980 w/ -fsanitize={kernel-,}address. Say sorry.
22982 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22985 * tree-loop-distribution.c (classify_partition): Only assert on
22986 numer of iterations.
22987 (merge_dep_scc_partitions): Delete prameter. Update function call.
22988 (distribute_loop): Remove code handling loop with unknown niters.
22989 (pass_loop_distribution::execute): Skip loop with unknown niters.
22991 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22994 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22995 function sort_partitions_by_post_order.
22997 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22999 PR tree-optimization/81374
23000 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23001 the max index of basic blocks, rather than number of basic blocks.
23003 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23005 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23007 (arc_legitimate_pic_operand_p): Likewise.
23008 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23010 (arc_needs_pcl_p): Likewise.
23011 (arc_legitimate_pc_offset_p): Likewise.
23012 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23013 function is also used in constrains.md.
23014 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23015 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23016 PLUS. Only return true/false in known cases, otherwise assert.
23017 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23018 is already called in arc_legitimate_constant_p.
23019 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23021 (LEGITIMATE_PIC_OPERAND_P): Use
23022 arc_raw_symbolic_reference_mentioned_p function.
23023 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23028 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23029 Andrew Burgess <andrew.burgess@embecosm.com>
23031 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23032 (arc_return_address_register): New function.
23033 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23034 (arc_handle_fndecl_attribute): Add naked attribute.
23035 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23036 (TARGET_WARN_FUNC_RETURN): Likewise.
23037 (arc_allocate_stack_slots_for_args): New function.
23038 (arc_warn_func_return): Likewise.
23039 (machine_function): Change type fn_type.
23040 (arc_compute_function_type): Consider new naked function type,
23041 change function return type.
23042 (arc_must_save_register): Adapt to handle new
23043 arc_compute_function_type's return type.
23044 (arc_expand_prologue): Likewise.
23045 (arc_expand_epilogue): Likewise.
23046 (arc_return_address_regs): Delete.
23047 (arc_return_address_register): New function.
23048 (arc_epilogue_uses): Use above function.
23049 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23050 (arc_function_type): Change encoding, add naked type.
23051 (ARC_INTERRUPT_P): Change to handle the new encoding.
23052 (ARC_FAST_INTERRUPT_P): Likewise.
23053 (ARC_NORMAL_P): Define.
23054 (ARC_NAKED_P): Likewise.
23055 (arc_compute_function_type): Delete prototype.
23056 * config/arc/arc.md (in_ret_delay_slot): Use
23057 arc_return_address_register function.
23058 (simple_return): Likewise.
23059 (p_return_i): Likewise.
23061 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23063 PR tree-optimization/81428
23064 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23065 can't be built for those types.
23067 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23069 Remove stuff dead since r239246.
23071 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23072 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23073 (avr_inform_devices): Remove dead stuff.
23075 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23077 * config/arm/arm_neon.h: Fix softp typo.
23079 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23081 PR tree-optimization/81365
23082 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23083 aggregate moves onto bb predecessor edges, make sure there are no
23084 loads that could alias the lhs in between the start of bb and the
23087 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23090 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23091 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23092 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23094 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23096 PR tree-optimization/81396
23097 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23098 (init_symbolic_number): Initialize it to 1.
23099 (perform_symbolic_merge): Add n_ops from both operands into the new
23101 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23102 without base_addr as useless if they need more than one operation.
23103 (bswap_replace): Handle !bswap case for NULL base_addr.
23105 2017-07-17 Tom de Vries <tom@codesourcery.com>
23108 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23111 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23113 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23114 conditional builtin define __FIX_LEON3FT_B2BST.
23116 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23118 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23119 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23122 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23124 PR rtl-optimization/81424
23125 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23126 to remove potential trapping from operands if -fnon-call-exceptions.
23128 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23130 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23131 profile_proability for scalling.
23132 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23136 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23138 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23140 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23141 fixpoint arithmetics.
23143 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23145 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23146 fixpoint arithmetics.
23148 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23150 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23151 fixpoint arithmetics.
23153 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23155 * profile-count.h (profile_probability::from_reg_br_prob_note,
23156 profile_probability::to_reg_br_prob_note): New functions.
23157 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23158 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23159 * predict.c (probability_reliable_p): Update.
23160 (edge_probability_reliable_p): Update.
23161 (br_prob_note_reliable_p): Update.
23162 (invert_br_probabilities): Update.
23163 (add_reg_br_prob_note): New function.
23164 (combine_predictions_for_insn): Update.
23165 * asan.c (asan_clear_shadow): Update.
23166 * cfgbuild.c (compute_outgoing_frequencies): Update.
23167 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23168 (update_br_prob_note): Update.
23169 (rtl_verify_edges): Update.
23170 (purge_dead_edges): Update.
23171 (fixup_reorder_chain): Update.
23172 * emit-rtl.c (try_split): Update.
23173 * ifcvt.c (cond_exec_process_insns): Update.
23174 (cond_exec_process_if_block): Update.
23175 (dead_or_predicable): Update.
23176 * internal-fn.c (expand_addsub_overflow): Update.
23177 (expand_neg_overflow): Update.
23178 (expand_mul_overflow): Update.
23179 * loop-doloop.c (doloop_modify): Update.
23180 * loop-unroll.c (compare_and_jump_seq): Update.
23181 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23182 * predict.h: Update.
23183 * reorg.c (mostly_true_jump): Update.
23185 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23186 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23187 * config/arc/arc.c: (emit_unlikely_jump): Update.
23188 * config/arm/arm.c: (emit_unlikely_jump): Update.
23189 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23190 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23191 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23192 (ix86_print_operand): Update.
23193 (ix86_split_fp_branch): Update.
23194 (predict_jump): Update.
23195 * config/ia64/ia64.c (ia64_print_operand): Update.
23196 * config/mmix/mmix.c (mmix_print_operand): Update.
23197 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23198 (rs6000_expand_split_stack_prologue): Update.
23199 * config/rs6000/rs6000.c: Update.
23200 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23201 (s390_expand_vec_movstr): Update.
23202 (s390_expand_cs_tdsi): Update.
23203 (s390_expand_split_stack_prologue): Update.
23204 * config/sh/sh.c (sh_print_operand): Update.
23205 (expand_cbranchsi4): Update.
23206 (expand_cbranchdi4): Update.
23207 * config/sparc/sparc.c (output_v9branch): Update.
23208 * config/spu/spu.c (get_branch_target): Update.
23209 (ea_load_store_inline): Update.
23210 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23211 * config/tilepro/tilepro.c: Update.
23213 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23215 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23216 (gimplify_save_expr): Likewise.
23218 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23220 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23222 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23224 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23226 * timevar.def (TV_IPA_FNSUMMARY): Define.
23228 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23230 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23231 to back store errata sensitive sequence from being generated.
23232 (sqrtdf2_fix): Likewise.
23234 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23236 * tree-ssa-threadupdate.c (compute_path_counts,
23237 update_joiner_offpath_counts): Use profile_probability.
23239 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23242 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23244 * config/arm/arm-c.c (arm_cpu_builtins): Define
23245 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23247 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23250 array entries to represent __ieee128 versions of the
23251 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23252 scalar_extract_sig, and scalar_insert_exp built-in functions.
23253 (altivec_resolve_overloaded_builtin): Add special case handling
23254 for the __builtin_scalar_insert_exp function, as represented by
23255 the P9V_BUILTIN_VEC_VSIEDP constant.
23256 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23257 exponent support for __ieee128 argument.
23258 (VSESQP): Add scalar extract signature support for __ieee128
23260 (VSTDCNQP): Add scalar test negative support for __ieee128
23262 (VSIEQP): Add scalar insert exponent support for __int128 argument
23263 with __ieee128 result.
23264 (VSIEQPF): Add scalar insert exponent support for __ieee128
23265 argument with __ieee128 result.
23266 (VSTDCQP): Add scalar test data class support for __ieee128
23268 (VSTDCNQP): Add overload support for scalar test negative with
23269 __ieee128 argument.
23270 (VSTDCQP): Add overload support for scalar test data class
23271 __ieee128 argument.
23272 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23273 UNSPEC_VSX_SXSIGDP.
23274 (UNSPEC_VSX_SIEXPQP): New constant.
23275 (xsxexpqp): New insn for VSX scalar extract exponent quad
23277 (xsxsigqp): New insn for VSX scalar extract significand quad
23279 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23280 precision with floating point argument.
23281 (xststdcqp): New expand for VSX scalar test data class quad
23283 (xststdcnegqp): New expand for VSX scalar test negative quad
23285 (xststdcqp): New insn to match expansions for VSX scalar test data
23286 class quad precision and VSX scalar test negative quad precision.
23287 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23288 special case operand checking to enforce that second operand of
23289 VSX scalar test data class with quad precision argument is a 7-bit
23291 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23292 prototypes and descriptions of __ieee128 versions of
23293 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23294 scalar_test_data_class, and scalar_test_neg built-in functions.
23296 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23298 PR tree-optimization/81162
23299 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23300 replace a negate with an add.
23302 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23304 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23306 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23308 * config/arm/arm-c.c (arm_cpu_builtins): Define
23309 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23311 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23313 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23314 (armv8-r): Set ARM Cortex-R52 as default CPU.
23315 * config/arm/arm-tables.opt: Regenerate.
23316 * config/arm/arm-tune.md: Regenerate.
23317 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23319 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23320 extension for -mcpu=cortex-r52.
23322 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23324 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23325 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23326 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23327 (fp-armv8): Define it as FP_ARMv8 only.
23328 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23329 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23331 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23332 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23333 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23334 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23335 than TARGET_FPU_ARMV8.
23336 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23337 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23338 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23340 * config/arm/neon.md (neon_vrint): Likewise.
23341 (neon_vcvt): Likewise.
23342 (neon_<fmaxmin_op><mode>): Likewise.
23343 (<fmaxmin><mode>3): Likewise.
23344 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23345 * config/arm/predicates.md (arm_cond_move_operator): Check against
23346 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23348 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23350 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23351 to top of function.
23353 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23355 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23356 loop in comment with memset.
23358 2017-07-14 Martin Liska <mliska@suse.cz>
23360 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23361 * dwarf2out.c (is_java): Remove the function.
23362 (output_pubname): Remove usage of the function.
23363 (lower_bound_default): Remove usage of DW_LANG_Java.
23364 (gen_compile_unit_die): Likewise.
23365 * gcc.c: Remove compiler defaults for .java and .zip files.
23366 * gimple-expr.c (remove_suffix): Change as there's no longer
23367 extension than 4-letter one.
23368 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23369 (gimplify_save_expr): Likewise.
23370 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23371 as it's possible even for other languages than Java.
23372 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23373 * lto-opts.c (lto_write_options): Remove reference to Java.
23374 * opts.c (strip_off_ending): Update file extension handling.
23375 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23376 * tree-eh.c (lower_resx): Likewise.
23377 * tree.c (free_lang_data_in_type): Remove dead code.
23378 (find_decls_types_r): Likewise.
23379 (build_common_builtin_nodes): Remove Java from a comment.
23380 (verify_type): Remove dead code.
23381 * varasm.c (assemble_external): Remove Java from a comment.
23383 2017-07-14 Martin Liska <mliska@suse.cz>
23385 * opts.c (finish_options): Add quotes.
23386 (common_handle_option): Likewise.
23388 2017-07-14 Martin Liska <mliska@suse.cz>
23390 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23391 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23392 Remove N_SO_PASCAL.
23393 * dwarf2out.c (lower_bound_default): Do not handle
23395 (gen_compile_unit_die): Likewise.
23396 * gcc.c: Remove default extension binding for GNU Pascal.
23397 * stmt.c: Remove Pascal language from a comment.
23398 * xcoffout.c: Likewise.
23400 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23403 * diagnostic-show-locus.c (fixit_cmp): New function.
23404 (layout::layout): Sort m_fixit_hints.
23405 (column_range::column_range): Assert that the values are valid.
23406 (struct char_span): New struct.
23407 (correction::overwrite): New method.
23408 (struct source_line): New struct.
23409 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23410 calls in terms of classes source_line and char_span, and
23411 correction::overwrite.
23412 (selftest::test_overlapped_fixit_printing_2): New function.
23413 (selftest::diagnostic_show_locus_c_tests): Call it.
23415 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23417 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23418 early if there is no lhs.
23420 2017-07-13 Martin Liska <mliska@suse.cz>
23422 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23423 (gen_reference_type_die): Likewise.
23424 * stor-layout.c: Remove Pascal-related comment.
23426 2017-07-13 Martin Liska <mliska@suse.cz>
23428 * opts.c (finish_options): Add quotes to error messages.
23429 (parse_sanitizer_options): Likewise.
23431 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23433 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23435 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23437 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23439 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23441 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23443 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23445 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23448 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23449 provides the hardware capability bits, define the macro
23450 __BUILTIN_CPU_SUPPORTS__.
23451 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23452 if GLIBC does not provide the hardware capability bits. Add a
23453 gcc_unreachable call if the built-in cpu function is neither
23454 __builtin_cpu_is nor __builtin_cpu_supports.
23455 (rs6000_get_function_versions_dispatcher): Change the warning
23456 that an old GLIBC is used which does not export the capability
23457 bits to be an error.
23458 * doc/extend.texi (target_clones attribute): Document the
23459 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23460 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23461 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23462 the macros defined by GCC if the newer GLIBC is available.
23464 2017-07-12 Jeff Law <law@redhat.com>
23466 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23467 remaining includes slightly.
23468 * config/riscv/riscv-builtins.c: Include profile-count.h.
23470 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23473 * config/avr/avr.c (avr_set_current_function): In diagnostic
23474 messages: Quote keywords and (parts of) identifiers.
23475 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23478 2017-07-12 Carl Love <cel@us.ibm.com>
23480 * config/rs6000/rs6000-c.c: Add support for built-in functions
23481 vector bool char vec_revb (vector bool char);
23482 vector bool short vec_revb (vector short char);
23483 vector bool int vec_revb (vector bool int);
23484 vector bool long long vec_revb (vector bool long long);
23485 * doc/extend.texi: Update the built-in documentation file for the
23486 new built-in functions.
23488 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23490 * config/s390/s390.md: Remove movcc splitter.
23492 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23494 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23495 load/store on condition.
23497 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23500 * config/avr/avr.c (avr_encode_section_info)
23501 [progmem && !TREE_READONLY]: Error if progmem object needs
23504 2017-07-11 Michael Collison <michael.collison@arm.com>
23506 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23509 2017-07-11 Carl Love <cel@us.ibm.com>
23511 * config/rs6000/rs6000-c.c: Add support for builtins
23512 vector unsigned int vec_parity_lsbb (vector signed int);
23513 vector unsigned int vec_parity_lsbb (vector unsigned int);
23514 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23515 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23516 vector unsigned long long vec_parity_lsbb (vector signed long long);
23517 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23518 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23519 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23520 * doc/extend.texi: Update the built-in documentation file for the
23521 new built-in functions.
23523 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23525 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23526 (layout::m_primary_loc): New field.
23527 (layout::layout): Initialize new field. Move location filtering
23528 logic from here to...
23529 (layout::maybe_add_location_range): ...this new method. Add
23530 support for filtering to just the lines already specified by other
23532 (layout::will_show_line_p): New method.
23533 (gcc_rich_location::add_location_if_nearby): New method.
23534 (selftest::test_add_location_if_nearby): New test function.
23535 (selftest::diagnostic_show_locus_c_tests): Call it.
23536 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23539 2017-07-11 Tom de Vries <tom@codesourcery.com>
23541 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23542 (bb_first_real_insn): New function.
23543 (nvptx_single): Add extra initialization of broadcasted condition
23546 2017-07-11 Nathan Sidwell <nathan@acm.org>
23548 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23550 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23552 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23553 Binutils doc as TEXI will mess them up.
23554 * doc/invoke.texi (AVR Options): Same here.
23556 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23558 * config/sparc/sparc.opt (mfix-ut700): New option.
23559 (mfix-gr712rc): Likewise.
23560 (sparc_fix_b2bst): New variable.
23561 * doc/invoke.texi (SPARC options): Document them.
23562 (ARM options): Fix warnings.
23563 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23564 instructions to prevent sequences that can trigger the store-store
23565 errata for certain LEON3FT processors.
23566 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23567 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23568 * config/sparc/sparc.md (fix_b2bst): New attribute.
23569 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23571 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23574 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23576 (*rsqrtsf2_sse): Ditto.
23578 (div<mode>3): Macroize insn from divdf3 and divsf3
23579 using MODEF mode iterator.
23581 2017-07-10 Martin Sebor <msebor@redhat.com>
23583 PR tree-optimization/80397
23584 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23585 instead of testing for equality to INTEGER_TYPE.
23587 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23589 * config.gcc: Remove uclibc from arc target spec.
23591 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23593 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23595 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23598 * lto-wrapper.c (remove_option): New function.
23599 (merge_and_complain): Merge PIC/PIE options more realistically.
23601 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23603 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23607 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23608 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23609 * config.in: Regenerate.
23610 * configure: Regenerate.
23611 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23612 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23613 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23614 (TARGET_GASISR_PROLOGUES): ...target mask.
23615 * common/config/avr/avr-common.c
23616 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23617 Set -mgas-isr-prologues.
23618 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23619 INSERT_PASS_BEFORE for it.
23620 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23621 * config/avr/avr.c (avr_option_override)
23622 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23623 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23624 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23625 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23626 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23628 (make_avr_pass_pre_proep): New function.
23629 (emit_push_sfr) <treg>: Add argument to function and use it
23630 instead of TMP_REG.
23631 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23632 and set machine->gasisr.yes.
23633 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23634 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23635 __gcc_isr.n_pushed to .L__stack_usage.
23636 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23637 (avr_asm_final_postscan_insn): ...this new static function.
23638 * config/avr/avr.h (machine_function)
23639 <is_no_gccisr, use_L__stack_usage>: New fields.
23640 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23641 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23642 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23643 (gasisr, *gasisr): New expander and insn.
23644 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23645 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23646 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23648 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23650 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23653 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23655 Move jump-tables out of .text again.
23658 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23659 (ASM_OUTPUT_ADDR_VEC): New function.
23660 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23661 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23662 INSN_ADDRESSes as asm comment.
23663 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23664 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23665 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23666 * config/avr/avr.md (*tablejump): Adjust comment.
23667 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23668 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23670 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23671 (avr_output_addr_vec): New proto.
23672 (avr_log_t) <insn_addresses>: New field.
23674 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23677 * config/i386/i386.c (ix86_function_arg_advance): Set
23678 outgoing_args_on_stack to true if there are outgoing arguments
23680 (ix86_function_arg): Likewise.
23681 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23682 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23683 * config/i386/i386.h (machine_function): Add
23684 outgoing_args_on_stack.
23686 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23688 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23689 supporting pthreds.
23690 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23692 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23694 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23695 (REAL_H): Remove $(MACHMODE_H).
23696 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23698 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23699 $(MACHMODE_H) and double-int.h.
23700 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23702 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23705 2017-07-07 Andrew Pinski <apinski@cavium.com>
23707 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23708 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23710 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23712 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23713 Add warning if GCC was not configured to link against a GLIBC that
23714 exports the hardware capability bits.
23715 (make_resolver_func): Make resolver function private and not a
23716 COMDAT function. Create the name with clone_function_name instead
23717 of make_unique_name.
23720 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23721 correct operand in doing the split.
23723 2017-07-07 Carl Love <cel@us.ibm.com>
23725 * config/rs6000/rs6000-c: Add support for built-in function
23726 vector unsigned short vec_pack_to_short_fp32 (vector float,
23728 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23729 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23730 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23731 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23732 (convert_4f32_8i16): Add define_expand.
23733 * doc/extend.texi: Update the built-in documentation file for the
23734 new built-in function.
23736 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23738 * config/sparc/m8.md: New file.
23739 * config/sparc/sparc.md: Include m8.md.
23741 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23743 * config/sparc/sparc.opt: New option -mvis4b.
23744 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23745 (sparc_option_override): Handle VIS4B.
23746 (enum sparc_builtins): Define
23747 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23748 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23749 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23750 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23751 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23752 (check_constant_argument): New function.
23753 (sparc_vis_init_builtins): Define builtins
23754 __builtin_vis_dictunpack{8,16,32},
23755 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23756 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23757 __builtin_vis_fpcmpde{8,16,32}shl and
23758 __builtin_vis_fpcmpur{8,16,32}shl.
23759 (sparc_expand_builtin): Check that the constant operands to
23760 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23761 constant and in range.
23762 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23764 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23765 (SPARC_IMM5_P): Likewise.
23766 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23767 (enabled): Handle vis4b.
23768 (UNSPEC_DICTUNPACK): New unspec.
23769 (UNSPEC_FPCMPSHL): Likewise.
23770 (UNSPEC_FPUCMPSHL): Likewise.
23771 (UNSPEC_FPCMPDESHL): Likewise.
23772 (UNSPEC_FPCMPURSHL): Likewise.
23773 (cpu_feature): New CPU feature `vis4b'.
23774 (dictunpack{8,16,32}): New insns.
23775 (FPCSMODE): New mode iterator.
23776 (fpcscond): New code iterator.
23777 (fpcsucond): Likewise.
23778 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23779 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23780 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23781 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23782 * config/sparc/constraints.md: Define constraints `q' for unsigned
23783 2-bit integer constants and `t' for unsigned 5-bit integer
23785 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23787 (imm5_operand_dictunpack16): Likewise.
23788 (imm5_operand_dictunpack32): Likewise.
23789 (imm2_operand): Likewise.
23790 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23791 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23792 ditunpack* and fpcmp*shl builtins.
23794 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23796 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23797 * config.in: Add HAVE_AS_SPARC6 define.
23798 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23800 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23802 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23803 (CPP_CPU_SPEC): Handle m8.
23804 (ASM_CPU_SPEC): Likewise.
23805 * config/sparc/sparc-opts.h (enum processor_type): Add
23807 * config/sparc/sparc.c (m8_costs): New struct.
23808 (sparc_option_override): Handle TARGET_CPU_m8.
23809 (sparc32_initialize_trampoline): Likewise.
23810 (sparc64_initialize_trampoline): Likewise.
23811 (sparc_issue_rate): Likewise.
23812 (sparc_register_move_cost): Likewise.
23813 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23814 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23815 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23816 (CPP_CPU_SPEC): Handle M8.
23817 (ASM_CPU_SPEC): Likewise.
23818 (AS_M8_FLAG): Define.
23819 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23820 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23821 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23823 * configure: Regenerate.
23824 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23827 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23829 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23831 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23832 ("*movdi_insn_sp32"): Do not set v3pipe.
23833 ("*movsi_insn"): Likewise.
23834 ("*movdi_insn_sp64"): Likewise.
23835 ("*movsf_insn"): Likewise.
23836 ("*movdf_insn_sp32"): Likewise.
23837 ("*movdf_insn_sp64"): Likewise.
23838 ("*zero_extendsidi2_insn_sp64"): Likewise.
23839 ("*sign_extendsidi2_insn"): Likewise.
23840 ("*mov<VM32:mode>_insn"): Likewise.
23841 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23842 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23843 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23844 ("<vlop:code><VL:mode>3"): Likewise.
23845 ("*not_<vlop:code><VL:mode>3"): Likewise.
23846 ("*nand<VL:mode>_vis"): Likewise.
23847 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23848 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23849 ("one_cmpl<VL:mode>2"): Likewise.
23850 ("faligndata<VM64:mode>_vis"): Likewise.
23851 ("alignaddrsi_vis"): Likewise.
23852 ("alignaddrdi_vis"): Likweise.
23853 ("alignaddrlsi_vis"): Likewise.
23854 ("alignaddrldi_vis"): Likewise.
23855 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23856 ("bmaskdi_vis"): Likewise.
23857 ("bmasksi_vis"): Likewise.
23858 ("bshuffle<VM64:mode>_vis"): Likewise.
23859 ("cmask8<P:mode>_vis"): Likewise.
23860 ("cmask16<P:mode>_vis"): Likewise.
23861 ("cmask32<P:mode>_vis"): Likewise.
23862 ("pdistn<P:mode>_vis"): Likewise.
23863 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23865 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23867 * config/sparc/sparc.md ("subtype"): New insn attribute.
23868 ("*wrgsr_sp64"): Set insn subtype.
23869 ("*rdgsr_sp64"): Likewise.
23870 ("alignaddrsi_vis"): Likewise.
23871 ("alignaddrdi_vis"): Likewise.
23872 ("alignaddrlsi_vis"): Likewise.
23873 ("alignaddrldi_vis"): Likewise.
23874 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23875 ("fexpand_vis"): Likewise.
23876 ("fpmerge_vis"): Likewise.
23877 ("faligndata<VM64:mode>_vis"): Likewise.
23878 ("bshuffle<VM64:mode>_vis"): Likewise.
23879 ("cmask8<P:mode>_vis"): Likewise.
23880 ("cmask16<P:mode>_vis"): Likewise.
23881 ("cmask32<P:mode>_vis"): Likewise.
23882 ("fchksm16_vis"): Likewise.
23883 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23884 ("fmean16_vis"): Likewise.
23885 ("fp<plusminus_insn>64_vis"): Likewise.
23886 ("<plusminus_insn>v8qi3"): Likewise.
23887 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23888 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23889 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23890 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23891 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23892 ("*movqi_insn"): Likewise.
23893 ("*movhi_insn"): Likewise.
23894 ("*movsi_insn"): Likewise.
23895 ("movsi_pic_gotdata_op"): Likewise.
23896 ("*movdi_insn_sp32"): Likewise.
23897 ("*movdi_insn_sp64"): Likewise.
23898 ("movdi_pic_gotdata_op"): Likewise.
23899 ("*movsf_insn"): Likewise.
23900 ("*movdf_insn_sp32"): Likewise.
23901 ("*movdf_insn_sp64"): Likewise.
23902 ("*zero_extendhisi2_insn"): Likewise.
23903 ("*zero_extendqihi2_insn"): Likewise.
23904 ("*zero_extendqisi2_insn"): Likewise.
23905 ("*zero_extendqidi2_insn"): Likewise.
23906 ("*zero_extendhidi2_insn"): Likewise.
23907 ("*zero_extendsidi2_insn_sp64"): Likewise.
23908 ("ldfsr"): Likewise.
23909 ("prefetch_64"): Likewise.
23910 ("prefetch_32"): Likewise.
23911 ("tie_ld32"): Likewise.
23912 ("tie_ld64"): Likewise.
23913 ("*tldo_ldub_sp32"): Likewise.
23914 ("*tldo_ldub1_sp32"): Likewise.
23915 ("*tldo_ldub2_sp32"): Likewise.
23916 ("*tldo_ldub_sp64"): Likewise.
23917 ("*tldo_ldub1_sp64"): Likewise.
23918 ("*tldo_ldub2_sp64"): Likewise.
23919 ("*tldo_ldub3_sp64"): Likewise.
23920 ("*tldo_lduh_sp32"): Likewise.
23921 ("*tldo_lduh1_sp32"): Likewise.
23922 ("*tldo_lduh_sp64"): Likewise.
23923 ("*tldo_lduh1_sp64"): Likewise.
23924 ("*tldo_lduh2_sp64"): Likewise.
23925 ("*tldo_lduw_sp32"): Likewise.
23926 ("*tldo_lduw_sp64"): Likewise.
23927 ("*tldo_lduw1_sp64"): Likewise.
23928 ("*tldo_ldx_sp64"): Likewise.
23929 ("*mov<VM32:mode>_insn"): Likewise.
23930 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23931 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23933 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23935 * config/sparc/sparc.md ("type"): New insn type viscmp.
23936 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23938 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23939 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23940 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23941 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23943 ("n7_vis_logical_11cycle"): Likewise.
23944 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23945 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23946 * config/sparc/niagara.md ("niag_vis"): Likewise.
23947 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23948 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23950 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23952 * config/sparc/sparc.md: New instruction type `bmask'.
23953 (bmaskdi_vis): Use the `bmask' type.
23954 (bmasksi_vis): Likewise.
23955 * config/sparc/ultra3.md (us3_array): Likewise.
23956 * config/sparc/niagara7.md (n7_array): Likewise.
23957 * config/sparc/niagara4.md (n4_array): Likewise.
23958 * config/sparc/niagara2.md (niag2_vis): Likewise.
23959 (niag3_vis): Likewise.
23960 * config/sparc/niagara.md (niag_vis): Likewise.
23962 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23964 * ipa-comdats.c: Remove optimize check from gate.
23965 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23966 for functions not optimized.
23967 (ipa_fn_summary_read): Skip optimize check.
23968 (ipa_fn_summary_write): Likewise.
23969 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23971 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23973 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23974 (check_callers): Give up on caller which is not optimized.
23975 (inline_small_functions): Likewise.
23976 (ipa_inline): Do not give up when not optimizing.
23977 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23978 away unoptimizes cdtors.
23979 (whole_program_function_and_variable_visibility): Do
23980 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23981 * ipa.c (process_references): Do not check optimize.
23982 (symbol_table::remove_unreachable_nodes): Update optimize check.
23983 (set_writeonly_bit): Update optimize check.
23984 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23985 (pass_ipa_single_use::gate): Remove.
23987 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23989 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23990 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23991 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23992 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23993 permute_load, permute_store, adjust_extract, adjust_splat,
23994 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23995 replace_swap_with_copy, dump_swap_insn_table,
23996 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23997 recombine_lvx_pattern, recombine_stvx_pattern,
23998 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23999 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24000 to file rs6000-p8swap.c.
24001 * config/rs6000/rs6000-p8swap.c: New file.
24002 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24003 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24004 and rs6000*-*-* targets.
24006 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24008 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24010 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24012 * lto-wrapper.c (merge_and_complain): Do not merge
24013 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24014 fsigned_zeros, ftrapping_math, fwrapv.
24015 (append_compiler_options): Do not track these options.
24016 (append_linker_options): Likewie
24018 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24020 * cgraphunit.c (cgraph_node::finalize_function): When
24021 !flag_toplevel_reorde set no_reorder flag.
24022 (varpool_node::finalize_decl): Likewise.
24023 (symbol_table::compile): Drop no toplevel reorder path.
24025 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24027 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24028 edges; zero probability is not better than uninitialized.
24030 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24032 * asan.h (asan_sanitize_allocas_p): Declare.
24033 * asan.c (asan_sanitize_allocas_p): New function.
24034 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24035 (handle_builtin_alloca): Likewise.
24036 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24037 if !asan_sanitize_allocas_p.
24038 * params.def (asan-instrument-allocas): Add new option.
24039 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24040 * opts.c (common_handle_option): Disable allocas sanitization for
24043 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24045 * asan.c: Include gimple-fold.h.
24046 (get_last_alloca_addr): New function.
24047 (handle_builtin_stackrestore): Likewise.
24048 (handle_builtin_alloca): Likewise.
24049 (asan_emit_allocas_unpoison): Likewise.
24050 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24051 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24052 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24053 (instrument_builtin_call): Pass gimple iterator to
24054 get_mem_refs_of_builtin_call.
24055 (last_alloca_addr): New global.
24056 * asan.h (asan_emit_allocas_unpoison): Declare.
24057 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24058 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24059 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24060 if function calls alloca.
24061 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24062 * gimple-fold.h (replace_call_with_value): Declare.
24063 * internal-fn.c: Include asan.h.
24064 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24065 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24067 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24069 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24070 (C_SELFTEST_FLAGS): New.
24071 (CPP_SELFTEST_FLAGS): New.
24072 (SELFTEST_DEPS): New, from deps of s-selftest.
24073 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24074 (CPP_SELFTEST_DEPS): New.
24075 (selftest): Add dependency on s-selftest-c++.
24076 (s-selftest): Rename to...
24077 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24078 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24079 than SELFTEST_FLAGS.
24080 (selftest-gdb): Rename to...
24081 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24083 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24084 (selftest-valgrind): Rename to...
24085 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24087 (selftest-valgrind): Reintroduce as an alias for
24088 selftest-c-valgrind.
24089 (s-selftest-c++): New.
24090 (selftest-c++-gdb): New.
24091 (selftest-c++-valgrind): New.
24093 2017-07-06 Olivier Hainque <hainque@adacore.com>
24095 * gcc.c (process_command): When deciding if undefined variables
24096 should be ignored when processing specs, accept "gcc -v" as well.
24098 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24100 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24101 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24103 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24105 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24106 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24107 * config/arm/arm-tables.opt: Regenerate.
24108 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24110 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24112 2017-07-06 Carl Love <cel@us.ibm.com>
24114 * ChangeLog: Clean up from mid air collision
24116 2017-07-06 Carl Love <cel@us.ibm.com>
24118 * config/rs6000/rs6000-c.c: Add support for built-in functions
24119 vector signed int vec_subc (vector signed int, vector signed int);
24120 vector signed __int128 vec_subc (vector signed __int128,
24121 vector signed __int128);
24122 vector unsigned __int128 vec_subc (vector unsigned __int128,
24123 vector unsigned __int128);
24124 vector signed int vec_sube (vector signed int, vector signed int,
24125 vector signed int);
24126 vector unsigned int vec_sube (vector unsigned int,
24127 vector unsigned int,
24128 vector unsigned int);
24129 vector signed __int128 vec_sube (vector signed __int128,
24130 vector signed __int128,
24131 vector signed__int128);
24132 vector unsigned __int128 vec_sube (vector unsigned __int128,
24133 vector unsigned __int128,
24134 vector unsigned __int128);
24135 vector signed int vec_subec (vector signed int, vector signed int,
24136 vector signed int);
24137 vector unsigned int vec_subec (vector unsigned int,
24138 vector unsigned int,
24139 vector unsigned int);
24140 vector signed __int128 vec_subec (vector signed __int128,
24141 vector signed __int128,
24142 vector signed__int128);
24143 vector unsigned __int128 vec_subec (vector unsigned __int128,
24144 vector unsigned __int128,
24145 vector unsigned __int128);
24146 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24147 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24148 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24149 BU_ALTIVEC_OVERLOAD_X definitions.
24150 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24151 * doc/extend.texi: Update the built-in documentation file for the new
24152 built-in functions.
24154 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24157 * diagnostic-show-locus.c (layout::layout): Use start and finish
24158 spelling location for the start and finish of each range.
24159 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24160 Add unused aspect param.
24161 * input.c (expand_location_1): Add "aspect" param, and use it
24162 to access the correct part of the location.
24163 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24165 (expand_location_to_spelling_point): Likewise.
24166 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24167 param, and pass it to expand_location_1.
24169 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24171 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24172 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24173 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24174 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24175 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24176 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24177 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24178 _mm_maskz_getmant_ss): New intrinsics.
24179 (__builtin_ia32_getexpss128_mask): Changed to ...
24180 __builtin_ia32_getexpss128_round ... this.
24181 (__builtin_ia32_getexpsd128_mask): Changed to ...
24182 __builtin_ia32_getexpsd128_round ... this.
24183 * config/i386/i386-builtin-types.def
24184 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24185 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24186 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24187 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24188 __builtin_ia32_getmantss_mask_round): New builtins.
24189 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24190 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24191 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24192 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24193 * config/i386/sse.md
24194 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24195 avx512f_sgetexp<mode><mask_scalar_name>
24196 <round_saeonly_scalar_name> ... this.
24197 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24198 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24199 vgetexp<ssescalarmodesuffix>
24200 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24201 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24202 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24203 avx512f_vgetmant<mode><mask_scalar_name>
24204 <round_saeonly_scalar_name> ... this.
24205 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24206 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24207 vgetmant<ssescalarmodesuffix>
24208 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24209 %0<mask_scalar_operand4>, %1, %2
24210 <round_saeonly_scalar_mask_op4>, %3} ... this.
24211 * config/i386/subst.md (mask_scalar_operand4,
24212 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24213 round_saeonly_scalar_nimm_predicate): New subst attributes.
24215 2017-07-06 Julia Koval <julia.koval@intel.com>
24217 * config/i386/i386.c (ix86_erase_embedded_rounding):
24218 Remove code for old rounding pattern.
24220 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24222 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24224 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24226 * doc/sourcebuild.texi (Test Directives, Variants of
24227 dg-require-support): Add documentation for dg-require-stack-check.
24229 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24231 * config/i386/subst.md (mask_scalar, round_scalar,
24232 round_saeonly_scalar): New meta-templates.
24233 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24234 round_scalar_mask_operand3, round_scalar_mask_op3,
24235 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24236 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24237 round_saeonly_scalar_constraint,
24238 round_saeonly_scalar_prefix): New subst attribute.
24239 * config/i386/sse.md
24240 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24241 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24242 <round_scalar_name> ... this.
24243 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24244 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24245 <round_scalar_name> ... this.
24246 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24247 <sse>_vm<code><mode>3<mask_scalar_name>
24248 <round_saeonly_scalar_name> ... this.
24249 (v<plusminus_mnemonic><ssescalarmodesuffix>
24250 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24251 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24252 v<plusminus_mnemonic><ssescalarmodesuffix>
24253 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24254 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24255 (v<multdiv_mnemonic><ssescalarmodesuffix>
24256 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24257 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24258 v<multdiv_mnemonic><ssescalarmodesuffix>
24259 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24260 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24261 (v<maxmin_float><ssescalarmodesuffix>
24262 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24263 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24264 v<maxmin_float><ssescalarmodesuffix>
24265 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24266 %0<mask_scalar_operand3>, %1, %<iptr>2
24267 <round_saeonly_scalar_mask_op3>} ... this.
24269 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24271 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24272 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24274 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24275 Alan Hayward <alan.hayward@arm.com>
24276 David Sherwood <david.sherwood@arm.com>
24278 * combine.c (simplify_if_then_else): Remove "enum" before
24280 * compare-elim.c (can_eliminate_compare): Likewise.
24281 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24283 (aarch64_lookup_simd_builtin_type): Likewise.
24284 (aarch64_simd_builtin_type): Likewise.
24285 (aarch64_init_simd_builtin_types): Likewise.
24286 (aarch64_simd_expand_args): Likewise.
24287 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24289 (aarch64_reverse_mask): Likewise.
24290 (aarch64_simd_emit_reg_reg_move): Likewise.
24291 (aarch64_gen_adjusted_ldpstp): Likewise.
24292 (aarch64_ccmp_mode_to_code): Likewise.
24293 (aarch64_operands_ok_for_ldpstp): Likewise.
24294 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24295 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24297 (aarch64_min_divisions_for_recip_mul): Likewise.
24298 (aarch64_reassociation_width): Likewise.
24299 (aarch64_get_condition_code_1): Likewise.
24300 (aarch64_simd_emit_reg_reg_move): Likewise.
24301 (aarch64_simd_attr_length_rglist): Likewise.
24302 (aarch64_reverse_mask): Likewise.
24303 (aarch64_operands_ok_for_ldpstp): Likewise.
24304 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24305 (aarch64_gen_adjusted_ldpstp): Likewise.
24306 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24308 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24309 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24310 (arm_lookup_simd_builtin_type): Likewise.
24311 (arm_simd_builtin_type): Likewise.
24312 (arm_init_simd_builtin_types): Likewise.
24313 (arm_expand_builtin_args): Likewise.
24314 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24315 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24316 (ft32_setup_incoming_varargs): Likewise.
24317 (ft32_function_arg): Likewise.
24318 (ft32_function_arg_advance): Likewise.
24319 (ft32_pass_by_reference): Likewise.
24320 (ft32_arg_partial_bytes): Likewise.
24321 (ft32_valid_pointer_mode): Likewise.
24322 (ft32_addr_space_pointer_mode): Likewise.
24323 (ft32_addr_space_legitimate_address_p): Likewise.
24324 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24326 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24327 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24328 (iamcu_alignment): Likewise.
24329 (canonicalize_vector_int_perm): Likewise.
24330 (ix86_noce_conversion_profitable_p): Likewise.
24331 (ix86_mpx_bound_mode): Likewise.
24332 (ix86_operands_ok_for_move_multiple): Likewise.
24333 * config/microblaze/microblaze-protos.h
24334 (microblaze_expand_conditional_branch_reg): Likewise.
24335 * config/microblaze/microblaze.c
24336 (microblaze_expand_conditional_branch_reg): Likewise.
24337 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24339 (rs6000_reassociation_width): Likewise.
24340 (rs6000_invalid_binary_op): Likewise.
24341 (fusion_p9_p): Likewise.
24342 (emit_fusion_p9_load): Likewise.
24343 (emit_fusion_p9_store): Likewise.
24344 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24346 (riscv_hard_regno_mode_ok_p): Likewise.
24347 (riscv_address_insns): Likewise.
24348 (riscv_split_symbol): Likewise.
24349 (riscv_legitimize_move): Likewise.
24350 (riscv_function_value): Likewise.
24351 (riscv_hard_regno_nregs): Likewise.
24352 (riscv_expand_builtin): Likewise.
24353 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24354 (riscv_build_integer): Likewise.
24355 (riscv_split_integer): Likewise.
24356 (riscv_legitimate_constant_p): Likewise.
24357 (riscv_cannot_force_const_mem): Likewise.
24358 (riscv_regno_mode_ok_for_base_p): Likewise.
24359 (riscv_valid_base_register_p): Likewise.
24360 (riscv_valid_offset_p): Likewise.
24361 (riscv_valid_lo_sum_p): Likewise.
24362 (riscv_classify_address): Likewise.
24363 (riscv_legitimate_address_p): Likewise.
24364 (riscv_address_insns): Likewise.
24365 (riscv_load_store_insns): Likewise.
24366 (riscv_force_binary): Likewise.
24367 (riscv_split_symbol): Likewise.
24368 (riscv_force_address): Likewise.
24369 (riscv_legitimize_address): Likewise.
24370 (riscv_move_integer): Likewise.
24371 (riscv_legitimize_const_move): Likewise.
24372 (riscv_legitimize_move): Likewise.
24373 (riscv_address_cost): Likewise.
24374 (riscv_subword): Likewise.
24375 (riscv_output_move): Likewise.
24376 (riscv_canonicalize_int_order_test): Likewise.
24377 (riscv_emit_int_order_test): Likewise.
24378 (riscv_function_arg_boundary): Likewise.
24379 (riscv_pass_mode_in_fpr_p): Likewise.
24380 (riscv_pass_fpr_single): Likewise.
24381 (riscv_pass_fpr_pair): Likewise.
24382 (riscv_get_arg_info): Likewise.
24383 (riscv_function_arg): Likewise.
24384 (riscv_function_arg_advance): Likewise.
24385 (riscv_arg_partial_bytes): Likewise.
24386 (riscv_function_value): Likewise.
24387 (riscv_pass_by_reference): Likewise.
24388 (riscv_setup_incoming_varargs): Likewise.
24389 (riscv_print_operand): Likewise.
24390 (riscv_elf_select_rtx_section): Likewise.
24391 (riscv_save_restore_reg): Likewise.
24392 (riscv_for_each_saved_reg): Likewise.
24393 (riscv_register_move_cost): Likewise.
24394 (riscv_hard_regno_mode_ok_p): Likewise.
24395 (riscv_hard_regno_nregs): Likewise.
24396 (riscv_class_max_nregs): Likewise.
24397 (riscv_memory_move_cost): Likewise.
24398 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24399 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24400 (rl78_addr_space_address_mode): Likewise.
24401 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24403 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24404 (rs6000_reassociation_width): Likewise.
24405 (rs6000_invalid_binary_op): Likewise.
24406 (fusion_p9_p): Likewise.
24407 (emit_fusion_p9_load): Likewise.
24408 (emit_fusion_p9_store): Likewise.
24409 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24410 (ok_for_simple_move_operands): Likewise.
24411 (ok_for_simple_move_strict_operands): Likewise.
24412 (ok_for_simple_arith_logic_operands): Likewise.
24413 (visium_legitimize_reload_address): Likewise.
24414 (visium_select_cc_mode): Likewise.
24415 (output_cbranch): Likewise.
24416 (visium_split_double_move): Likewise.
24417 (visium_expand_copysign): Likewise.
24418 (visium_expand_int_cstore): Likewise.
24419 (visium_expand_fp_cstore): Likewise.
24420 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24421 (visium_function_arg): Likewise.
24422 (visium_function_arg_advance): Likewise.
24423 (visium_libcall_value): Likewise.
24424 (visium_setup_incoming_varargs): Likewise.
24425 (visium_legitimate_constant_p): Likewise.
24426 (visium_legitimate_address_p): Likewise.
24427 (visium_legitimize_address): Likewise.
24428 (visium_secondary_reload): Likewise.
24429 (visium_register_move_cost): Likewise.
24430 (visium_memory_move_cost): Likewise.
24431 (prepare_move_operands): Likewise.
24432 (ok_for_simple_move_operands): Likewise.
24433 (ok_for_simple_move_strict_operands): Likewise.
24434 (ok_for_simple_arith_logic_operands): Likewise.
24435 (visium_function_value_1): Likewise.
24436 (rtx_ok_for_offset_p): Likewise.
24437 (visium_legitimize_reload_address): Likewise.
24438 (visium_split_double_move): Likewise.
24439 (visium_expand_copysign): Likewise.
24440 (visium_expand_int_cstore): Likewise.
24441 (visium_expand_fp_cstore): Likewise.
24442 (visium_split_cstore): Likewise.
24443 (visium_select_cc_mode): Likewise.
24444 (visium_split_cbranch): Likewise.
24445 (output_cbranch): Likewise.
24446 (visium_print_operand_address): Likewise.
24447 * expmed.c (flip_storage_order): Likewise.
24448 * expmed.h (emit_cstore): Likewise.
24449 (flip_storage_order): Likewise.
24450 * genrecog.c (validate_pattern): Likewise.
24451 * hsa-gen.c (gen_hsa_addr): Likewise.
24452 * internal-fn.c (expand_arith_overflow): Likewise.
24453 * ira-color.c (allocno_copy_cost_saving): Likewise.
24454 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24455 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24456 (process_invariant_for_inheritance): Likewise.
24457 * lra-eliminations.c (move_plus_up): Likewise.
24458 * omp-low.c (lower_oacc_reductions): Likewise.
24459 * simplify-rtx.c (simplify_subreg): Likewise.
24460 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24461 (TARGET_CHKP_BOUND_MODE): Likewise..
24462 * targhooks.c (default_chkp_bound_mode): Likewise.
24463 (default_setup_incoming_vararg_bounds): Likewise.
24464 * targhooks.h (default_chkp_bound_mode): Likewise.
24465 (default_setup_incoming_vararg_bounds): Likewise.
24466 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24467 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24468 (have_whole_vector_shift): Likewise.
24469 * tree-vect-stmts.c (vectorizable_load): Likewise.
24470 * doc/tm.texi: Regenerate.
24472 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24474 Graceful degrade if Binutils PR21472 is not available.
24477 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24478 .rodata in flash test fails.
24479 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24480 * confgure: Regenerate.
24481 * config.in: Regenerate.
24482 * config/avr/avr.c (avr_asm_named_section)
24483 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24484 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24485 (avr_asm_init_sections): Same.
24487 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24489 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24490 (fma<VH:mode>4_intrinsic): Likewise.
24491 (*fmsub<VCVTF:mode>4): Likewise.
24492 (*fmsub<VH:mode>4_intrinsic): Likewise.
24494 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24497 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24498 Don't depend on "optimize > 0".
24499 (out_movhi_r_mr, out_movqi_mr_r): Same.
24500 (out_movhi_mr_r, out_movqi_r_mr): Same.
24501 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24502 io_address_operand on "optimize > 0".
24504 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24506 * tree-loop-distribution.c: Add general explanantion on the pass.
24507 (generate_loops_for_partition): Mark distributed loop.
24508 (pg_add_dependence_edges): New parameter. Handle alias data
24509 dependence specially and record it in the parameter if asked.
24510 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24511 (init_partition_graph_vertices, add_partition_graph_edge): New.
24512 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24513 (free_partition_graph_vdata, build_partition_graph): New.
24514 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24515 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24516 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24517 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24518 (version_for_distribution_p, finalize_partitions): New.
24519 (distribute_loop): Handle alias data dependence specially. Factor
24520 out loop fusion code as functions and call these functions.
24522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24524 * tree-loop-distribution.c (classify_partition): New parameter and
24525 better handle reduction statement.
24526 (rdg_build_partitions): Revise comment.
24527 (distribute_loop): Compute statements in all partitions and pass it
24528 to classify_partition.
24530 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24532 * tree-loop-distribution.c (enum partition_type): New.
24533 (struct partition): New field type.
24534 (partition_merge_into): Add parameter. Update partition type.
24535 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24536 (build_rdg_partition_for_vertex): Compute partition type.
24537 (rdg_build_partitions): Dump partition type.
24538 (distribute_loop): Update calls to partition_merge_into.
24540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24542 * tree-loop-distribution.c (struct ddr_hasher): New.
24543 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24545 (classify_partition): Call get_data_dependence.
24546 (pg_add_dependence_edges): Ditto.
24547 (distribute_loop): Release data dependence hash table.
24549 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24551 * tree-loop-distribution.c (ref_base_address): Delete.
24552 (similar_memory_accesses): Rename ...
24553 (share_memory_accesses): ... to this. Check if partitions access
24554 the same memory reference.
24555 (distribute_loop): Call share_memory_accesses.
24557 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24559 * tree-loop-distribution.c (struct partition): New field recording
24560 its data reference.
24561 (partition_alloc, partition_free): Init and release data refs.
24562 (partition_merge_into): Merge data refs.
24563 (build_rdg_partition_for_vertex): Collect data refs for partition.
24564 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24566 (distribute_loop): Remve data refs from vertice data of partition
24569 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24571 * tree-loop-distribution.c (params.h): Include header file.
24572 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24573 (datarefs_vec): New global var.
24574 (create_rdg_vertices): Use datarefs_vec directly.
24575 (free_rdg): Don't free data references.
24576 (build_rdg): Update use. Don't free data references.
24577 (distribute_loop): Compute global variable for data references.
24578 Bail out if there are too many data references.
24580 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24582 * tree-loop-distribution.c (loop_nest): New global var.
24583 (build_rdg): Use loop directly, rather than loop nest.
24584 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24586 (distribute_loop): Compute global variable loop nest. Update use.
24588 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24590 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24591 (partition_merge_into): New parameter. Dump reason for fusion.
24592 (distribute_loop): Update use of partition_merge_into.
24594 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24596 * tree-loop-distribution.c (bb_top_order_index): New.
24597 (bb_top_order_index_size, bb_top_order_cmp): New.
24598 (stmts_from_loop): Use topological order.
24599 (pass_loop_distribution::execute): Compute and release topological
24600 order for basic blocks.
24602 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24604 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24607 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24609 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24610 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24611 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24612 * internal-fn.def (LOOP_DIST_ALIAS): New.
24613 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24614 (fold_loop_internal_call): ... this.
24615 (vect_loop_dist_alias_call): New function.
24616 (set_uid_loop_bbs): Call fold_loop_internal_call.
24617 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24620 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24623 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24624 Require dead FLAGS_REG at the beginning of a peephole.
24626 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24629 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24630 arguments in the call to __builtin_ia32_sbb_u32.
24631 (_subborrow_u64): Swap _X and _Y arguments in the call to
24632 __builtin_ia32_sbb_u64.
24634 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24637 * tree-vrp.c (compare_assert_loc): Turn into a function template
24638 with stable template parameter. Only test if a->e is NULL,
24639 !a->e == !b->e has been verified already. Use e == NULL or
24640 e != NULL instead of e or ! e tests. If stable is true, don't use
24641 iterative_hash_expr, on the other side allow a or b or both NULL
24642 and sort the NULLs last.
24643 (process_assert_insertions): Sort using compare_assert_loc<false>
24644 instead of compare_assert_loc, later sort using
24645 compare_assert_loc<true> before calling process_assert_insertions_for
24646 in a loop. Use break instead of continue once seen NULL pointer.
24648 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24650 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24651 Cortex-R7 and Cortex-R8 processors.
24653 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24655 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24656 uninitialized while src is not.
24658 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24660 * common/config/arm/arm-common.c: Adjust include path for
24662 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24663 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24664 (arm-cpu-data.h): Likewise.
24665 (arm-cpu-cdata.h): Likewise.
24666 * config/arm/arm-cpu.h: Delete.
24667 * config/arm/arm-cpu-cdata.h: Delete.
24668 * config/arm/arm-cpu-data.h: Delete.
24670 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24672 * config/arm/arm-cpus.in (cortex-a55): New.
24673 (cortex-a75): Likewise.
24674 (cortex-a75.cortex-a55): Likewise.
24675 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24677 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24678 * config/arm/arm-cpu-cdata.h: Regenerate.
24679 * config/arm/arm-cpu-data.h: Regenerate.
24680 * config/arm/arm-cpu.h: Regenerate.
24681 * config/arm/arm-tables.opt: Regenerate.
24682 * config/arm/arm-tune.md: Regenerate.
24684 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24686 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24688 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24690 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24693 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24695 PR tree-optimization/81292
24696 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24697 full_string_p, also call adjust_related_strinfos if the adjustment
24698 is simple, otherwise invalidate related strinfos.
24700 2017-07-04 Martin Liska <mliska@suse.cz>
24703 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24704 newly created variable as DECL_IGNORED_P.
24706 2017-07-04 Martin Liska <mliska@suse.cz>
24709 * ipa-inline.c (inline_small_functions):
24710 Use xstrdup_for_dump.
24712 2017-07-04 Tom de Vries <tom@codesourcery.com>
24714 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24716 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24719 * config/darwin.c (darwin_function_switched_text_sections):
24722 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24724 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24726 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24728 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24730 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24732 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24733 min_profitable_iters, and th as inclusive lower bounds.
24734 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24735 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24736 for min_profitable_iters and min_profitable_estimate.
24737 (vect_transform_loop): Treat th as an inclusive lower bound.
24738 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24740 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24743 * config/darwin.c (darwin_function_switched_text_sections):
24744 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24745 in two pieces, and suppress the use of buf.
24747 2017-07-03 Nathan Sidwell <nathan@acm.org>
24749 * hash-table.h (hash_table_mod1): Fix indentation.
24751 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24753 PR middle-end/81290
24754 * predict.c (force_edge_cold): Be more careful about propagation
24756 * profile-count.h (profile_probability::guessed,
24757 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24759 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24761 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24763 * doc/invoke.texi (rcpc architecture extension): Document it.
24765 2017-07-03 Richard Biener <rguenther@suse.de>
24767 PR tree-optimization/60510
24768 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24769 the scalar reduction PHI and use it.
24770 (vectorizable_reduction): Properly guard the single_defuse_cycle
24771 path for non-SLP reduction chains where we cannot use it.
24772 Rework reduc_def/index and vector type deduction. Rework
24773 vector operand gathering during reduction op code-gen.
24774 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24775 chains dissolve the chain and leave it to non-SLP reduction
24778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24780 * tree-data-ref.h (dr_alignment): Declare.
24781 * tree-data-ref.c (dr_alignment): New function.
24782 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24783 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24785 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24787 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24789 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24790 and base_misalignment fields.
24791 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24792 * tree-data-ref.c: Include builtins.h.
24793 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24794 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24795 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24796 * tree-vect-data-refs.c: Include tree-cfg.h.
24797 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24798 fields instead of calculating an alignment here.
24799 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24800 innermost_loop_behavior fields.
24802 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24804 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24806 (DR_STEP_ALIGNMENT): New macro.
24807 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24808 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24809 (create_data_ref): Print it.
24810 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24811 to tell whether the step preserves vector (mis)alignment.
24812 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24813 Move the check for an integer step and generalise to all INTEGER_CST.
24814 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24815 Print the outer step alignment.
24817 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24819 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24820 with offset_alignment.
24821 (DR_ALIGNED_TO): Delete.
24822 (DR_OFFSET_ALIGNMENT): New macro.
24823 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24824 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24825 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24826 (create_data_ref): Likewise.
24827 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24828 (vect_analyze_data_refs): Likewise.
24829 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24830 creating dummy innermost behavior.
24832 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24834 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24835 with a "innermost_loop_behavior *" and refeence tree.
24836 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24837 (create_data_ref): Update call accordingly.
24838 * tree-predcom.c (find_looparound_phi): Likewise.
24840 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24842 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24843 fields with dr_wrt_vec_loop.
24844 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24845 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24846 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24847 (vect_dr_behavior): New function.
24848 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24849 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24850 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24851 track whether the step preserves the misalignment.
24852 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24853 Use vect_dr_behavior.
24854 (vect_setup_realignment): Update call accordingly.
24855 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24856 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24857 call to vect_create_addr_base_for_vector_ref.
24858 (vect_create_cond_for_align_checks): Likewise.
24859 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24860 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24861 (vect_recog_mask_conversion_pattern): Likewise.
24862 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24863 (new_stmt_vec_info): Remove redundant zeroing.
24865 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24867 * common/config/arm/arm-common.c (arm_be8_option): New function.
24868 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24869 (ISA_ARMv6): Add isa_bit_be8.
24870 * config/arm/arm.h (arm_be8_option): Add prototype.
24871 (BE8_SPEC_FUNCTION): New define.
24872 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24873 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24874 (mlittle-endian): Similarly.
24875 (mbe8, mbe32): New options.
24876 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24877 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24879 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24881 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24883 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24885 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24886 (cleanup_tree_cfg_bb): Use it.
24887 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24889 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24891 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24894 * loop-doloop.c (add_test): Update profile.
24896 2017-07-03 Martin Liska <mliska@suse.cz>
24899 * sanopt.c (rewrite_usage_of_param): New function.
24900 (sanitize_rewrite_addressable_params): Likewise.
24901 (pass_sanopt::execute): Call rewrite_usage_of_param.
24903 2017-07-03 Richard Biener <rguenther@suse.de>
24905 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24906 back to using VIEW_CONVERT_EXPR.
24908 2017-07-03 Martin Liska <mliska@suse.cz>
24911 * doc/extend.texi: Document when a resolver function is
24912 generated for target_clones.
24914 2017-07-03 Martin Liska <mliska@suse.cz>
24916 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24917 and shadow memory of auto variables which are subject of
24918 use-after-scope sanitization.
24919 (asan_expand_mark_ifn): Add do set only when is_poison.
24921 2016-07-03 Richard Biener <rguenther@suse.de>
24923 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24925 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24926 (vectorizable_reduction): Perform reduction PHI creation when
24927 visiting a reduction PHI and adjust and simplify code generation
24928 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24929 (vect_transform_loop): Visit reduction PHIs.
24930 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24931 defs into the SLP tree.
24932 (vect_build_slp_tree): Reduction defs terminate the recursion.
24933 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24935 (vect_get_vec_defs_for_stmt_copy): Export.
24936 (vect_get_vec_defs): Likewise.
24937 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24939 (vect_get_vec_defs_for_stmt_copy): Declare.
24940 (vect_get_vec_defs): Likewise.
24942 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24944 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24945 parameter with a "loop" parameter and use it instead of the
24946 loop containing DR_STMT. Don't check simple_iv when doing
24947 BB analysis. Describe the two analysis modes in the comment.
24949 2017-07-03 Tom de Vries <tom@codesourcery.com>
24951 PR tree-optimization/69468
24952 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24953 (find_same_succ_bb): Handle ignore_edge_flags.
24955 2017-07-03 Tom de Vries <tom@codesourcery.com>
24957 PR tree-optimization/81192
24958 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24960 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24962 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24964 2017-07-03 Tom de Vries <tom@codesourcery.com>
24966 PR tree-optimization/81192
24967 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24968 BB_SAME_SUCC (bb) == NULL.
24970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24972 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24975 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24977 * dumpfile.c: Include profile-count.h
24978 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24980 (insert_cond_bb): Update profile.
24981 * tree-cfg.h (insert_cond_bb): Update prototype.
24982 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24983 * tree-dump.c: Do not include tree-cfg.
24985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24987 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24989 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24991 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24994 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24996 * tree-complex.c (expand_complex_div_wide): update profile.
24998 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24999 Alan Hayward <alan.hayward@arm.com>
25000 David Sherwood <david.sherwood@arm.com>
25002 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25003 (CORETYPES_H): New define.
25004 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25005 (insn-modes-inline.h, s-modes-inline-h): New rules.
25006 (generated_files): Add insn-modes-inline.h.
25007 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25008 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25009 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25010 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25011 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25012 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25013 (build/gencodes.o, build/genconditions.o): Likewise.
25014 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25015 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25016 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25017 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25018 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25019 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25020 * coretypes.h: Include everything up to real.h for generators.
25021 Include insn-modes.h first. Include wide-int-print.h after
25022 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25023 * machmode.h: Don't include insn-modes.h here.
25024 * function-tests.c: Remove includes of signop.h, machmode.h,
25025 double-int.h and wide-int.h.
25027 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25029 * optc-save-gen.awk: Likewise.
25030 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25031 * godump.c: Remove include of wide-int-print.h.
25032 * pretty-print.h: Likewise.
25033 * wide-int-print.cc: Likewise.
25034 * wide-int.cc: Likewise.
25035 * hash-map-tests.c: Remove include of signop.h.
25036 * hash-set-tests.c: Likewise.
25037 * rtl-tests.c: Likewise.
25038 * mkconfig.sh: Remove include of machmode.h.
25039 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25041 (emit_insn_modes_inline_h): ...this new function. Emit the code
25042 into an insn-modes-inline.h header file, adding appropriate
25043 include guards and end comments.
25044 (emit_insn_modes_c_header): Remove include of machmode.h.
25045 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25047 (main): Handle -i flag and call emit_insn_modes_inline_h when
25050 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25052 * tree-ssa-strlen.c (strinfo): Rename the length field to
25053 nonzero_chars. Add a full_string_p field.
25054 (compare_nonzero_chars, zero_length_string_p): New functions.
25055 (get_addr_stridx): Add an offset_out parameter.
25056 Use compare_nonzero_chars.
25057 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25058 (new_strinfo): Update after above changes to strinfo.
25059 (set_endptr_and_length): Set full_string_p.
25060 (get_string_length): Update after above changes to strinfo.
25061 (unshare_strinfo): Update call to new_strinfo.
25062 (maybe_invalidate): Likewise.
25063 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25064 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25065 as a uhwi instead of an shwi. Update after above changes to
25066 strinfo and new_strinfo.
25067 (zero_length_string): Assert that chainsi contains full strings.
25068 Use zero_length_string_p. Update call to new_strinfo.
25069 (adjust_related_strinfos): Update after above changes to strinfo.
25070 Copy full_string_p from origsi.
25071 (adjust_last_stmt): Use zero_length_string_p.
25072 (handle_builtin_strlen): Update after above changes to strinfo and
25073 new_strinfo. Install the lhs as the string length if the previous
25074 entry didn't describe a full string.
25075 (handle_builtin_strchr): Update after above changes to strinfo
25077 (handle_builtin_strcpy): Likewise.
25078 (handle_builtin_strcat): Likewise.
25079 (handle_builtin_malloc): Likewise.
25080 (handle_pointer_plus): Likewise.
25081 (handle_builtin_memcpy): Likewise. Track nonzero characters
25082 that aren't necessarily followed by a nul terminator.
25083 (handle_char_store): Likewise.
25085 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25087 PR tree-optimization/80769
25088 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25089 for malloc and calloc. Document the new invariant that all related
25090 strinfos have delayed lengths or none do.
25091 (verify_related_strinfos): Move earlier in file.
25092 (set_endptr_and_length): New function, split out from...
25093 (get_string_length): ...here. Also set the lengths of related
25095 (zero_length_string): Assert that chainsi has known (rather than
25097 (adjust_related_strinfos): Likewise.
25099 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25101 PR tree-optimization/81136
25102 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25103 assert that two references with the same misalignment have the same
25104 compile-time misalignment if those compile-time misalignments
25107 2017-07-01 Andi Kleen <ak@linux.intel.com>
25109 * print-tree.c (print_node): Print all attributes.
25111 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25113 * cfg.c (scale_bbs_frequencies): New function.
25114 * cfg.h (scale_bbs_frequencies): Declare it.
25115 * cfgloopanal.c (single_likely_exit): Cleanup.
25116 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25118 (scale_loop_profile): Likewise.
25119 (loop_version): Likewise.
25120 (create_empty_loop_on_edge): Update.
25121 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25122 scale_loop_frequencies, scale_loop_profile, loopify,
25123 loop_version): Update prototypes.
25124 * modulo-sched.c (sms_schedule): Update.
25125 * predict.c (unlikely_executed_edge_p): Also check probability.
25126 (probably_never_executed_edge_p): Fix typo.
25127 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25128 * tree-parloops.c (gen_parallel_loop): Update.
25129 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25130 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25131 * tree-ssa-loop-split.c (split_loop): Update.
25132 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25133 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25134 (vect_loop_versioning): Update.
25135 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25139 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25141 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25143 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25144 to keep profile consistent.
25146 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25148 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25149 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25150 * profile-count.h (max_safe_multiplier): Make unsigned.
25151 (profile_count::guessed_zero): New.
25153 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25155 * bb-reorder.c (fix_up_crossing_landing_pad,
25156 fix_crossing_conditional_branches): Use make_single_succ_edge
25157 to keep profile consistent.
25159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25161 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25164 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25167 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25168 the right scopes, make sure cond_jump isn't preserved between multiple
25169 iterations. Search for fallthru edge whenever there are 3+ edges and
25170 use find_fallthru_edge for it.
25172 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25174 Patch by Alexander Monakov <amonakov@ispras.ru>
25175 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25176 probabilities consistently.
25178 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25180 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25181 emit_cmp_and_jump_insns.
25183 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25186 * tree-inline.c (expand_call_inline): Combine profile statuses.
25188 2017-06-30 Andrew Pinski <apinski@cavium.com>
25190 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25191 fold_stmt returned true.
25193 2017-06-30 Nathan Sidwell <nathan@acm.org>
25195 * ggc.h (empty_string): Delete.
25196 * cfgexpand.c (expand_asm_stmt): Use plain "".
25197 * optabs.c (expand_asm_memory_barrier): Likewise.
25198 * stringpool.c (empty_string): Delete.
25199 (digit_vector, digit_string): Delete.
25200 (ggc_alloc_string): Use plain "", don't optimize single digit
25201 strings. Use ggc_alloc_atomic.
25203 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25205 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25206 comparison set and one other set, use the cost of the non-comparison
25209 2017-06-30 Nathan Sidwell <nathan@acm.org>
25211 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25214 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25216 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25217 loops. Remove now unneeded calls to gimple_switch_set_label() that
25218 just set removed labels to NULL_TREE.
25220 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25222 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25223 (set_range_info): ...here. Only call set_range_info_raw if domain
25225 (set_nonzero_bits): Call set_range_info_raw.
25226 * tree-ssanames.h (set_range_info_raw): New.
25228 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25231 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25232 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25233 of nonimmediate_operand and <store_mask_constraint> instead of m
25234 for the input operand. For V8FI iterator, always split if input
25235 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25236 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25237 <store_mask_predicate> instead of register_operand and
25238 <store_mask_constraint> instead of v for the input operand. Make
25239 sure both operands aren't MEMs for if not <mask_applied>.
25241 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25243 * lto-wrapper.c (copy_file) Close both file descriptors before
25246 2017-06-30 Martin Liska <mliska@suse.cz>
25249 * multiple_target.c (create_dispatcher_calls): Make ifunc
25250 also for function that don't have calls or are not referenced.
25252 2017-06-30 Richard Biener <rguenther@suse.de>
25254 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25255 analyze the first scalar stmt. Move vector type computation
25256 for the BB case here from ...
25257 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25258 live operation processing in the SLP case properly.
25260 2017-06-30 Richard Biener <rguenther@suse.de>
25262 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25264 2017-06-30 Martin Liska <mliska@suse.cz>
25267 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25268 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25270 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25272 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25275 2017-06-30 Richard Biener <rguenther@suse.de>
25277 PR tree-optimization/81249
25278 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25279 condition reduction result to original scalar type.
25281 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25283 * profile-count.h (enum profile_quality): Fix typos and whitespace
25286 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25288 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25289 type for branch probabilities.
25291 2017-06-29 Julian Brown <julian@codesourcery.com>
25292 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25294 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25295 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25296 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25297 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25299 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25301 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25302 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25303 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25304 CC usage from generic code to here.
25305 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25306 CC usage into the target macros.
25308 2017-06-29 Maya Rashish <coypu@sdf.org>
25310 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25313 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25315 * arm/arm-builtins.c: Include profile-count.h
25316 * except.c (sjlj_emit_function_enter): Use
25317 profile_probability::unlikely.
25319 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25321 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25322 and tocrel_offset be pointer args rather than implicitly using
25324 (legitimate_constant_pool_address_p, rs6000_emit_move,
25325 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25326 tocrel_offset and use in toc_relative_expr_p call.
25327 (print_operand, print_operand_address): Use static tocrel_base_oac
25328 and tocrel_offset_oac.
25329 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25332 2017-06-29 Maya Rashish <coypu@sdf.org>
25334 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25336 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25338 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25339 objects, take into account only the alignment of 'op0' and 'mode1' if
25342 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25344 * ccmp.c (ccmp_tree_comparison_p): New function.
25345 (ccmp_candidate_p): Update to use above function.
25346 (get_compare_parts): New function.
25347 (expand_ccmp_next): Update to use new functions.
25348 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25350 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25351 take mode as argument.
25352 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25353 * expr.c (expand_expr_real_1): Pass mode as argument.
25355 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25357 * combine.c (combine_instructions): Print insns to dump_file, together
25360 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25362 * asan.c (asan_emit_stack_protection): Update.
25363 (create_cond_insert_point): Update.
25364 * auto-profile.c (afdo_propagate_circuit): Update.
25365 * basic-block.h (struct edge_def): Turn probability to
25366 profile_probability.
25367 (EDGE_FREQUENCY): Update.
25368 * bb-reorder.c (find_traces_1_round): Update.
25369 (better_edge_p): Update.
25370 (sanitize_hot_paths): Update.
25371 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25372 (make_single_succ_edge): Update.
25373 (check_bb_profile): Update.
25374 (dump_edge_info): Update.
25375 (update_bb_profile_for_threading): Update.
25376 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25377 probabilitycount to 0.
25378 * cfgbuild.c (compute_outgoing_frequencies): Update.
25379 * cfgcleanup.c (try_forward_edges): Update.
25380 (outgoing_edges_match): Update.
25381 (try_crossjump_to_edge): Update.
25382 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25383 (expand_gimple_tailcall): Update.
25384 (construct_init_block): Use make_single_succ_edge.
25385 (construct_exit_block): Use make_single_succ_edge.
25386 * cfghooks.c (verify_flow_info): Update.
25387 (redirect_edge_succ_nodup): Update.
25388 (split_edge): Update.
25389 (account_profile_record): Update.
25390 * cfgloopanal.c (single_likely_exit): Update.
25391 * cfgloopmanip.c (scale_loop_profile): Update.
25392 (set_zero_probability): Remove.
25393 (duplicate_loop_to_header_edge): Update.
25394 * cfgloopmanip.h (loop_version): Update prototype.
25395 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25396 (force_nonfallthru_and_redirect): Update.
25397 (update_br_prob_note): Update.
25398 (rtl_verify_edges): Update.
25399 (purge_dead_edges): Update.
25400 (rtl_lv_add_condition_to_bb): Update.
25401 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25402 * cgraphunit.c (init_lowered_empty_function): Update.
25403 (cgraph_node::expand_thunk): Update.
25404 * cilk-common.c: Include profile-count.h
25405 * dojump.c (inv): Remove.
25406 (jumpifnot): Update.
25407 (jumpifnot_1): Update.
25408 (do_jump_1): Update.
25410 (do_jump_by_parts_greater_rtx): Update.
25411 (do_compare_rtx_and_jump): Update.
25412 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25413 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25414 * dwarf2cfi.c: Include profile-count.h
25415 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25416 (sjlj_emit_dispatch_table): Likewise.
25417 * explow.c: Include profile-count.h
25418 * expmed.c (emit_store_flag_force): Update.
25419 (do_cmp_and_jump): Update.
25420 * expr.c (compare_by_pieces_d::generate): Update.
25421 (compare_by_pieces_d::finish_mode): Update.
25422 (emit_block_move_via_loop): Update.
25423 (store_expr_with_bounds): Update.
25424 (store_constructor): Update.
25425 (expand_expr_real_2): Update.
25426 (expand_expr_real_1): Update.
25427 * expr.h (try_casesi, try_tablejump): Update prototypes.
25428 * gimple-pretty-print.c (dump_probability): Update.
25429 (dump_profile): New.
25430 (dump_gimple_label): Update.
25431 (dump_gimple_bb_header): Update.
25432 * graph.c (draw_cfg_node_succ_edges): Update.
25433 * hsa-gen.c (convert_switch_statements): Update.
25434 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25435 (find_if_case_1): Update.
25436 (find_if_case_2): Update.
25437 * internal-fn.c (expand_arith_overflow_result_store): Update.
25438 (expand_addsub_overflow): Update.
25439 (expand_neg_overflow): Update.
25440 (expand_mul_overflow): Update.
25441 (expand_vector_ubsan_overflow): Update.
25442 * ipa-cp.c (good_cloning_opportunity_p): Update.
25443 * ipa-split.c (split_function): Use make_single_succ_edge.
25444 * ipa-utils.c (ipa_merge_profiles): Update.
25445 * loop-doloop.c (add_test): Update.
25446 (doloop_modify): Update.
25447 * loop-unroll.c (compare_and_jump_seq): Update.
25448 (unroll_loop_runtime_iterations): Update.
25449 * lra-constraints.c (lra_inheritance): Update.
25450 * lto-streamer-in.c (input_cfg): Update.
25451 * lto-streamer-out.c (output_cfg): Update.
25452 * mcf.c (adjust_cfg_counts): Update.
25453 * modulo-sched.c (sms_schedule): Update.
25454 * omp-expand.c (expand_omp_for_init_counts): Update.
25455 (extract_omp_for_update_vars): Update.
25456 (expand_omp_ordered_sink): Update.
25457 (expand_omp_for_ordered_loops): Update.
25458 (expand_omp_for_generic): Update.
25459 (expand_omp_for_static_nochunk): Update.
25460 (expand_omp_for_static_chunk): Update.
25461 (expand_cilk_for): Update.
25462 (expand_omp_simd): Update.
25463 (expand_omp_taskloop_for_outer): Update.
25464 (expand_omp_taskloop_for_inner): Update.
25465 * omp-simd-clone.c (simd_clone_adjust): Update.
25466 * optabs.c (expand_doubleword_shift): Update.
25467 (expand_abs): Update.
25468 (emit_cmp_and_jump_insn_1): Update.
25469 (expand_compare_and_swap_loop): Update.
25470 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25471 * predict.c (predictable_edge_p): Update.
25472 (edge_probability_reliable_p): Update.
25473 (set_even_probabilities): Update.
25474 (combine_predictions_for_insn): Update.
25475 (combine_predictions_for_bb): Update.
25476 (propagate_freq): Update.
25477 (estimate_bb_frequencies): Update.
25478 (force_edge_cold): Update.
25479 * profile-count.c (profile_count::dump): Add missing space into dump.
25480 (profile_count::debug): Add newline.
25481 (profile_count::differs_from_p): Explicitly convert to unsigned.
25482 (profile_count::stream_in): Update.
25483 (profile_probability::dump): New member function.
25484 (profile_probability::debug): New member function.
25485 (profile_probability::differs_from_p): New member function.
25486 (profile_probability::differs_lot_from_p): New member function.
25487 (profile_probability::stream_in): New member function.
25488 (profile_probability::stream_out): New member function.
25489 * profile-count.h (profile_count_quality): Rename to ...
25490 (profile_quality): ... this one.
25491 (profile_probability): New.
25492 (profile_count): Update.
25493 * profile.c (compute_branch_probabilities): Update.
25494 * recog.c (peep2_attempt): Update.
25495 * sched-ebb.c (schedule_ebbs): Update.
25496 * sched-rgn.c (find_single_block_region): Update.
25497 (compute_dom_prob_ps): Update.
25498 (schedule_region): Update.
25499 * sel-sched-ir.c (compute_succs_info): Update.
25500 * stmt.c (struct case_node): Update.
25501 (do_jump_if_equal): Update.
25502 (get_outgoing_edge_probs): Update.
25503 (conditional_probability): Update.
25504 (emit_case_dispatch_table): Update.
25505 (expand_case): Update.
25506 (expand_sjlj_dispatch_table): Update.
25507 (emit_case_nodes): Update.
25508 * targhooks.c: Update.
25509 * tracer.c (better_p): Update.
25510 (find_best_successor): Update.
25511 * trans-mem.c (expand_transaction): Update.
25512 * tree-call-cdce.c: Update.
25513 * tree-cfg.c (gimple_split_edge): Upate.
25514 (move_sese_region_to_fn): Upate.
25515 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25516 * tree-eh.c (lower_resx): Upate.
25517 (cleanup_empty_eh_move_lp): Upate.
25518 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25519 * tree-inline.c (copy_edges_for_bb): Update.
25520 (copy_cfg_body): Update.
25521 * tree-parloops.c (gen_parallel_loop): Update.
25522 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25523 (gimple_gen_time_profiler): Update.
25524 * tree-ssa-dce.c (remove_dead_stmt): Update.
25525 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25526 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25527 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25528 (unloop_loops): Update.
25529 (try_peel_loop): Update.
25530 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25531 * tree-ssa-loop-split.c (connect_loops): Update.
25532 (split_loop): Update.
25533 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25534 (hoist_guard): Update.
25535 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25536 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25537 (value_replacement): Update.
25538 * tree-ssa-reassoc.c (branch_fixup): Update.
25539 * tree-ssa-tail-merge.c (replace_block_by): Update.
25540 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25541 (create_edge_and_update_destination_phis): Update.
25542 (compute_path_counts): Update.
25543 (recompute_probabilities): Update.
25544 (update_joiner_offpath_counts): Update.
25545 (freqs_to_counts_path): Update.
25546 (duplicate_thread_path): Update.
25547 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25548 (struct switch_conv_info): Update.
25549 (gen_inbound_check): Update.
25550 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25551 (vect_do_peeling): Update.
25552 (vect_loop_versioning): Update.
25553 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25554 (optimize_mask_stores): Update.
25555 * ubsan.c (ubsan_expand_null_ifn): Update.
25556 * value-prof.c (gimple_divmod_fixed_value): Update.
25557 (gimple_divmod_fixed_value_transform): Update.
25558 (gimple_mod_pow2): Update.
25559 (gimple_mod_pow2_value_transform): Update.
25560 (gimple_mod_subtract): Update.
25561 (gimple_mod_subtract_transform): Update.
25562 (gimple_ic): Update.
25563 (gimple_stringop_fixed_value): Update.
25564 (gimple_stringops_transform): Update.
25565 * value-prof.h: Update.
25567 2017-06-29 Carl Love <cel@us.ibm.com>
25569 * config/rs6000/rs6000-c.c: Add support for built-in functions
25570 vector signed int vec_signed (vector float);
25571 vector signed long long vec_signed (vector double);
25572 vector signed int vec_signed2 (vector double, vector double);
25573 vector signed int vec_signede (vector double);
25574 vector signed int vec_signedo (vector double);
25575 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25576 instruction generator.
25577 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25578 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25579 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25581 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25582 vunsignede_v2df): Add define_expands.
25583 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25584 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25585 VEC_UNSIGNEDO): Add definitions.
25586 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25587 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25588 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25589 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25590 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25591 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25592 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25593 vec_unsignede, vec_unsignedo): Add builtin defines.
25594 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25596 * doc/extend.texi: Update the built-in documentation file for the
25597 new built-in functions.
25599 2017-06-29 Richard Biener <rguenther@suse.de>
25601 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25602 reduction chains to LOOP_VINFO_REDUCTIONS.
25603 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25604 SLP reductions after processing reduction chains.
25606 2017-06-29 Nathan Sidwell <nathan@acm.org>
25608 * builtins.c (fold_builtin_FUNCTION): Use
25609 lang_hooks.decl_printable_name.
25611 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25613 PR middle-end/81194
25614 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25615 with only one label.
25616 * stmt.c (expand_case): Assert NCASES is greater than one.
25618 2017-06-29 Richard Biener <rguenther@suse.de>
25620 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25622 (group_case_labels): Likewise.
25623 (find_taken_edge): Push sanity checking on val to workers...
25624 (find_taken_edge_cond_expr): ... here
25625 (find_taken_edge_switch_expr): ... and here, handle cases
25626 with just a default label.
25627 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25628 (group_case_labels): Likewise.
25629 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25630 group_case_labels does anything cleanup the CFG again.
25632 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25634 PR tree-optimization/81196
25635 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25636 exit condition comparing two IVs.
25638 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25640 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25641 profile to the dummy entry at the end of the list of architectures.
25642 * config/arm/arm-cpu-cdata.h: Regenerated.
25644 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25645 Michael Collison <michael.collison@arm.com>
25648 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25650 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25651 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25652 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25653 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25654 of shift when the shift amount is masked with constant equal to
25655 the size of the mode.
25656 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25659 2017-06-29 Martin Liska <mliska@suse.cz>
25661 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25663 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25666 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25669 (edge_growth_cache_entry::edge_growth_cache_entry): New
25671 (reset_edge_growth_cache): Update to use constructor.
25673 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25675 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25676 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25677 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25679 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25681 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25682 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25684 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25686 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25687 (*-linux-uclibc*): Add t-uclibc tmake_file.
25688 * config/t-musl: New.
25689 * config/t-uclibc: New.
25691 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25693 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25695 (gen_comm_data): Emit architectural setting of arch_prof.
25696 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25698 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25699 (armv8-m.base, armv8-m.main): Likewise.
25700 * arm-protos.h (arm_build_target): Add profile field.
25701 (arch_option): Likewise.
25702 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25704 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25705 arm_active_target.profile.
25707 2017-06-28 Richard Biener <rguenther@suse.de>
25709 PR middle-end/81227
25710 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25711 TYPE_OVERFLOW_WRAPS.
25712 * match.pd (negate_expr_p): Likewise.
25713 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25714 fold_build2, not fold_binary.
25716 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25718 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25719 Convert memory address to Pmode.
25720 (aarch64_print_operand): Assert MEM operands are always Pmode.
25722 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25725 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25726 Remove redundant if.
25727 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25728 * config/arm/aarch-common-protos.h
25729 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25730 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25732 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25735 * multiple_target.c (create_dispatcher_calls): Set the default
25736 clone to be static, not public.
25738 2017-06-28 Richard Biener <rguenther@suse.de>
25740 * tree-vect-loop.c (vectorizable_reduction): Move special
25741 cond reduction IV var creation ...
25742 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25743 parameter. Use STMT_VINFO_VECTYPE.
25744 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25747 2017-06-28 Martin Liska <mliska@suse.cz>
25750 * ipa-visibility.c (non_local_p): Handle visibility.
25752 2017-06-28 Martin Liska <mliska@suse.cz>
25755 * common.opt: Add IntegerRange to various options.
25756 * opt-functions.awk (integer_range_info): New function.
25757 * optc-gen.awk: Add integer_range_info to cl_options struct.
25758 * opts-common.c (decode_cmdline_option): Handle
25759 CL_ERR_INT_RANGE_ARG.
25760 (cmdline_handle_error): Likewise.
25761 * opts.c (print_filtered_help): Show valid interval in
25762 when --help is provided.
25763 * opts.h (struct cl_option): Add range_min and range_max fields.
25764 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25766 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25768 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25769 (x * C EQ/NE y * C): New transformation.
25771 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25773 * genmultilib (combination_space): Accept '+' in option names.
25775 2017-06-28 Martin Liska <mliska@suse.cz>
25778 * asan.c (instrument_derefs): Bail out inner references
25779 that are hard register variables.
25781 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25784 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25785 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25787 2017-06-28 Richard Biener <rguenther@suse.de>
25789 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25790 (vect_get_slp_defs): Adjust.
25791 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25793 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25795 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25796 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25797 (vectorizable_reduction): Adjust.
25798 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25800 (vect_get_slp_defs): Likewise.
25801 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25802 (vectorizable_bswap): Adjust.
25803 (vectorizable_call): Likewise.
25804 (vectorizable_conversion): Likewise.
25805 (vectorizable_assignment): Likewise.
25806 (vectorizable_shift): Likewise.
25807 (vectorizable_operation): Likewise.
25808 (vectorizable_store): Likewise.
25809 (vectorizable_condition): Likewise.
25810 (vectorizable_comparison): Likewise.
25812 2017-06-28 Michael Collison <michael.collison@arm.com>
25815 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25817 (arm_gen_movmemqi): Removed unused variable 'i'.
25818 Convert 'for' loop into 'while' loop.
25819 (arm_expand_prologue): Remove last unnecessary set of insn.
25820 (thumb_pop): Remove unused variable 'pushed_words'.
25821 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25823 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25825 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25826 * config/s390/s390-protos.h: Add prototype of
25827 s390_rel_address_ok_p.
25828 * config/s390/s390.c (s390_got_symbol): New function.
25829 (s390_rel_address_ok_p): New function.
25830 (legitimize_pic_address): Use s390_rel_address_ok_p.
25831 (s390_load_got): Use s390_got_symbol.
25832 (s390_option_override): Issue error if
25833 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25834 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25836 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25838 2017-06-27 Andrew Pinski <apinski@cavium.com>
25840 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25841 (X * copysign (1.0, X)): New pattern.
25842 (X * copysign (1.0, -X)): New pattern.
25843 (copysign (-1.0, CST)): New pattern.
25845 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25847 * genmultilib (combination_space): Remove variable.
25848 Validate reuse rules against regular expression for any sequence
25849 of multilib options in any order.
25851 2017-06-27 Michael Collison <michael.collison@arm.com>
25853 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25854 call aarch64_split_simd_combine.
25855 * (aarch64_combine_internal<mode>): Delete pattern.
25856 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25857 Allow register and subreg operands.
25859 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25861 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25862 specific need, just fallback on defaults.
25863 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25865 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25866 Olivier Hainque <hainque@adacore.com>
25868 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25870 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25871 targets. Pick a default if no particular attempt applied.
25872 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25875 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25877 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25878 (x86_64-wrs-vxworks7): Likewise.
25880 2017-06-27 Marek Polacek <polacek@redhat.com>
25883 * ubsan.c (instrument_null): Check get_base_address's result for null.
25885 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25887 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25889 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25891 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25892 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25893 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25894 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25895 New function types.
25896 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25897 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25898 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25899 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25900 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25901 BUILT_IN_FEUPDATEENV): New builtins.
25902 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25903 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25904 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25905 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25907 (builtin_structptr_types): Adjust size.
25908 * tree.c (builtin_structptr_types): Add four entries.
25910 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25911 Olivier Hainque <hainque@adacore.com>
25913 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25914 (TLS_SYM): New local macro, forcing reference to __tls__ on
25915 link command lines for VxWorks 7 RTPs, triggering initialization
25917 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25918 OS features TLS support, true for RTPs on VxWorks 7.
25919 * config/vxworks.c (vxworks_override_options): Setup emutls
25922 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25924 * predict.c (test_prediction_value_range): Use -1U instead of -1
25925 to avoid narrowing conversion warning.
25926 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25927 to avoid narrowing conversion warning.
25928 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25930 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25932 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25934 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25935 64bit configurations.
25936 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25937 (SIZE_TYPE): Likewise.
25938 * config/vxworks.c (vxworks_emutls_var_fields): Use
25939 long_unsigned_type_node instead of unsigned_type_node as the offset
25940 field type, which is "pointer" mode in emutls.c.
25942 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25945 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25947 PR middle-end/81207
25948 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25949 gimple_vuse copying separately from gimple_vdef copying.
25951 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25953 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25956 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25957 Olivier Hainque <hainque@adacore.com>
25959 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25960 for all vxworks7 targets.
25961 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25962 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25963 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25964 variations for VX6/VX7 and 32/64bits later on in ...
25965 (VXWORKS_LIB_SPEC): Leverage new macros.
25966 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25967 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25969 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25971 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25972 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25974 2017-06-26 Carl Love <cel@us.ibm.com>
25976 * config/rs6000/rs6000-c.c: Add support for built-in functions
25977 vector bool char vec_reve (vector bool char);
25978 vector signed char vec_reve (vector signed char);
25979 vector unsigned char vec_reve (vector unsigned char);
25980 vector bool int vec_reve (vector bool int);
25981 vector signed int vec_reve (vector signed int);
25982 vector unsigned int vec_reve (vector unsigned int);
25983 vector bool long long vec_reve (vector bool long long);
25984 vector signed long long vec_reve (vector signed long long);
25985 vector unsigned long long vec_reve (vector unsigned long long);
25986 vector bool short vec_reve (vector bool short);
25987 vector signed short vec_reve (vector signed short);
25988 vector double vec_reve (vector double);
25989 vector float vec_reve (vector float);
25990 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25991 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25992 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25993 (altivec_vreve): New pattern.
25994 * config/rs6000/altivec.h (vec_reve): New define.
25995 * doc/extend.texi (vec_rev): Update the built-in documentation file
25996 for the new built-in functions.
25998 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26000 PR tree-optimization/71815
26001 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26003 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26005 (slsr_process_phi): Likewise.
26006 (replace_uncond_cands_and_profitable_phis): Don't replace a
26007 multiply candidate with a stride of 1 (copy or cast).
26008 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26010 (lowest_cost_path): Likewise.
26011 (total_savings): Likewise.
26013 2017-06-26 Richard Biener <rguenther@suse.de>
26016 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26017 Use def_builtin_pure for all gather builtins.
26019 2017-06-26 Richard Biener <rguenther@suse.de>
26021 PR tree-optimization/81203
26022 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26023 non-dominating BBs.
26025 2017-06-26 Marek Polacek <polacek@redhat.com>
26028 * doc/invoke.texi: Document -Wmultistatement-macros.
26030 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26032 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26033 arm_neon_ok_no_float_abi effective target.
26035 2017-06-26 Richard Biener <rguenther@suse.de>
26037 PR tree-optimization/80928
26038 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26039 (copy_bbs): Set BB_DUPLICATED flag early.
26040 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26042 (execute_on_shrinking_pred): Likewise.
26043 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26044 BB_DUPLICATED blocks.
26045 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26046 iterate over all PHIs considering removal of *gsi.
26048 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26050 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26053 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26055 * config/rs6000/rs6000-string.c: (expand_block_clear,
26056 do_load_for_compare, select_block_compare_mode,
26057 compute_current_alignment, expand_block_compare,
26058 expand_strncmp_align_check, expand_strn_compare,
26059 expand_block_move, rs6000_output_load_multiple)
26060 Move functions related to string/block move/compare
26061 to a separate file.
26062 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26063 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26064 for this function which is now used in two files.
26065 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26066 * config.gcc: Add rs6000-string.o to extra_objs for
26067 targets powerpc*-*-* and rs6000*-*-*.
26069 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26072 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26073 32-bit, since indexed is not valid for DImode.
26074 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26075 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26076 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26077 (define_peephole2 for Altivec d-form store): Likewise.
26080 * multiple_target.c (create_dispatcher_calls): Only create the
26081 dispatcher call if the function is the default clone of a
26082 versioned function.
26084 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26086 PR middle-end/80902
26087 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26088 a call, force the call to not be a tail call.
26090 2017-06-23 Jeff Law <law@redhat.com>
26092 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26095 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26097 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26098 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26099 handling for early expansion of vector shifts (sl,sr,sra,rl).
26100 (builtin_function_type): Add vector shift right instructions
26101 to the unsigned argument list.
26103 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26105 rtl-optimizatoin/79286
26106 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26107 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26108 trap. PIC register plus a const unspec without offset can never trap.
26110 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26112 * tree.h (builtin_structptr_type): New type.
26113 (builtin_structptr_types): Declare new array.
26114 * tree.c (builtin_structptr_types): New array.
26115 (free_lang_data, build_common_tree_nodes): Use it.
26117 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26120 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26123 2017-06-22 Matt Turner <mattst88@gmail.com>
26125 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26126 Lake models to skylake case. Assume skylake for unknown
26127 models with clflushopt.
26129 2017-06-22 Jeff Law <law@redhat.com>
26131 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26132 frame sizes that do not satisfy aarch64_uimm12_shift.
26134 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26136 * profile-count.h (apply_probability,
26137 apply_scale, probability_in): Fix checks for zero.
26139 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26141 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26142 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26144 2016-06-22 Richard Biener <rguenther@suse.de>
26146 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26147 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26148 REDUC_MAX_EXPR support.
26149 (vectorizable_reduction): Likewise.
26150 (vect_create_epilog_for_reduction): Likewise.
26152 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26154 * match.pd (A / (1 << B) -> A >> B): New.
26155 * generic-match-head.c: Include optabs-tree.h.
26156 * gimple-match-head.c: Likewise.
26157 * optabs-tree.h (target_supports_op_p): New.
26158 * optabs-tree.c (target_supports_op_p): New.
26160 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26162 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26163 $gcc_cv_ld --help output.
26164 (gcc_cv_ld_demangle): Likewise.
26165 (gcc_cv_ld_eh_frame_hdr): Likewise.
26166 (gcc_cv_ld_pie): Likewise.
26167 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26168 (gcc_cv_ld_buildid): Likewise.
26169 (gcc_cv_ld_sysroot): Likewise.
26170 (ld_bndplt_support): Likewise.
26171 (ld_pushpopstate_support): Likewise.
26172 * configure: Regenerate.
26173 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26175 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26178 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26179 operands indexes to avoid gap between operands and match_dups.
26181 2017-06-21 Andrew Pinski <apinski@cavium.com>
26183 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26184 Increment Arith_shift and Arith_shift_reg by 1.
26185 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26187 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26188 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26189 (aarch64_strip_extend): Add new argument and test for it.
26190 (aarch64_cheap_mult_shift_p): New function.
26191 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26192 add a cost if it is true.
26193 Update calls to aarch64_strip_extend.
26194 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26196 2017-06-21 Andrew Pinski <apinski@cavium.com>
26198 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26200 (thunderxt88): Likewise.
26201 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26202 (thunderx_prefetch_tune): New variable.
26203 (thunderx2t99_prefetch_tune): Update for the correct values.
26204 (thunderxt88_tunings): New variable.
26205 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26206 generic_prefetch_tune.
26207 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26209 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26211 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26212 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26213 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26214 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26215 (aarch64_atomic_cas<mode>, GPI): Likewise.
26217 2017-06-21 Martin Liska <mliska@suse.cz>
26219 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26220 statements on cold and hot labels.
26221 * predict.c (tree_estimate_probability_bb): Remove the
26222 prediction from this place.
26224 2017-06-21 Martin Liska <mliska@suse.cz>
26226 PR tree-optimization/79489
26227 * gimplify.c (maybe_add_early_return_predict_stmt): New
26229 (gimplify_return_expr): Call the function.
26230 * predict.c (tree_estimate_probability_bb): Remove handling
26232 * predict.def: Update comment about early return predictor.
26233 * gimple-predict.h (is_gimple_predict): New function.
26234 * predict.def: Change default value of early return to 66.
26235 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26237 * passes.def: Put pass_strip_predict_hints to the beginning of
26240 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26242 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26243 FUNCTION_DECL declarations.
26244 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26246 (dwaf2out_decl): Likewise.
26247 * godump.c (go_early_global_decl): Skip call to the real debug hook
26248 for FUNCTION_DECL declarations.
26249 * passes.c (rest_of_decl_compilation): Skip call to the
26250 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26251 -fdump-go-spec is passed.
26253 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26255 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26256 Reorder for compactness.
26257 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26258 (def_builtin_pure, def_builtin_pure2): New functions.
26259 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26261 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26263 * match.pd (nop_convert): New predicate.
26264 ((A +- CST1) +- CST2): Allow some NOP conversions.
26266 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26269 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26270 with ctors/dtors if GOVD_SHARED is set.
26272 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26274 * config/aarch64/aarch64.md (movti_aarch64):
26275 Emit mov rather than orr.
26276 (movtf_aarch64): Likewise.
26277 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26278 Emit mov rather than orr.
26280 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26282 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26283 Swap alternatives, make integer dup more expensive.
26285 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26287 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26288 Return true for non-tls symbols.
26290 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26292 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26293 (cortex-a75): Likewise.
26294 (cortex-a75.cortex-a55): Likewise.
26295 * config/aarch64/aarch64-tune.md: Regenerate.
26296 * doc/invoke.texi (-mtune): Document new values for -mtune.
26298 2017-06-21 Tom de Vries <tom@codesourcery.com>
26300 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26301 stack_size feature.
26302 (Effective-Target Keywords, Other attributes): Suggest using
26303 dg-add-options stack_size feature to get stack limit in stack_size
26304 effective target documentation.
26306 2017-06-21 Julian Brown <julian@codesourcery.com>
26307 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26309 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26310 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26311 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26313 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26314 attribute type list for neon_multiply.
26315 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26316 attribute type list for neon_multiply.
26317 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26318 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26319 attribute type list for neon_multiply.
26320 * config/arm/types.md (crypto_pmull): Add.
26321 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26322 attribute type list.
26324 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26326 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26329 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26331 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26332 to make sure not to dereference a NULL cost_classes_ptr pointer.
26334 2017-06-20 Carl Love <cel@us.ibm.com>
26336 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26337 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26338 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26340 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26341 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26342 VMULOSW): New enum "unspec" values.
26343 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26344 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26345 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26346 altivec_vmulosw): New patterns.
26347 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26348 VMULOSW): Add definitions.
26350 2017-06-20 Julia Koval <julia.koval@intel.com>
26352 * config/i386/i386.c: Fix rounding expand for new pattern.
26353 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26355 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26357 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26358 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26360 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26362 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26365 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26367 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26368 architecture, then by implementer ID.
26369 * config/aarch64/aarch64-tune.md: Regenerate.
26371 2017-06-20 Richard Biener <rguenther@suse.de>
26373 PR middle-end/81097
26374 * fold-const.c (split_tree): Fold to type before negating.
26376 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26378 * diagnostic-show-locus.c
26379 (selftest::test_fixit_deletion_affecting_newline): New function.
26380 (selftest::diagnostic_show_locus_c_tests): Call it.
26382 2017-06-20 Andreas Schwab <schwab@suse.de>
26385 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26388 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26390 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26391 __ARM_FEATURE_COPROC according to support.
26393 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26395 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26396 Rewritten to avoid overflow for > 32-bit pointers.
26399 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26400 by removing enum keyword.
26401 (ubsan_type_descriptor): Likewise. Formatting fix.
26404 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26405 splitter): Require TARGET_SSE2 in the condition.
26407 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26410 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26411 for doing vector set of SFmode on ISA 3.0.
26412 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26413 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26415 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26416 SFmode value into a V4SF variable that was extracted from another
26417 V4SF variable without converting the element to double precision
26418 and back to single precision vector format.
26419 (vsx_insert_extract_v4sf_p9_2): Likewise.
26421 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26423 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26424 in UWHI to avoid undefined overflow.
26427 * ubsan.h (enum ubsan_encode_value_phase): New.
26428 (ubsan_encode_value): Change second argument to
26429 enum ubsan_encode_value_phase with default value of
26430 UBSAN_ENCODE_VALUE_GENERIC.
26431 * ubsan.c (ubsan_encode_value): Change second argument to
26432 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26433 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26434 create_tmp_var_raw instead of create_tmp_var and use a
26436 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26437 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26438 ubsan_encode_value callers.
26441 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26442 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26443 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26445 2017-06-19 Richard Biener <rguenther@suse.de>
26447 PR middle-end/81118
26448 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26449 estimates if we changed anything.
26451 2017-06-19 Richard Biener <rguenther@suse.de>
26453 PR tree-optimization/80887
26454 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26455 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26456 simplified lookups, then reset mprts_hook.
26457 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26459 (try_to_simplify): Likewise.
26461 2017-06-19 Martin Liska <mliska@suse.cz>
26464 * gimplify.c (gimplify_switch_expr):
26465 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26467 2017-06-19 Martin Liska <mliska@suse.cz>
26469 * doc/install.texi: Document that PGO runs in 4 stages.
26471 2017-06-19 Martin Liska <mliska@suse.cz>
26474 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26475 to dispatcher function name.
26476 * multiple_target.c (replace_function_decl): New function.
26477 (create_dispatcher_calls): Redirect both edges and references.
26479 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26481 * profile-count.c (profile_count::dump): Dump quality.
26482 (profile_count::differs_from_p): Update for unsigned val.
26483 * profile-count.h (profile_count_quality): New enum.
26484 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26486 2017-06-19 Richard Biener <rguenther@suse.de>
26488 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26489 struct function as arg.
26490 (estimate_numbers_of_iterations): Export overload with loop arg.
26491 (free_numbers_of_iterations_estimates_loop): Use an overload of
26492 free_numbers_of_iterations_estimates instead.
26493 * tree-cfg.c (remove_bb): Adjust.
26494 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26495 * tree-parloops.c (gen_parallel_loop): Likewise.
26496 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26498 (tree_unroll_loops_completely): Likewise.
26499 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26500 Use an overload instead and export.
26501 (estimated_loop_iterations): Adjust.
26502 (max_loop_iterations): Likewise.
26503 (likely_max_loop_iterations): Likewise.
26504 (estimate_numbers_of_iterations): Take struct function as arg
26506 (loop_exits_before_overflow): Adjust.
26507 (free_numbers_of_iterations_estimates_loop): Use an overload.
26508 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26509 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26511 2017-06-19 Richard Biener <rguenther@suse.de>
26514 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26515 RANGE_EXPR conservatively.
26517 2017-06-16 Carl Love <cel@us.ibm.com>
26519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26520 definitions for vec_float, vec_float2, vec_floato,
26521 vec_floate built-ins.
26522 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26523 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26525 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26526 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26527 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26528 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26529 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26530 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26531 vec_floato): Add builtin defines.
26532 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26533 Update the built-in documentation file for the new built-in
26536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26538 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26539 (mthumb): Mark as the negative of -marm.
26541 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26543 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26545 (ARM Options, -mtune): Document that this accepts the same
26546 extension options as -mcpu.
26547 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26551 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26552 permitted extensions.
26554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26556 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26557 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26558 (armv8-m.main): Add option +nodsp.
26559 * config/arm/arm-cpu-cdata.h: Regenerated.
26561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26563 * config/arm/t-fuchsia: New file.
26564 * config.gcc (arm*-*-fuchsia*): Use it.
26566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26568 * config/arm/t-symbian: Rewrite for new option infrastructure.
26570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26572 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26573 (MULTILIB_REQUIRED): Likewise.
26575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26577 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26578 (MULTILIB_RESUE): Likewise.
26579 (MULTILIB_MATCHES): Likewise.
26580 (MULTLIB_REQUIRED): Likewise.
26582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26584 * config/arm/t-rtems: Rewrite for new option framework.
26586 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26588 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26589 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26590 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26591 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26592 * config/arm/t-multilib: ... here.
26593 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26594 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26595 armv7-a and armv8*-a when A-profile libraries have not been built.
26596 * config/arm/t-rmprofile: Rewrite.
26598 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26600 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26601 with a backslash. Remove the backslash after substituting unescaped
26603 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26607 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26608 the driver to pass through the default setting of -mfloat-abi.
26609 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26611 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26612 (all_feat_combs): New rule.
26613 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26615 * config/arm/t-aprofile: Rewrite.
26617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26619 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26620 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26621 fpu is not specified by the user/command-line.
26622 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26623 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26624 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26625 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26626 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26627 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26631 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26632 * config/arm/t-arm-elf: Rewritten.
26634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26636 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26637 have some floating-point instructions.
26638 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26639 (TARGET_MAYBE_HARD_FLOAT): New macro.
26640 * config/arm/arm-builtins.c (arm_init_builtins): Use
26641 TARGET_MAYBE_HARD_FLOAT.
26642 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26644 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26646 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26647 (configargs.h): Include it.
26648 (arm_print_hint_for_fpu_option): New function.
26649 (arm_parse_fpu_option): New function.
26650 (candidate_extension): New class.
26651 (arm_canon_for_multilib): New function.
26652 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26653 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26654 (ARCH_CANONICAL_SPECS): New macro.
26655 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26659 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26660 are set after handling multilib fragments. Set target_cpu_default2
26663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26665 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26667 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26668 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26670 (check_arch): Likewise.
26671 * config/arm/arm.c (arm_configure_build_target): Handle
26672 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26673 options in the default.
26675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26677 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26678 when an option is an alias of another.
26679 * config/arm/parsecpu.awk (optalias): New parser token.
26680 (gen_comm_data): Mark non-alias options as such. Emit entries
26681 for extension aliases.
26682 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26683 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26684 (armv6kz, armv6zk, armv6t2): Likewise.
26685 (armv7): Make vfpv3-d16 an alias.
26686 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26688 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26689 Sort in canonical order.
26690 (armv8-a): Sort in canonical order.
26691 (armv8.1-a, armv8.2-a): Likewise.
26692 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26694 (cortex-a9): Sort in canonical order.
26695 * config/arm/arm.c (selftests.h): Include it.
26696 (arm_test_cpu_arch_data): New function.
26697 (arm_run_self_tests): New function.
26698 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26699 (targetm): Move declaration to the end of the file.
26700 * arm-cpu-cdata.h: Regenerated.
26702 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26704 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26705 call to target_mode_check describing the type of option passed.
26706 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26707 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26708 arm_parse_cpu_option_name to match parameters against list of
26710 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26711 arm_arch_core_flags data structure.
26712 * config/arm/arm-cpu_cdata.h: Regenerated.
26714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26716 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26718 (arm_print_hint_for_cpu_option): Likewise.
26719 (arm_print_hint_for_arch_option): Likewise.
26720 (arm_parse_cpu_option_name): Likewise.
26721 (arm_parse_arch_option_name): Likewise.
26722 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26723 of entries in the all_fpus list.
26724 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26725 (arm_parse_cpu_option_name): Declare.
26726 (arm_parse_arch_option_name): Declare.
26727 (arm_parse_option_features): Declare.
26728 (arm_intialize_isa): Declare.
26729 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26731 (gen_comm_data): ... here. Make definitions non-static.
26732 * config/arm/arm-cpu-data.h: Regenerated.
26733 * config/arm/arm-cpu-cdata.h: Regenerated.
26735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26737 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26738 (cpu_arch_extension): New structure.
26739 (cpu_arch_option, arch_option, cpu_option): New structures.
26740 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26741 architecture types.
26742 (gen_data): Generate new format data tables.
26743 * config/arm/arm.c (cpu_tune): New structure.
26744 (cpu_option, processors): Delete.
26745 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26746 (arm_print_hint_for_cpu_option): ... this and ...
26747 (arm_print_hint_for_arch_option): ... this.
26748 (arm_parse_arch_cpu_name): Delete. Replace with ...
26749 (arm_parse_cpu_option_name): ... this and ...
26750 (arm_parse_arch_option_name): ... this.
26751 (arm_unrecognized_feature): Change type of target parameter to
26753 (arm_parse_arch_cpu_features): Delete. Replace with ...
26754 (arm_parse_option_features): ... this.
26755 (arm_configure_build_target): Rework to use new configuration data
26757 (arm_print_tune_info): Rework for new configuration data tables.
26758 * config/arm/arm-cpu-data.h: Regenerated.
26759 * config/arm/arm-cpu.h: Regenerated.
26761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26763 * Makefile.in (OBJS): Move sbitmap.o from here ...
26764 (OBJS-libcommon): ... to here.
26766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26768 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26769 (ISA_ALL_CRYPTO): New macro.
26770 (ISA_ALL_SIMD): New macro
26771 (ISA_ALL_FP): New macro.
26772 * config/arm/arm.c (fpu_bitlist): Update initializer.
26773 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26775 (arm9e): Add fpu. Add option for nofp
26776 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26777 (arm926ej-s, arm1026ej-s): Likewise.
26778 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26779 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26780 neon-fp16, neon-vfpv4, nofp and nosimd.
26781 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26782 (cortex-a8): Add fpu. Add option for nofp.
26783 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26784 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26785 (cortex-r4f): Add fpu.
26786 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26787 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26789 (cortex-r8): Likewise.
26790 (cortex-m4): Add fpu. Add option for nofp.
26791 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26792 (cortex-a17.cortex-a7): Likewise.
26793 (cortex-a32): Add fpu. Add options for crypto and nofp.
26794 (cortex-a35, cortex-a53): Likewise.
26795 (cortex-a57): Add fpu. Add option for crypto.
26796 (cortex-a72, cortex-a73): Likewise.
26797 (exynos-m1): Likewise.
26798 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26799 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26800 (cortex-m33): Add fpu. Add option for nofp.
26801 * config/arm/arm-cpu-cdata.h: Regenerated
26802 * config/arm/arm-cpu-data.h: Regenerated.
26804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26806 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26807 (armv5te, armv5tej): Likewise.
26808 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26809 (armv7): Add options fp and vfpv3-d16.
26810 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26811 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26813 (armv7ve): Likewise.
26814 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26815 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26816 (armv8-a): Add nocrypto option.
26817 (armv8.1-a, armv8.2-a): Likewise.
26818 (armv8-m.main): add options fp, fp.dp and nofp.
26820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26822 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26824 (armv8-a+crc): Delete.
26825 (armv8.1-a): Add options simd, crypto and nofp.
26826 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26827 (armv8.2-a+fp16): Delete.
26828 (armv8-m.main): Add option dsp.
26829 (armv8-m.main+dsp): Delete.
26830 (cortex-a8): Add fpu. Add nofp option.
26831 (cortex-a9): Add fpu. Add nofp and nosimd options.
26832 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26833 link to main cpu and architecture data structures.
26834 (gen_comm_data): Only put isa attributes from the main architecture
26836 (option): New statement for architecture and CPU entries.
26837 * arm.c (struct cpu_option): New structure.
26838 (struct processors): Add entry for options.
26839 (arm_unrecognized_feature): New function.
26840 (arm_parse_arch_cpu_name): Ignore any characters after the first
26842 (arm_parse_arch_cpu_feature): New function.
26843 (arm_configure_build_target): Separate out any CPU and architecture
26844 features and parse separately. Don't error out if -mfpu=auto is
26845 used with only an architecture string.
26846 (arm_print_asm_arch_directives): New function.
26847 (arm_file_start): Call it.
26848 * config/arm/arm-cpu-cdata.h: Regenerated.
26849 * config/arm/arm-cpu-data.h: Likewise.
26850 * config/arm/arm-tables.opt: Likewise.
26852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26854 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26855 assembler when it is not -mfpu=auto.
26857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26859 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26860 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26861 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26862 (ASM_CPU_SPEC): Rewrite.
26863 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26864 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26865 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26866 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26867 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26868 copied string is NUL-terminated. Also strip any characters prefixed
26870 (arm_rewrite_selected_arch): New function.
26871 (arm_rewrite_march): New function.
26873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26875 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26876 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26877 (march, mcpu, mtune): Convert to string-based options.
26878 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26879 (arm_parse_arch_cpu_name): New function.
26880 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26881 identify selected architecture or CPU.
26882 (arm_option_save): New function.
26883 (TARGET_OPTION_SAVE): Redefine.
26884 (arm_option_restore): Restore string options.
26885 (arm_option_print): Print string options.
26887 2017-06-16 Martin Sebor <msebor@redhat.com>
26889 PR tree-optimization/80933
26890 PR tree-optimization/80934
26891 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26892 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26893 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26894 (gimple_fold_builtin): Call them.
26896 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26898 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26899 as unlikely; update profile.
26901 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26903 * predict.c (force_edge_cold): Handle declaring edges impossible
26906 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26908 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26910 (try_unroll_loop_completely): Fix reporting.
26912 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26914 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26916 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26919 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26920 if given a non-constant argument for an intrinsic which requires a
26923 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26925 * profile.c (compare_freqs): New function.
26926 (branch_prob): Sort edge list.
26927 (find_spanning_tree): Assume that the list is priority sorted.
26929 2017-06-16 Richard Biener <rguenther@suse.de>
26931 PR tree-optimization/81090
26932 * passes.def (pass_record_bounds): Remove.
26933 * tree-pass.h (make_pass_record_bounds): Likewise.
26934 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26935 make_pass_record_bounds): Likewise.
26936 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26937 not free niter estimates at the beginning but at the end.
26938 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26940 2017-06-16 Richard Biener <rguenther@suse.de>
26942 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26943 initializer to workaround ICE in host GCC 4.8.
26945 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26947 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26949 (clone_inlined_nodes): Update.
26951 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26953 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26954 prefetch settings, and enable prefetching by default at -O3.
26956 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26958 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26959 Set flag_prefetch_loop_arrays according to tuning data.
26961 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26963 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26964 New tune structure.
26965 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26966 [Unrelated to main purpose of the patch] Place the pointer field last
26967 to enable type checking errors when tune structure are wrongly merged.
26968 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26969 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26970 (thunderx2t99_prefetch_tune): New tune constants.
26971 (tune_params *_tunings): Update all tunings (no functional change).
26972 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26973 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26974 from tunings structures.
26976 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26979 * ubsan.c (instrument_null): Add T argument, use it instead
26980 of computing it based on IS_LHS.
26981 (instrument_object_size): Likewise.
26982 (pass_ubsan::execute): Adjust instrument_null and
26983 instrument_object_size callers to pass gimple_get_lhs or
26984 gimple_assign_rhs1 result to it. Use instrument_null instead of
26985 calling get_base_address and instrument_mem_ref. Handle
26986 aggregate call arguments for object-size sanitization.
26988 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26990 PR tree-optimization/81089
26991 * tree-vrp.c (is_masked_range_test): Validate operands of
26994 2017-06-15 Martin Sebor <msebor@redhat.com>
26997 * dumpfile.c (dump_register): Avoid calling memset to initialize
26998 a class with a default ctor.
26999 * gcc.c (struct compiler): Remove const qualification.
27000 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27001 * hash-table.h: Ditto.
27002 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27004 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27005 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27007 * params.h (struct param_info): Make struct members non-const.
27008 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27009 with default initialization.
27010 * vec.h (vec_copy_construct, vec_default_construct): New helper
27012 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27013 with vec_copy_construct.
27014 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27015 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27016 * doc/invoke.texi (-Wclass-memaccess): Document.
27018 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27020 * emit-rtl.h (is_leaf): Update comment about local
27021 register allocator.
27023 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27026 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27027 for a variable to have a section before checking if the section has a
27029 Set section to.persistent if persistent attribute is set.
27030 Warn if .persistent attribute is used on an automatic variable.
27032 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27034 PR rtl-optimization/80474
27035 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27037 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27039 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27042 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27044 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27045 (casesi): Emit bounds checking as RTL.
27046 (casesi_internal_mips16_<mode>): Remove bounds checking.
27048 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27050 * config/xtensa/xtensa.c (xtensa_option_override): Append
27051 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27052 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27053 xtensa_doloop_hooks): Define unconditionally.
27054 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27056 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27057 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27058 for it in xtensa_option_override.
27059 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27060 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27062 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27064 * doc/cppopts.texi: Document '-' special value to -MF.
27066 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27068 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27069 (cortex_a53_fconst): Likewise.
27070 (cortex_a53_fpmul): Likewise.
27071 (cortex_a53_f_load_64): Likewise.
27072 (cortex_a53_f_load_many): Likewise.
27073 (cortex_a53_advsimd_alu): Likewise.
27074 (cortex_a53_advsimd_alu_q): Likewise.
27075 (cortex_a53_advsimd_mul): Likewise.
27076 (cortex_a53_advsimd_mul_q): Likewise.
27077 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27078 Add missing fmul, r2f_cvt and fconst cases.
27080 2017-06-14 Richard Biener <rguenther@suse.de>
27082 PR middle-end/81088
27083 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27085 (fold_binary_loc): When associating do not treat pre-existing
27086 TREE_OVERFLOW on literal constants as a reason to allow
27087 TREE_OVERFLOW on associated literal constants.
27089 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27091 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27092 (MASK_FEATURES): New macro.
27093 * config/sparc/sparc.c (sparc_option_override): Remove the special
27094 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27096 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27098 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27099 a division of 0 if non-call exceptions are enabled.
27101 2017-06-14 Andrew Pinski <apinski@cavium.com>
27102 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27105 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27106 Improve vector initialization code gen for only variable case.
27108 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27110 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27112 2017-06-14 Richard Biener <rguenther@suse.de>
27114 PR tree-optimization/81083
27115 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27120 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27121 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27122 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27123 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27124 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27125 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27129 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27130 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27132 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27134 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27136 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27138 * config/rs6000/t-rtems: Don't handle SPE.
27140 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27142 * config/rs6000/t-linux: Don't handle SPE.
27144 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27146 * config/rs6000/eabispe.h: Delete file.
27148 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27150 * config/rs6000/t-spe: Delete file.
27152 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27154 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27155 (rs6000_legitimate_offset_address_p): Return false for anything in
27156 V2SImode or V2SFmode.
27158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27160 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27161 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27162 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27164 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27165 (rs6000_legitimate_offset_address_p): Ditto.
27166 (rs6000_emit_move): Ditto.
27167 (rs6000_init_builtins): Remove V4HI_type_node.
27169 2017-06-13 Martin Liska <mliska@suse.cz>
27172 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27173 (gate_asan): Likewise.
27174 * asan.h (asan_no_sanitize_address_p): Remove the function.
27175 (sanitize_flags_p): New function.
27176 * builtins.def: Fix coding style.
27177 * common.opt: Use renamed enum value.
27178 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27179 * doc/extend.texi: Document no_sanitize attribute.
27180 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27181 to SANITIZE_UNDEFINED_NONDEFAULT.
27182 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27183 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27184 Use sanitize_flags_p.
27185 * gimplify.c (gimplify_function_tree): Likewise.
27186 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27187 * opts.c (parse_no_sanitize_attribute): New function.
27188 (common_handle_option): Use renamed enum value.
27189 * opts.h (parse_no_sanitize_attribute): Declare.
27190 * tree.c (sanitize_flags_p): New function.
27191 * tree.h: Declared here.
27192 * tsan.c: Use sanitize_flags_p.
27193 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27194 (instrument_mem_ref): Likewise.
27195 (instrument_bool_enum_load): Likewise.
27196 (do_ubsan_in_current_function): Remove the function.
27197 (pass_ubsan::execute): Use sanitize_flags_p.
27198 * ubsan.h: Remove do_ubsan_in_current_function
27199 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27200 (dump_function_to_file): Use it here.
27202 2017-06-13 Martin Jambor <mjambor@suse.cz>
27204 PR tree-optimization/80803
27205 PR tree-optimization/81063
27206 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27207 (propagate_subaccesses_across_link): Enqueue subtree whenever
27208 necessary instead of relying on the caller.
27210 2017-06-13 Martin Jambor <mjambor@suse.cz>
27212 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27213 that have a first_link.
27214 (sort_and_splice_var_accesses): Do not check first_link before
27216 (subtree_mark_written_and_enqueue): Likewise.
27217 (propagate_all_subaccesses): Likewise and do not stop at first
27218 parent with a first_link.
27220 2017-06-13 Martin Jambor <mjambor@suse.cz>
27222 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27225 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27227 * match.pd: New pattern.
27229 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27231 * tree-vrp.c (is_masked_range_test): New function.
27232 (register_edge_assert_for): Determine ranges for
27235 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27237 PR tree-optimization/67328
27238 * fold-const.c (maskable_range_p): New function.
27239 (build_range_check): Generate bittests if possible.
27241 2017-06-13 Martin Liska <mliska@suse.cz>
27243 * gimple-pretty-print.c (dump_probability): Add new argument.
27244 (dump_edge_probability): Dump both probability and count.
27245 (dump_gimple_label): Likewise.
27246 (dump_gimple_bb_header): Likewise.
27248 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27251 * config/avr/avr-devices.c: Fix indentation.
27252 * config/avr/gen-avr-mmcu-specs.c: Dito.
27254 2017-06-13 Richard Biener <rguenther@suse.de>
27256 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27257 instead get vector type from stmt_info.
27258 (vectorizable_reduction): Adjust. Remove dead code.
27260 2017-06-13 Richard Biener <rguenther@suse.de>
27262 PR middle-end/81065
27263 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27264 case of C * (x * C2 + C3).
27265 (fold_addr_of_array_ref_difference): Properly fold index difference.
27267 2017-06-12 David S. Miller <davem@davemloft.net>
27270 * config/sparc/sparc.md (return expander): Emit frame blockage if
27271 function uses alloca.
27273 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27275 * combine.c (make_field_assignment): Check len rather than the mode
27276 precision when calling force_to_mode.
27278 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27280 Support multilibs and devices that see flash in RAM address range.
27283 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27284 (avr_mcu_t) <flash_pm_offset>: New field.
27285 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27286 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27287 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27288 (AVR_TINY_PM_OFFSET): Remove macro.
27289 * config/avr/avr.opt (-mshort-calls): New option.
27290 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27291 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27292 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27293 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27294 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27295 instead of avr_arch->have_jmp_call.
27296 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27297 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27298 avr_arch->flash_pm_offset to define.
27299 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27300 new field flash_pm_offset. Add entry for avrxmega3.
27301 (avr_texinfo): Add entry for avrxmega3.
27302 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27303 attiny212, attiny214,
27304 attiny412, attiny414, attiny416, attiny417,
27305 attiny814, attiny816, attiny817,
27306 attiny1614, attiny1616, attiny1617,
27307 attiny3214, attiny3216, attiny3217.
27308 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27309 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27310 (avr_print_operand_address) [AVR_TINY]: Same.
27311 (avr_asm_init_sections) <readonly_data_section>: Only patch
27312 callback if avr_arch->flash_pm_offset = 0.
27313 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27314 for rodata if avr_arch->flash_pm_offset != 0.
27315 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27316 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27317 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27318 (m_options): Append opt_rcall.
27319 (m_dirnames): Append dir_rcall.
27320 * config/avr/t-multilib: Regenerate.
27322 * configure.ac [target=avr]: Check whether avrxmega3 default
27323 linker description file works as needed.
27324 * configure: Regenerate.
27325 * doc/avr-mmcu.texi: Regenerate.
27326 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27327 <__AVR_ARCH__>: Document avrxmega3 and 103.
27328 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27329 <__AVR_SHORT_CALLS__>: Document it.
27330 <__AVR_PM_BASE_ADDRESS__>: Document it.
27331 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27332 (AVR Variable Attributes) <progmem>: Document this is
27333 not needed for avrxmega3.
27334 (AVR Named Address Spaces) <__flash>: Dito.
27336 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27338 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27340 2017-06-12 Doug Rupp <rupp@adacore.com>
27342 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27343 Append vxworks-stdint.h to the tm_file list.
27344 * config/vxworks-stdint.h: New file.
27346 2017-06-12 Martin Liska <mliska@suse.cz>
27348 PR tree-optimization/81041
27349 * tree-profile.c (gimple_gen_ic_func_profiler):
27350 Create an extra BB in profile-generate
27351 (gimple_gen_time_profiler): Likewise.
27353 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27355 PR tree-optimization/81003
27356 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27357 (update_range_test): Use it instead of force_gimple_operand_gsi.
27359 2017-06-12 Richard Biener <rguenther@suse.de>
27361 PR tree-optimization/81053
27362 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27363 with backedge value not defined in loop. Simplify def stmt
27366 2017-06-11 Tom de Vries <tom@codesourcery.com>
27369 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27371 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27372 nvptx_cannot_force_const_mem.
27374 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27376 * opts.c (finish_options): Move test for flag_split_stack after
27377 it has been initialized.
27379 2017-06-11 Jason Merrill <jason@redhat.com>
27381 * tree.h (id_equal): New.
27382 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27383 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27384 instead of strcmp of IDENTIFIER_POINTER.
27386 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27388 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27389 (mark_all_inlined_calls_cdtor): Fix formating.
27390 (inline_transform): Rescale profile before inlining.
27392 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27394 * cgraph.h (cgraph_edge::clone): Update prototype.
27395 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27396 (cgraph_node::create_clone): Update.
27397 (cgraph_node::create_version_clone): Update.
27398 * tree-inline.c (copy_bb): Update.
27399 (expand_call_inline): Update.
27401 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27403 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27404 factored out from ...
27405 (rs6000_emit_prologue): ... here.
27407 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27409 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27410 factored out from ...
27411 (rs6000_emit_prologue): ... here.
27413 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27415 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27417 (handle_missing_profiles): Fix computation of tp_first_run.
27418 (counts_to_freqs): Do not touch freqs when count is 0.
27420 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27422 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27425 2017-06-10 Tom de Vries <tom@codesourcery.com>
27427 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27428 attributes): Document signal effective target.
27430 2017-06-10 Tom de Vries <tom@codesourcery.com>
27432 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27433 Document effective target stack_size.
27435 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27437 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27438 to the edit_context if they can be auto-applied.
27440 2017-06-9 Ian Lance Taylor <iant@golang.org>
27442 * opts.c (finish_options): If -fsplit-stack, disable implicit
27443 -forder-blocks-and-partition.
27444 * doc/invoke.texi (Optimize Options): Document that when using
27445 -fsplit-stack -forder-blocks-and-partition is not implicitly
27448 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27450 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27451 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27452 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27453 * builtins.def (abort, trap, unreachable): Declare cold.
27454 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27455 * tree-core.h (ECF_COLD): New.
27456 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27457 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27459 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27461 * predict.c (unlikely_executed_stmt_p): Cleanup.
27463 2017-06-09 Richard Biener <rguenther@suse.de>
27465 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27466 model if the ref is always written to.
27468 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27470 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27472 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27474 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27477 2017-06-09 Tom de Vries <tom@codesourcery.com>
27480 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27481 "target cannot support label values" when encountering LABEL_REF.
27483 2017-06-09 Martin Liska <mliska@suse.cz>
27485 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27486 (gimple_gen_ic_func_profiler): Emit direct comparison
27487 of __gcov_indirect_call_callee with NULL.
27488 (gimple_gen_time_profiler): Change probability from
27489 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27491 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27493 * profile.c (edge_gcov_counts): Turn to pointer.
27494 (compute_branch_probabilities, compute_branch_probabilities): Update.
27495 (branch_prob): Do not clear edge_gcov_count.
27496 * profile.h (edge_gcov_counts): Turn to pointer.
27497 (edge_gcov_count): Update.
27499 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27501 * gimple.h (gimple_check_failed): Mark cold.
27503 2017-06-09 Richard Biener <rguenther@suse.de>
27505 PR tree-optimization/66623
27506 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27507 refactor check_reduction into two parts, properly computing
27508 whether we have to check reduction validity for outer loop
27511 2017-06-09 Richard Biener <rguenther@suse.de>
27513 PR tree-optimization/79483
27514 * graphite-scop-detection.c (order): New global.
27515 (get_order): Compute bb to order mapping that satisfies code
27516 generation constraints.
27517 (cmp_pbbs): New helper.
27518 (build_scops): Start domwalk at entry block, sort generated
27521 2017-06-09 Richard Biener <rguenther@suse.de>
27523 PR middle-end/81007
27524 * ipa-polymorphic-call.c
27525 (ipa_polymorphic_call_context::restrict_to_inner_class):
27526 Skip FIELD_DECLs with error_mark_node type.
27527 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27530 2017-06-09 Martin Liska <mliska@suse.cz>
27532 * predict.c (struct branch_predictor): New struct.
27533 (test_prediction_value_range): New test.
27534 (predict_c_tests): New function.
27535 * selftest-run-tests.c (selftest::run_tests): Run the function.
27536 * selftest.h: Declare new tests.
27538 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27541 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27542 gen_add3_insn did not fail.
27543 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27544 r0, construct that number in a temporary reg and add that reg to r0.
27545 If asked to put the result in r0 as well, fail.
27547 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27549 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27550 for early expansion of vec_eqv.
27552 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27554 PR middle-end/81005
27555 * ubsan.c (instrument_null): Avoid pointless code temporary.
27556 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27558 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27562 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27565 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27568 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27570 * predict.c (unlikely_executed_edge_p): Move ahead.
27571 (probably_never_executed_edge_p): Use it.
27573 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27575 PR middle-end/79988
27576 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27577 gimple_call_builtin_p call.
27579 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27581 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27582 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27583 rtl_check_failed_type2, rtl_check_failed_code1,
27584 rtl_check_failed_code2, rtl_check_failed_code_mode,
27585 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27586 rtvec_check_failed_bounds, rtl_check_failed_flag,
27587 _fatal_insn_not_found, _fatal_insn): Likewise.
27588 * tree.h (tree_contains_struct_check_failed,
27589 tree_check_failed, tree_not_check_failed,
27590 tree_class_check_failed, tree_range_check_failed,
27591 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27592 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27593 tree_operand_check_failed, omp_clause_check_failed,
27594 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27597 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27599 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27600 flag_branch_probabilities.
27601 * ipa-inline.c (edge_badness): Likewise.
27602 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27603 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27604 * predict.c (maybe_hot_frequency_p): Likewise.
27605 (probably_never_executed): Likewise.
27606 * sched-ebb.c (schedule_ebbs): Likewise.
27607 * sched-rgn.c (find_single_block_region): Likewise.
27608 * tracer.c (tail_duplicate): Likewise.
27610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27612 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27613 longer requires x_flag_profile_use.
27615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27617 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27618 instead of flag_reorder_blocks_and_partition.
27619 * dbxout.c (dbxout_function_end): Likewise.
27620 * dwarf2out.c (gen_subprogram_die): Likewise.
27621 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27622 * hw-doloop.c (reorg_loops): Likewise.
27623 * varasm.c (assemble_start_function,
27624 assemble_end_function): Likewise.
27625 (decide_function_section): Do not check for
27626 flag_reorder_blocks_and_partition.
27628 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27630 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27632 (chkp_get_hard_register_fake_addr_expr): Ditto.
27633 (chkp_build_addr_expr): Add check for hard reg case.
27634 (chkp_parse_array_and_component_ref): Ditto.
27635 (chkp_find_bounds_1): Ditto.
27636 (chkp_process_stmt): Don't generate bounds store for
27639 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27641 * predict.c (maybe_hot_bb_p): Do not check profile status.
27642 (maybe_hot_edge_p): Likewise.
27643 (probably_never_executed): Check for zero counts even if profile
27645 (unlikely_executed_edge_p): New function.
27646 (unlikely_executed_stmt_p): New function.
27647 (unlikely_executed_bb_p): New function.
27648 (set_even_probabilities): Use unlikely predicates.
27649 (combine_predictions_for_bb): Likewise.
27650 (predict_paths_for_bb): Likewise.
27651 (predict_paths_leading_to_edge): Likewise.
27652 (determine_unlikely_bbs): New function.
27653 (estimate_bb_frequencies): Use it.
27654 (compute_function_frequency): Use zero counts even if profile is
27656 * profile-count.h: Fix typo.
27658 2017-08-08 Julia Koval <julia.koval@intel.com>
27660 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27661 _mm512_mask_cvtsepi16_storeu_epi8,
27662 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27663 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27664 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27665 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27666 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27667 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27668 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27669 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27670 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27671 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27672 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27673 __builtin_ia32_pmovuswb256mem_mask,
27674 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27675 __builtin_ia32_pmovwb512mem_mask): New builtins.
27677 2017-08-08 Julia Koval <julia.koval@intel.com>
27679 PR target/73350,80862
27680 * config/i386/subst.md (round): Fix round pattern.
27681 * config/i386/i386.c (ix86_erase_embedded_rounding):
27682 Fix erasing rounding for the fixed pattern.
27684 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27686 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27688 2017-06-08 Martin Liska <mliska@suse.cz>
27690 PR gcov-profile/80911
27691 * gcov.c (block_info::block_info): New constructor.
27693 2017-06-07 Carl Love <cel@us.ibm.com>
27695 * config/rs6000/rs6000-c: The return type of the following
27696 built-in functions was implemented as int not long long. Fix sign
27697 of return value for the unsigned version of vec_mulo and vec_mule.
27698 vector unsigned long long vec_bperm (vector unsigned long long,
27699 vector unsigned char)
27700 vector signed long long vec_mule (vector signed int,
27702 vector unsigned long long vec_mule (vector unsigned int,
27703 vector unsigned int)
27704 vector signed long long vec_mulo (vector signed int,
27706 vector unsigned long long vec_mulo (vector unsigned int,
27707 vector unsigned int)
27708 * doc/extend.texi: Fix the documentation for the built-in
27711 2017-06-07 Carl Love <cel@us.ibm.com>
27714 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27717 2017-06-07 Carl Love <cel@us.ibm.com>
27719 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27720 support, Generate doublehv for signed int/float for BE case only.
27722 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27724 * doc/invoke.texi (mcx16): Rewrite.
27726 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27728 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27729 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27730 *mov<mode>_softfloat, and an anonymous splitter): Use
27731 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27735 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27737 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27738 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27739 (rs6000_debug_reg_global): Adjust.
27740 (rs6000_init_hard_regno_mode_ok): Adjust.
27741 (rs6000_dbx_register_number): Adjust.
27742 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27743 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27744 Remove SPE_ACC and SPEFSCR.
27745 (REG_ALLOC_ORDER): Ditto.
27746 (FRAME_POINTER_REGNUM): Change to 111.
27747 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27748 (REG_CLASS_NAMES): Ditto.
27749 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27750 (REGISTER_NAMES): Ditto.
27751 (ADDITIONAL_REG_NAMES): Ditto.
27752 (rs6000_reg_names): Ditto.
27753 * config/rs6000/rs6000.md: Renumber some register number
27756 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27758 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27760 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27761 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27763 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27764 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27765 Delete the SPE high registers.
27766 (REG_ALLOC_ORDER): Ditto.
27767 (enum reg_class): Remove SPE_HIGH_REGS.
27768 (REG_CLASS_NAMES): Ditto.
27769 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27770 (REGISTER_NAMES): Ditto.
27771 (rs6000_reg_names): Ditto.
27772 * doc/tm.texi.in: Remove SPE as example.
27773 * doc/tm.texi: Regenerate.
27775 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27777 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27778 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27779 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27780 * config/rs6000/rs6000.md (type): Remove "brinc".
27782 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27784 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27785 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27786 * config/rs6000/linuxspe.h: Delete file.
27787 * config/rs6000/rs6000.md: Don't include spe.md.
27788 * config/rs6000/spe.h: Delete file.
27789 * config/rs6000/spe.md: Delete file.
27790 * config/rs6000/t-rs6000: Remove spe.md.
27792 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27794 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27795 (reg_or_none500mem_operand): Delete.
27796 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27797 instead of reg_or_none500mem_operand.
27799 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27802 handling of SPE flags.
27803 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27807 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27809 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27810 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27811 paired_divv2sf3): Similar.
27812 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27813 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27814 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27816 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27817 Rename the paired_* instruction patterns.
27818 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27820 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27821 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27822 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27823 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27824 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27825 PAIRED_VECTOR_MODE.
27826 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27827 (spe_func_has_64bit_regs_p): Delete.
27828 (spe_expand_predicate_builtin): Delete.
27829 (spe_expand_evsel_builtin): Delete.
27830 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27831 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27832 (invalid_e500_subreg): Delete.
27833 (rs6000_legitimize_address): Always force_reg op2 as well, for
27834 paired single memory accesses.
27835 (rs6000_member_type_forces_blk): Delete.
27836 (rs6000_spe_function_arg): Delete.
27837 (rs6000_expand_unop_builtin): Delete SPE handling.
27838 (rs6000_expand_binop_builtin): Ditto.
27839 (spe_expand_stv_builtin): Delete.
27840 (bdesc_2arg_spe): Delete.
27841 (spe_expand_builtin): Delete.
27842 (spe_expand_predicate_builtin): Delete.
27843 (spe_expand_evsel_builtin): Delete.
27844 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27845 (spe_init_builtins): Delete.
27846 (spe_func_has_64bit_regs_p): Delete.
27847 (savres_routine_name): Delete "info" parameter. Adjust callers.
27848 (rs6000_emit_stack_reset): Ditto.
27849 (rs6000_dwarf_register_span): Delete.
27850 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27851 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27852 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27854 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27856 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27857 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27858 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27859 mulv2sf3, divv2sf3): Delete expanders.
27861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27863 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27865 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27867 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27868 * config/rs6000/rs6000.c: Ditto.
27870 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27872 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27873 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27874 comparison_operator.
27876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27878 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27879 * config/rs6000/rs6000.opt: Ditto.
27880 * config/rs6000/t-rtems: Ditto.
27882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27884 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27885 TARGET_E500_SINGLE by 0, simplify.
27886 * config/rs6000/rs6000.c: Ditto.
27887 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27888 (spe_build_register_parallel): Delete.
27889 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27890 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27891 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27892 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27893 (E500_CONVERT): Delete.
27894 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27896 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27898 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27899 * config/rs6000/dfp.md: Ditto.
27900 (negdd2, *negdd2_fpr): Merge.
27901 (absdd2, *absdd2_fpr): Merge.
27902 (negtd2, *negtd2_fpr): Merge.
27903 (abstd2, *abstd2_fpr): Merge.
27904 * config/rs6000/e500.h: Delete file.
27905 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27906 TARGET_FPRS by 1 and simplify.
27907 * config/rs6000/rs6000-c.c: Ditto.
27908 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27909 TARGET_DF_SPE by 0.
27910 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27912 * config/rs6000/rs6000.md: Ditto.
27913 (floatdidf2, *floatdidf2_fpr): Merge.
27914 (move_from_CR_gt_bit): Delete.
27915 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27916 (E500_CR_IOR_COMPARE): Delete.
27917 (All patterns that require !TARGET_FPRS): Delete.
27918 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27920 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27922 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27924 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27926 * graphds.c (add_edge): Intitialize edge's attached data.
27927 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27928 pointer parameter. Call pointed function on each edge during
27929 graph traversing. Skip traversing the edge when the function
27931 (graphds_dfs, graphds_scc): Ditto.
27932 (for_each_edge): New parameter. Pass the new parameter to callback
27934 * graphds.h (skip_edge_callback): New function pointer type.
27935 (graphds_dfs, graphds_scc): New function pointer parameter.
27936 (graphds_edge_callback, for_each_edge): New parameter.
27938 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27940 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27941 out code checking if runtime alias check is possible to below ...
27942 Call the new function.
27943 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27944 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27946 2017-06-07 Marek Polacek <polacek@redhat.com>
27949 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27950 TYPE_OVERFLOW_WRAPS check.
27952 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27954 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27955 if versioning is required.
27956 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27957 peeling with the check for versioning.
27959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27961 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27962 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27963 Set true to new parameter if new ssa variable is defined.
27964 (vect_gen_vector_loop_niters): Refactor. Set range information
27965 for the new vector loop bound variable.
27966 (vect_do_peeling): Ditto.
27968 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27970 * tree-affine.c (ssa.h): Include header file.
27971 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27972 has wrapping overflow behavior.
27974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27976 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27978 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27980 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27981 (tree_to_aff_combination): ... here.
27983 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27985 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27986 reg_pressure model function.
27987 (ivopts_global_cost_for_size): Delete.
27988 (determine_set_costs, iv_ca_recount_cost): Call new model function
27989 ivopts_estimate_reg_pressure.
27991 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27993 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27994 expensive than udiv. Remove floating point cases from mod.
27996 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27998 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27999 Increase idiv cost.
28001 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28003 * config/aarch64/aarch64.md
28004 (copysignsf3): Fix mask generation.
28006 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28008 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28010 (class dump_manager): Add register_dumps method.
28011 * dumpfile.c: Include langhooks.h.
28012 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28013 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28014 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28015 (dump_manager::dump_register): Start with 512 entries instead of 32.
28016 (dump_manager::register_dumps): New method.
28017 * toplev.c (general_init): Instead of invoking register_dumps
28018 langhook, invoke register_dumps method on the dump manager.
28019 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28022 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28024 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28025 Say that # requires an associated define_split to exist, and that
28026 the define_split must be suitable for use after register allocation.
28028 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28030 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28031 (compute_outgoing_frequencies): Also initialize zero counts.
28032 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28033 around loops; preserve more of profile when nothing changes.
28035 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28037 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28039 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28040 * config/arm/arm-cpu-cdata.h: Regenerate.
28041 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28042 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28043 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28044 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28045 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28047 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28049 * config/arm/t-rmprofile: Likewise.
28050 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28052 2017-06-06 David S. Miller <davem@davemloft.net>
28055 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28056 blockage if function uses alloca.
28058 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28060 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28061 New "uid" fields to hold pretty-print IDs of group and ref.
28062 Memory references are now identified as <group_id>:<ref_id>
28063 instead of using [random] addresses.
28064 (dump_mem_details): Simplify, no functional change.
28065 (dump_mem_ref): Simplify and make output more concise.
28066 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28067 (find_or_create_group): Initialize group uid.
28068 (record_ref): Initialize ref uid. Improve debug output.
28069 (prune_group_by_reuse, should_issue_prefetch_p,)
28070 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28071 (mark_nontemporal_store, determine_loop_nest_reuse):
28072 Improve debug output.
28074 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28076 * dbgcnt.def (prefetch): New debug counter.
28077 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28078 (schedule_prefetches): Stop issueing prefetches if debug counter
28081 2017-06-06 Tom de Vries <tom@codesourcery.com>
28083 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28084 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28086 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28088 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28089 Use aarch64_reg_or_zero predicate for operand 4.
28090 (aarch64_compare_and_swap<mode> define_insn_and_split):
28091 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28092 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28094 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28096 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28097 (arm_compute_save_core_reg_mask): This.
28098 (thumb1_compute_save_reg_mask): Rename into ...
28099 (thumb1_compute_save_core_reg_mask): This.
28100 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28101 (arm_compute_frame_layout): Likewise.
28103 2017-06-06 Richard Biener <rguenther@suse.de>
28105 PR tree-optimization/80974
28106 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28107 keep or clear leaders SSA info.
28109 2017-06-06 Tom de Vries <tom@codesourcery.com>
28111 * config/nvptx/nvptx.c (split_mode_p): New function.
28112 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28114 2017-06-06 Tom de Vries <tom@codesourcery.com>
28116 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28118 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28121 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28124 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28126 * shrink-wrap.c (handle_simple_exit): Update profile.
28127 (try_shrink_wrapping): Upate profile.
28129 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28131 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28132 (tree_guess_outgoing_edge_probabilities): New.
28133 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28134 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28136 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28138 * ipa-split.c (split_function): Initialize return bb profile.
28140 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28142 * profile.c (compute_branch_probabilities): Also initialize
28143 EXIT_BLOCK profile.
28145 2017-06-06 Richard Biener <rguenther@suse.de>
28147 PR tree-optimization/80928
28148 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28149 (vect_analyze_loop_operations): Properly guard analysis for
28151 (vect_transform_loop): Likewise.
28152 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28153 (vect_model_induction_cost): Do not cost for pure SLP.
28154 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28155 of induction in inner loop vectorization.
28156 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28157 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28158 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28160 (vect_analyze_slp_cost_1): Cost induction.
28161 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28162 (vect_get_slp_vect_defs): Likewise.
28163 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28164 (vect_transform_stmt): Handle SLP reductions.
28165 * tree-vectorizer.h (vectorizable_induction): Adjust.
28167 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28169 * config/rs6000/rs6000.c (make_resolver_func): Update
28170 init_lowered_empty_function call.
28172 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28174 * doc/invoke.texi: Document the -fprofile-abs-path option.
28175 * common.opt (fprofile-abs-path): New option.
28176 * gcov-io.h (gcov_write_filename): Declare.
28177 * gcov-io.c (gcov_write_filename): New function.
28178 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28179 * profile.c (output_location): Likewise.
28181 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28183 * shring-wrap.c: Revert accidental commit.
28185 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28187 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28189 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28191 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28193 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28194 profile in callgraph edge.
28195 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28196 (apply_scale): Likewise.
28197 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28198 Also scale profile when inlining function with zero profile.
28199 (initialize_cfun): Update exit block profile even when it is zero.
28200 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28201 when profile is read.
28203 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28205 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28206 (CLONE_*): New constants to define the processors we can generate
28207 code for with the target_clone attribute.
28208 (rs6000_clone_map): New array to identify which clone processors
28209 the current program is running on.
28210 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28211 target_clone attribute.
28212 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28213 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28214 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28215 (cpu_expand_builtin): Add support for target_clone attribute.
28216 (rs6000_valid_attribute_p): Allow "default" attribute.
28217 (get_decl_name): New debug function to simplify printing the
28218 current function name in debugging statements.
28219 (rs6000_clone_priority): New functions to support the target_clone
28220 attribute, and be able to generate code to switch between ISA 2.05
28221 through ISA 3.0 (power6 through power9).
28222 (rs6000_compare_version_priority): Likewise.
28223 (rs6000_get_function_versions_dispatcher): Likewise.
28224 (make_resolver_func): Likewise.
28225 (add_condition_to_bb): Likewise.
28226 (dispatch_function_versions): Likewise.
28227 (rs6000_generate_version_dispatcher_body): Likewise.
28228 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28229 (fusion_gpr_load_p): Fix a spacing issue.
28230 * doc/extend.texi (Common Function Attributes): Document that the
28231 PowerPC supports the target_clone attribute.
28233 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28235 * config/arm/arm.h: explain F symbol found in description of ARM
28236 register allocation in its legend.
28238 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28240 * config/mips/frame-header-opt.c: Include profile-count.h.
28241 * config/riscv/riscv.c: Include profile-count.h
28243 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28245 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28247 (sm_set_flag_if_changed): Add bbs field.
28248 (execute_sm_if_changed_flag_set): Pass BBS.
28249 (execute_sm): Update.
28251 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28253 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28256 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28258 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28259 (peephole2): New peephole2 to emit the above.
28260 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28262 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28264 * config/aarch64/aarch64.c (define_peephole2 above
28265 *sub_<shift>_<mode>): New peephole.
28267 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28269 * config/i386/i386.c (make_resolver_func): Update.
28270 * Makefile.in: Add profile-count.h and profile-count.o
28271 * auto-profile.c (afdo_indirect_call): Update to new API.
28272 (afdo_set_bb_count): Update.
28273 (afdo_propagate_edge): Update.
28274 (afdo_propagate_circuit): Update.
28275 (afdo_calculate_branch_prob): Update.
28276 (afdo_annotate_cfg): Update.
28277 * basic-block.h: Include profile-count.h
28278 (struct edge_def): Turn count to profile_count.
28279 (struct basic_block_def): Likewie.
28280 (REG_BR_PROB_BASE): Move to profile-count.h
28281 (RDIV): Move to profile-count.h
28282 * bb-reorder.c (max_entry_count): Turn to profile_count.
28283 (find_traces): Update.
28284 (rotate_loop):Update.
28285 (connect_traces):Update.
28286 (sanitize_hot_paths):Update.
28287 * bt-load.c (migrate_btr_defs): Update.
28288 * cfg.c (RDIV): Remove.
28289 (init_flow): Use alloc_block.
28290 (alloc_block): Uninitialize count.
28291 (unchecked_make_edge): Uninitialize count.
28292 (check_bb_profile): Update.
28293 (dump_edge_info): Update.
28294 (dump_bb_info): Update.
28295 (update_bb_profile_for_threading): Update.
28296 (scale_bbs_frequencies_int): Update.
28297 (scale_bbs_frequencies_gcov_type): Update.
28298 (scale_bbs_frequencies_profile_count): New.
28299 * cfg.h (update_bb_profile_for_threading): Update.
28300 (scale_bbs_frequencies_profile_count): Declare.
28301 * cfgbuild.c (compute_outgoing_frequencies): Update.
28302 (find_many_sub_basic_blocks): Update.
28303 * cfgcleanup.c (try_forward_edges): Update.
28304 (try_crossjump_to_edge): Update.
28305 * cfgexpand.c (expand_gimple_tailcall): Update.
28306 (construct_exit_block): Update.
28307 * cfghooks.c (verify_flow_info): Update.
28308 (dump_bb_for_graph): Update.
28309 (split_edge): Update.
28310 (make_forwarder_block): Update.
28311 (duplicate_block): Update.
28312 (account_profile_record): Update.
28313 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28314 (get_estimated_loop_iterations): Update.
28315 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28316 (single_likely_exit): Update.
28317 * cfgloopmanip.c (scale_loop_profile): Update.
28319 (set_zero_probability): Update.
28320 (lv_adjust_loop_entry_edge): Update.
28321 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28322 (purge_dead_edges): Update.
28323 (rtl_account_profile_record): Update.
28324 * cgraph.c (cgraph_node::create): Uninitialize count.
28325 (symbol_table::create_edge): Uninitialize count.
28326 (cgraph_update_edges_for_call_stmt_node): Update.
28327 (cgraph_edge::dump_edge_flags): Update.
28328 (cgraph_node::dump): Update.
28329 (cgraph_edge::maybe_hot_p): Update.
28330 * cgraph.h: Include profile-count.h
28331 (create_clone), create_edge, create_indirect_edge): Update.
28332 (cgraph_node): Turn count to profile_count.
28333 (cgraph_edge0: Likewise.
28334 (make_speculative, clone): Update.
28335 (create_edge): Update.
28336 (init_lowered_empty_function): Update.
28337 * cgraphclones.c (cgraph_edge::clone): Update.
28338 (duplicate_thunk_for_node): Update.
28339 (cgraph_node::create_clone): Update.
28340 * cgraphunit.c (cgraph_node::analyze): Update.
28341 (cgraph_node::expand_thunk): Update.
28342 * final.c (dump_basic_block_info): Update.
28343 * gimple-streamer-in.c (input_bb): Update.
28344 * gimple-streamer-out.c (output_bb): Update.
28345 * graphite.c (print_global_statistics): Update.
28346 (print_graphite_scop_statistics): Update.
28347 * hsa-brig.c: Include basic-block.h.
28348 * hsa-dump.c: Include basic-block.h.
28349 * hsa-gen.c (T sum_slice): Update.
28350 (convert_switch_statements):Update.
28351 * hsa-regalloc.c: Include basic-block.h.
28352 * ipa-chkp.c (chkp_produce_thunks): Update.
28353 * ipa-cp.c (struct caller_statistics): Update.
28354 (init_caller_stats): Update.
28355 (gather_caller_stats): Update.
28356 (ipcp_cloning_candidate_p): Update.
28357 (good_cloning_opportunity_p): Update.
28358 (get_info_about_necessary_edges): Update.
28359 (dump_profile_updates): Update.
28360 (update_profiling_info): Update.
28361 (update_specialized_profile): Update.
28362 (perhaps_add_new_callers): Update.
28363 (decide_about_value): Update.
28364 (ipa_cp_c_finalize): Update.
28365 * ipa-devirt.c (struct odr_type_warn_count): Update.
28366 (struct decl_warn_count): Update.
28367 (struct final_warning_record): Update.
28368 (possible_polymorphic_call_targets): Update.
28369 (ipa_devirt): Update.
28370 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28371 * ipa-icf.c (sem_function::merge): Update.
28372 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28373 * ipa-inline.c (compute_uninlined_call_time): Update.
28374 (compute_inlined_call_time): Update.
28375 (want_inline_small_function_p): Update.
28376 (want_inline_self_recursive_call_p): Update.
28377 (edge_badness): Update.
28378 (lookup_recursive_calls): Update.
28379 (recursive_inlining): Update.
28380 (inline_small_functions): Update.
28381 (dump_overall_stats): Update.
28382 (dump_inline_stats): Update.
28383 * ipa-profile.c (ipa_profile_generate_summary): Update.
28384 (ipa_propagate_frequency): Update.
28385 (ipa_profile): Update.
28386 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28387 * ipa-utils.c (ipa_merge_profiles): Update.
28388 * loop-doloop.c (doloop_modify): Update.
28389 * loop-unroll.c (report_unroll): Update.
28390 (unroll_loop_runtime_iterations): Update.
28391 * lto-cgraph.c (lto_output_edge): Update.
28392 (lto_output_node): Update.
28393 (input_node): Update.
28394 (input_edge): Update.
28395 (merge_profile_summaries): Update.
28396 * lto-streamer-in.c (input_cfg): Update.
28397 * lto-streamer-out.c (output_cfg): Update.
28398 * mcf.c (create_fixup_graph): Update.
28399 (adjust_cfg_counts): Update.
28400 (sum_edge_counts): Update.
28401 * modulo-sched.c (sms_schedule): Update.
28402 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28403 * predict.c (maybe_hot_count_p): Update.
28404 (probably_never_executed): Update.
28405 (dump_prediction): Update.
28406 (combine_predictions_for_bb): Update.
28407 (propagate_freq): Update.
28408 (handle_missing_profiles): Update.
28409 (counts_to_freqs): Update.
28410 (rebuild_frequencies): Update.
28411 (force_edge_cold): Update.
28412 * predict.h: Include profile-count.h
28413 (maybe_hot_count_p, counts_to_freqs): UPdate.
28414 * print-rtl-function.c: Do not include cfg.h
28415 * print-rtl.c: Include basic-block.h
28416 * profile-count.c: New file.
28417 * profile-count.h: New file.
28418 * profile.c (is_edge_inconsistent): Update.
28419 (correct_negative_edge_counts): Update.
28420 (is_inconsistent): Update.
28421 (set_bb_counts): Update.
28422 (read_profile_edge_counts): Update.
28423 (compute_frequency_overlap): Update.
28424 (compute_branch_probabilities): Update; Initialize and deinitialize
28426 (branch_prob): Update.
28427 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28428 (edge_gcov_count): New.
28429 (bb_gcov_count): New.
28430 * shrink-wrap.c (try_shrink_wrapping): Update.
28431 * tracer.c (better_p): Update.
28432 * trans-mem.c (expand_transaction): Update.
28433 (ipa_tm_insert_irr_call): Update.
28434 (ipa_tm_insert_gettmclone_call): Update.
28435 * tree-call-cdce.c: Update.
28436 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28437 (gimple_duplicate_sese_tail): Update.
28438 (gimple_account_profile_record): Update.
28439 (execute_fixup_cfg): Update.
28440 * tree-inline.c (copy_bb): Update.
28441 (copy_edges_for_bb): Update.
28442 (initialize_cfun): Update.
28443 (freqs_to_counts): Update.
28444 (copy_cfg_body): Update.
28445 (expand_call_inline): Update.
28446 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28447 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28448 (try_unroll_loop_completely): Update.
28449 (try_peel_loop): Update.
28450 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28451 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28452 * tree-ssa-loop-split.c (connect_loops): Update.
28453 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28454 * tree-ssa-reassoc.c (branch_fixup): Update.
28455 * tree-ssa-tail-merge.c (replace_block_by): Update.
28456 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28457 (compute_path_counts): Update.
28458 (update_profile): Update.
28459 (recompute_probabilities): Update.
28460 (update_joiner_offpath_counts): Update.
28461 (estimated_freqs_path): Update.
28462 (freqs_to_counts_path): Update.
28463 (clear_counts_path): Update.
28464 (ssa_fix_duplicate_block_edges): Update.
28465 (duplicate_thread_path): Update.
28466 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28467 (struct switch_conv_info): Update.
28468 * tree-tailcall.c (decrease_profile): Update.
28469 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28470 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28471 * value-prof.c (check_counter): Update.
28472 (gimple_divmod_fixed_value): Update.
28473 (gimple_mod_pow2): Update.
28474 (gimple_mod_subtract): Update.
28475 (gimple_ic_transform): Update.
28476 (gimple_stringop_fixed_value): Update.
28477 * value-prof.h (gimple_ic): Update.
28479 2017-06-02 Carl Love <cel@us.ibm.com>
28481 * config/rs6000/rs6000-c: Add support for built-in functions
28482 vector double vec_doublee (vector signed int);
28483 vector double vec_doublee (vector unsigned int);
28484 vector double vec_doublee (vector float);
28485 vector double vec_doubleh (vector signed int);
28486 vector double vec_doubleh (vector unsigned int);
28487 vector double vec_doubleh (vector float);
28488 vector double vec_doublel (vector signed int);
28489 vector double vec_doublel (vector unsigned int);
28490 vector double vec_doublel (vector float);
28491 vector double vec_doubleo (vector signed int);
28492 vector double vec_doubleo (vector unsigned int);
28493 vector double vec_doubleo (vector float);.
28494 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28495 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28497 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28498 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28499 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28501 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28502 vec_doublel, vec_doubleh.
28503 * doc/extend.texi: Update the built-in documentation file for the
28504 new built-in functions.
28506 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28509 * ipa-inline-analysis.c (free_growth_caches): Set
28510 edge_removal_hook_holder to NULL after removing it.
28512 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28514 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28515 comparision with zero.
28517 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28518 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28519 for early expansion of vec_min and vec_max builtins.
28520 (builtin_function_type): Add min/max unsigned variants to those
28521 identified as having unsigned arguments.
28523 2017-06-02 Olivier Hainque <hainque@adacore.com>
28525 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28527 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28529 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28530 Use VALL_F16 iterator rather than VALL.
28532 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28534 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28535 Emit CBNZ inside loop when doing a strong exchange and comparing
28536 against zero. Generate the CC flags after the loop.
28538 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28540 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28541 (dl_section_ref): New.
28542 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28543 On AIX, append an expression to subtract the size of the
28544 section length to dl_section_ref.
28546 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28549 for early expansion of vector absolute builtins.
28551 2017-06-02 Richard Biener <rguenther@suse.de>
28553 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28554 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28556 2017-06-02 Richard Biener <rguenther@suse.de>
28558 PR tree-optimization/80948
28559 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28560 stmt order as well.
28562 2017-06-02 Richard Biener <rguenther@suse.de>
28564 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28566 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28567 for inductions as relevant.
28569 2017-06-02 Richard Biener <rguenther@suse.de>
28571 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28572 (vectorizable_induction): ... this. Remove dead code.
28574 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28576 * builtins. (expand_builtin_alloca): Remove second parameter and
28577 infer its value from the first parameter instead.
28578 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28580 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28582 PR rtl-optimization/80903
28583 * loop-doloop.c (add_test): Unshare sequence.
28585 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28587 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28589 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28591 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28593 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28594 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28595 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28596 (xlogue_layout::get_stub_rtx): Make static.
28597 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28598 (xlogue_layout::compute_stub_managed_regs): Rename to...
28599 (xlogue_layout::count_stub_managed_regs): ...this.
28600 (xlogue_layout::is_stub_managed_reg): New function.
28601 (xlogue_layout::m_stub_names): Rename to...
28602 (xlogue_layout::s_stub_names): ...this, make static.
28603 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28604 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28605 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28606 xlogue_layout::s_stub_names): Instantiate statics.
28607 (stub_managed_regs): Remove.
28608 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28609 (disable_call_ms2sysv_xlogues): Rename to...
28610 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28611 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28613 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28614 change after reload_completed.
28615 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28617 (ix86_expand_prologue): Likewise.
28618 (ix86_expand_epilogue): Likewise.
28619 (ix86_expand_split_stack_prologue): Likewise.
28620 (ix86_compute_frame_layout): Remove frame parameter ...
28621 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28622 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28624 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28625 (ix86_frame): Move from here ...
28626 * config/i386/i386.h (ix86_frame): ... to here.
28627 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28628 complete ix86_frame data structure instead. Remove some_ld_name.
28630 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28632 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28633 symbols that hold a DECL_VALUE_EXPR.
28635 2017-06-01 Martin Jambor <mjambor@suse.cz>
28637 PR tree-optimization/80898
28638 * tree-sra.c (process_subtree_disqualification): Removed.
28639 (disqualify_candidate): Do not acll
28640 process_subtree_disqualification.
28641 (subtree_mark_written_and_enqueue): New function.
28642 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28643 RHS has been disqualified and re-queue LHS if necessary. Apart
28644 from that, ignore disqualified RHS.
28646 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28648 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28649 address fetch for z10 or later.
28651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28653 * config/arc/arc.md (tst_movb): Add guard when splitting.
28655 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28657 * config/arc/arc.c (arc_can_eliminate): Test against
28658 arc_frame_pointer_needed.
28660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28662 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28663 to prevent store reordering.
28664 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28665 (type): Add block type.
28666 (stack_tie): Define special instruction to be used in
28669 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28671 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28672 constraint. It is not valid for the pattern.
28673 (noncommutative_binary_comparison): Likewise.
28675 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28677 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28680 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28682 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28683 be used by the reg-alloc.
28685 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28687 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28688 reg-alloc when having mul64 or mul32x16 instructions.
28689 (mulsidi3): Likewise.
28690 (umulsidi3): Likewise.
28691 (mulsi32x16): New pattern.
28692 (mulsi64): Likewise.
28693 (mulsidi64): Likewise.
28694 (umulsidi64): Likewise.
28695 (MUL32x16_REG): Define.
28696 (mul64_600): Use MUL32x16_REG.
28697 (mac64_600): Likewise.
28698 (umul64_600): Likewise.
28699 (umac64_600): Likewise.
28701 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28703 * config/arc/arc.md (mulsi3_700): Make it commutative.
28705 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28707 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28709 (*sign_extendsidi2_insn): Likewise for movstosw.
28711 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28713 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28714 the type of the input discriminant value. Convert the
28715 discriminant value of signedness vary.
28717 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28719 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28720 Add to -Wall section.
28722 2017-06-01 Richard Biener <rguenther@suse.de>
28724 PR middle-end/66313
28725 * fold-const.c (fold_plusminus_mult_expr): If the factored
28726 factor may be zero use a wrapping type for the inner operation.
28727 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28728 and handle moved defs.
28729 (process_assignment): Properly guard the unary op case. Return a
28730 tri-state indicating that moving the stmt before the call may allow
28731 to continue. Pass through to_move.
28732 (find_tail_calls): Handle moving unrelated defs before
28735 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28738 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28739 splitter result in the canonical way.
28741 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28743 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28744 also for 32bit target. Update insn attributes.
28745 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28747 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28749 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28750 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28751 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28752 (_mm_maskz_min_ss): New intrinsics.
28754 2017-05-31 Martin Liska <mliska@suse.cz>
28756 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28757 Change comment style to one we normally use.
28758 (vectorizable_reduction): Likewise.
28759 (vectorizable_induction): Likewise.
28760 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28761 (vectorizable_call): Likewise.
28762 (vectorizable_simd_clone_call): Likewise.
28763 (vectorizable_conversion): Likewise.
28764 (vectorizable_assignment): Likewise.
28765 (vectorizable_shift): Likewise.
28766 (vectorizable_operation): Likewise.
28767 (vectorizable_store): Likewise.
28768 (vectorizable_load): Likewise.
28769 * tree-vectorizer.h: Likewise.
28771 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28773 * passes.c (emergency_dump_function): New.
28774 * tree-pass.h (emergency_dump_function): Declare.
28775 * plugin.c (plugins_internal_error_function): Remove.
28776 * plugin.h (plugins_internal_error_function): Remove declaration.
28777 * toplev.c (internal_error_function): New static function. Use it...
28778 (general_init): ...here.
28780 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28782 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28783 (arc_rtx_costs): Add costs for new patterns.
28784 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28785 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28787 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28789 * tree-ssa-strlen.c (get_next_strinfo): New function.
28790 (get_stridx_plus_constant): Use it.
28791 (zero_length_string): Likewise.
28792 (adjust_related_strinfos): Likewise.
28793 (adjust_last_stmt): Likewise.
28795 2017-05-31 Richard Biener <rguenther@suse.de>
28798 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28799 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28801 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28803 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28804 loop_vinfo argument and use of dependence distance vectors.
28805 Check instead whether the two references differ only in their
28806 initial value and assume that they have the same alignment if the
28807 difference is a multiple of the vector alignment.
28808 (vect_analyze_data_refs_alignment): Update call accordingly.
28810 2017-05-31 Martin Liska <mliska@suse.cz>
28813 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28815 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28817 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28818 (create_intersect_range_checks): Move from ...
28819 * tree-data-ref.c (create_intersect_range_checks_index)
28820 (create_intersect_range_checks): ... to here.
28821 (create_runtime_alias_checks): New function factored from ...
28822 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28823 here. Call above function.
28824 * tree-data-ref.h (create_runtime_alias_checks): New function.
28826 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28828 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28829 segment length for dr_b and compute it in wide_int.
28831 2017-05-31 Richard Biener <rguenther@suse.de>
28833 PR tree-optimization/80906
28834 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28835 and pass through iv_map.
28836 (copy_bb_and_scalar_dependences): Adjust.
28837 (translate_pending_phi_nodes): Likewise.
28838 (copy_loop_close_phi_args): Handle code-generating IVs instead
28841 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28843 * diagnostic-color.c (color_dict): Add "type-diff".
28844 (parse_gcc_colors): Update comment.
28845 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28846 -fdiagnostics-show-template-tree and -fno-elide-type.
28847 (GCC_COLORS): Add type-diff to example.
28849 (-fdiagnostics-show-template-tree): New.
28850 (-fno-elide-type): New.
28851 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28852 the pp_format_decoder callback. Call any m_format_postprocessor's
28854 (pretty_printer::pretty_printer): Initialize
28855 m_format_postprocessor.
28856 (pretty_printer::~pretty_printer): Delete any
28857 m_format_postprocessor.
28858 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28859 (class format_postprocessor): New class.
28860 (struct pretty_printer::format_decoder): Document the new parameters.
28861 (struct pretty_printer::m_format_postprocessor): New field.
28862 * tree-diagnostic.c (default_tree_printer): Update for new
28863 bool and const char ** params.
28864 * tree-diagnostic.h (default_tree_printer): Likewise.
28866 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28868 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28869 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28870 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28871 rs6000_gen_cell_microcode code.
28872 (rs6000_final_prescan_insn): Delete.
28873 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28874 "warn-cell-microcode" entries.
28875 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28876 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28877 throughout. Change cc_reg_not_micro_cr0_operand to
28878 cc_reg_not_cr0_operand throughout.
28879 (*extendhi<mode>2_noload): Delete.
28880 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28881 (mwarn-cell-microcode): Delete.
28882 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28883 -mgen-cell-microcode and -mwarn-cell-microcode.
28885 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28888 * config/i386/constraints.md (Yd): New constraint.
28890 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28891 and (?Yd, r) alternatives. Update insn attributes.
28892 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28893 and (?*Yd, r) alternatives. Update insn attributes.
28894 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28896 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28898 * gimplify.c (gimplify_modify_expr): Don't create a
28899 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28902 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28904 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28906 2017-05-30 Richard Biener <rguenther@suse.de>
28908 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28909 and reduc_def fields.
28910 (STMT_VINFO_REDUC_TYPE): New define.
28911 (STMT_VINFO_REDUC_DEF): Likewise.
28912 (vect_force_simple_reduction): Adjust prototype.
28913 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28914 (vect_is_simple_reduction): Remove check_reduction argument.
28915 (vect_force_simple_reduction): Adjust and set
28916 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28917 (vectorizable_reduction): Do not re-do reduction analysis
28918 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28919 * tree-parloops.c (gather_scalar_reductions): Adjust.
28921 2017-05-30 Richard Biener <rguenther@suse.de>
28923 PR middle-end/80901
28924 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28927 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28929 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28930 Introduce unknown_misalignment parameter and remove vf.
28931 (vect_peeling_hash_get_lowest_cost):
28932 Pass unknown_misalignment parameter.
28933 (vect_enhance_data_refs_alignment):
28934 Fix unsupportable data ref treatment.
28936 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28938 * tree-vect-data-refs.c (vect_get_data_access_cost):
28939 Workaround for SLP handling.
28940 (vect_enhance_data_refs_alignment):
28941 Compute costs for doing no peeling at all, compare to the best
28942 peeling costs so far and avoid peeling if cheaper.
28944 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28946 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28947 Return peeling info and set costs to zero for unlimited cost
28949 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28950 with unknown misalignment. Compute and costs for unknown
28951 misalignment, compare them to the costs for known misalignment
28952 and choose the cheapest for peeling.
28954 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28956 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28957 (vect_get_peeling_costs_all_drs): Create function.
28958 (vect_peeling_hash_get_lowest_cost):
28959 Use vect_get_peeling_costs_all_drs.
28960 (vect_peeling_supportable): Create function.
28961 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28963 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28965 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28966 DR_HAS_NEGATIVE_STEP.
28967 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28968 (vect_enhance_data_refs_alignment): Use.
28969 (vect_duplicate_ssa_name_ptr_info): Use.
28970 * tree-vectorizer.h (dr_misalignment): Use.
28971 (known_alignment_for_access_p): Use.
28973 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28976 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28978 (has_section_name): New function.
28980 2017-05-30 Martin Liska <mliska@suse.cz>
28983 * auto-profile.c (get_function_decl_from_block): Fix
28986 2017-05-30 Richard Biener <rguenther@suse.de>
28988 PR middle-end/80876
28989 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28991 2017-05-30 Martin Liska <mliska@suse.cz>
28993 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28994 * dumpfile.h (struct dump_file_info): Remove ctors.
28996 2017-05-30 Martin Liska <mliska@suse.cz>
28998 * predict.def: Fix GNU coding style.
29000 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29002 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29003 Mark 'to' argument with ATTRIBUTE_UNUSED.
29005 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29007 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29008 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29009 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29012 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29014 * doc/install.texi (Options specification): Restore entry of
29015 --enable-sjlj-exceptions.
29017 2017-05-27 Michael Eager <eager@eagercon.com>
29020 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29022 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29024 * config/microblaze/microblaze.h
29025 (FIXED_REGISTERS): Update in macro.
29026 (CALL_USED_REGISTERS): Update in macro.
29028 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29030 * doc/install.texi: Add links to macOS binary distributions.
29032 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29036 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29038 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29040 2017-05-26 Martin Liska <mliska@suse.cz>
29042 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29044 2017-05-26 Martin Liska <mliska@suse.cz>
29046 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29047 always leading ';; '.
29048 (dump_bb_info): Likewise.
29049 (brief_dump_cfg): Likewise.
29050 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29051 * dumpfile.c: Remove usage of TDF_VERBOSE.
29052 * dumpfile.h (enum dump_kind): Likewise.
29053 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29054 * print-tree.c (debug_verbose): Remove.
29055 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29056 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29057 * tree-diagnostic.c (default_tree_printer): Replace
29058 TDF_DIAGNOSTIC with TDF_SLIM.
29060 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29062 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29063 in parameter loop, rather than loop_vinfo.
29064 (create_intersect_range_checks): Ditto.
29065 (vect_create_cond_for_alias_checks): Update call to above functions.
29067 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29069 PR tree-optimization/80815
29070 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29071 for merging runtime alias checks. Handle negative DR_STEPs.
29073 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29075 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29077 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29078 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29079 out code pruning runtime alias checks.
29080 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29081 factored out from above.
29082 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29084 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29086 (prune_runtime_alias_test_list): New decalaration.
29088 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29090 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29091 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29092 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29093 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29094 (operator==, comp_dr_with_seg_len_pair): Ditto.
29095 (vect_prune_runtime_alias_test_list): Ditto.
29097 2017-05-26 Martin Liska <mliska@suse.cz>
29100 * params.def: Bound partial-inlining-entry-probability param.
29102 2017-05-26 Marek Polacek <polacek@redhat.com>
29105 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29108 2017-05-26 Richard Biener <rguenther@suse.de>
29110 PR tree-optimization/80842
29111 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29114 2017-05-26 Richard Biener <rguenther@suse.de>
29116 PR tree-optimization/80844
29117 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29119 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29121 * doc/md.texi (Machine Constraints): Update x86 family
29122 machine constraints section to match 'config/i386/constraints.md'.
29124 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29126 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29128 2017-05-25 Nathan Sidwell <nathan@acm.org>
29130 * doc/invoke.texi (--enable-languages): Update documentation.
29132 2017-05-25 Martin Liska <mliska@suse.cz>
29134 * dumpfile.c: Add TDF_FOLDING.
29135 * dumpfile.h (enum dump_kind): Likewise.
29136 * genmatch.c (dt_simplify::gen_1): Use it.
29138 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29140 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29142 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29144 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29145 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29147 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29149 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29150 * match.pd (X == C): Rewrite it here.
29151 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29152 with_certain_nonzero_bits2): New predicates.
29153 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29155 2017-05-24 Nathan Sidwell <nathan@acm.org>
29157 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29160 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29163 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29165 * config/powerpcspe: New port. Files are copied from the rs6000
29166 port, with "rs6000" in filenames replaced by "powerpcspe".
29168 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29170 PR rtl-optimization/80754
29171 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29173 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29175 * config/sparc/sparc.md (length): Return the correct value for -mflat
29176 sibcalls to match output_sibcall.
29178 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29182 * config/rs6000/rs6000.c (struct machine_function): Add new field
29184 (rs6000_get_separate_components): Init that field, use it.
29185 (rs6000_components_for_bb): Use the field.
29187 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29189 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29191 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29193 PR middle-end/80823
29194 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29196 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29199 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29200 address against address_operand predicate.
29201 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29203 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29205 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29206 parameters passed indirectly.
29208 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29210 * config/i386/i386.md (*movdi_internal): Remove SSE4
29211 alternative 18 (?r, *v). Update insn attributes.
29212 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29213 Update insn attributes.
29214 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29215 Update insn attributes.
29216 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29217 alternative 1 (r, v). Remove isa attribute.
29218 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29219 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29220 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29222 2017-05-23 Tom de Vries <tom@codesourcery.com>
29224 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29227 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29229 * cgraphunit.c (symbol_table::process_new_functions): Update.
29230 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29231 (inline_generate_summary): Rename to ...
29232 (ipa_fn_summary_generate): ... this one.
29233 (inline_read_summary): Rename to ...
29234 (ipa_fn_summary_read): ... this one.
29235 (inline_write_summary): Rename to ...
29236 (ipa_fn_summary_write): ... this one.
29237 (inline_free_summary): Rename to ...
29238 (ipa_free_fn_summary): ... this one.
29239 (pass_data_local_fn_summary, pass_local_fn_summary,
29240 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29241 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29242 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29243 make_pass_ipa_fn_summary): New.
29244 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29245 inline_write_summary, inline_free_summary): Remove.
29246 (ipa_free_fn_summary) : New.
29247 * ipa-inline.c (ipa_inline): Update.
29248 (pass_ipa_inline): Do not generate summaries.
29249 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29251 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29252 and add pass_ipa_fn_summary.
29253 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29255 (make_pass_inline_parameters): Remove.
29257 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29259 * omp-low.c (struct omp_context): Remove "default_kind" member.
29262 * omp-offload.c (execute_oacc_device_lower): Remove the
29263 parallelism dimensions function attributes for unparallelized
29264 OpenACC kernels constructs.
29266 2017-05-23 Martin Liska <mliska@suse.cz>
29268 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29270 (cgraph_edge::make_speculative): Likewise.
29271 (cgraph_edge::resolve_speculation): Likewise.
29272 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29273 (cgraph_node::dump): Likewise.
29274 * cgraph.h: Likewise.
29275 * cgraphunit.c (analyze_functions): Likewise.
29276 (symbol_table::compile): Likewise.
29277 * ipa-cp.c (print_all_lattices): Likewise.
29278 (determine_versionability): Likewise.
29279 (initialize_node_lattices): Likewise.
29280 (ipcp_verify_propagated_values): Likewise.
29281 (estimate_local_effects): Likewise.
29282 (update_profiling_info): Likewise.
29283 (create_specialized_node): Likewise.
29284 (perhaps_add_new_callers): Likewise.
29285 (decide_about_value): Likewise.
29286 (decide_whether_version_node): Likewise.
29287 (identify_dead_nodes): Likewise.
29288 (ipcp_store_bits_results): Likewise.
29289 * ipa-devirt.c (dump_targets): Likewise.
29290 (ipa_devirt): Likewise.
29291 * ipa-icf.c (sem_item::dump): Likewise.
29292 (sem_function::equals): Likewise.
29293 (sem_variable::equals): Likewise.
29294 (sem_item_optimizer::read_section): Likewise.
29295 (sem_item_optimizer::execute): Likewise.
29296 (congruence_class::dump): Likewise.
29297 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29298 (dump_inline_summary): Likewise.
29299 (estimate_node_size_and_time): Likewise.
29300 (inline_analyze_function): Likewise.
29301 * ipa-inline-transform.c (inline_call): Likewise.
29302 * ipa-inline.c (report_inline_failed_reason): Likewise.
29303 (want_early_inline_function_p): Likewise.
29304 (edge_badness): Likewise.
29305 (update_edge_key): Likewise.
29306 (inline_small_functions): Likewise.
29307 * ipa-profile.c (ipa_profile): Likewise.
29308 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29309 (ipa_make_edge_direct_to_target): Likewise.
29310 (remove_described_reference): Likewise.
29311 (ipa_impossible_devirt_target): Likewise.
29312 (propagate_controlled_uses): Likewise.
29313 (ipa_print_node_params): Likewise.
29314 (ipcp_transform_function): Likewise.
29315 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29316 (propagate_pure_const): Likewise.
29317 * ipa-reference.c (generate_summary): Likewise.
29318 (read_write_all_from_decl): Likewise.
29319 (propagate): Likewise.
29320 (ipa_reference_read_optimization_summary): Likewise.
29321 * ipa-utils.c (ipa_merge_profiles): Likewise.
29322 * ipa.c (walk_polymorphic_call_targets): Likewise.
29323 (symbol_table::remove_unreachable_nodes): Likewise.
29324 (ipa_single_use): Likewise.
29325 * passes.c (execute_todo): Likewise.
29326 * predict.c (drop_profile): Likewise.
29327 * symtab.c (symtab_node::get_dump_name): New function.
29328 (symtab_node::dump_name): Likewise.
29329 (symtab_node::dump_asm_name): Likewise.
29330 (symtab_node::dump_references): Likewise.
29331 (symtab_node::dump_referring): Likewise.
29332 (symtab_node::dump_base): Likewise.
29333 (symtab_node::debug_symtab): Likewise.
29334 * tree-sra.c (convert_callers_for_node): Likewise.
29335 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29336 * value-prof.c (init_node_map): Likewise.
29338 2017-05-23 Martin Liska <mliska@suse.cz>
29340 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29341 and symtab_node::debug_symtab to symbol_table::debug.
29342 * cgraphunit.c (analyze_functions): Use the renamed function.
29343 (symbol_table::compile): Likewise.
29344 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29345 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29346 * passes.c (execute_todo): Likewise.
29347 * symtab.c (symbol_table::dump): New function.
29348 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29350 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29352 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29353 that nonconst implies exec.
29355 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29357 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29358 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29359 (inline_edge_summary_vec): Turn into ...
29360 (ipa_call_summaries): ... this one.
29361 (redirect_to_unreachable, edge_set_predicate,
29362 evaluate_properties_for_edge, inline_summary_alloc,
29363 reset_ipa_call_summary, reset_inline_summary,
29364 inline_summary_t::duplicate): Update.
29365 (inline_edge_duplication_hook): Turn to ...
29366 (ipa_call_summary_t::duplicate): ... this one.
29367 (inline_edge_removal_hook): Turn to ...
29368 (ipa_call_summary_t::remove): ... this one.
29369 (dump_inline_edge_summary): Turn to ...
29370 (dump_ipa_call_summary): ... this one.
29371 (estimate_function_body_sizes): Update.
29372 (inline_update_callee_summaries): Update.
29373 (remap_edge_change_prob): Update.
29374 (remap_edge_summaries): Update.
29375 (inline_merge_summary): Update.
29376 (do_estimate_edge_time): Update.
29377 (inline_generate_summary): Update.
29378 (inline_read_section): Update.
29379 (inline_read_summary): Update.
29380 (inline_free_summary): Update.
29381 * ipa-inline.c (can_inline_edge_p): Update.
29382 (compute_inlined_call_time): Update.
29383 (want_inline_small_function_p): Update.
29384 (edge_badness): Update.
29385 (early_inliner): Update.
29386 * ipa-inline.h (inline_edge_summary): Turn to ...
29387 (ipa_call_summary): ... this one.
29388 (ipa_call_summary_t): New class.
29389 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29390 (ipa_call_summaries): New.
29391 (inline_edge_summary): Remove.
29392 (estimate_edge_growth): Update.
29393 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29394 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29395 * ipa-split.c (execute_split_functions): Update.
29396 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29398 2017-05-23 Tom de Vries <tom@codesourcery.com>
29400 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29401 attributes): Document rdrand effective target.
29403 2017-05-23 Tom de Vries <tom@codesourcery.com>
29405 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29406 attributes): Sort alphabetically.
29408 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29410 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29412 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29415 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29416 V2DF/V2DI splat into two separate patterns, one that handles
29417 registers, and the other that only handles memory. Drop support
29418 for splatting from a GPR on ISA 2.07 and then splitting the
29419 splat into direct move and splat.
29420 (vsx_splat_<mode>_reg): Likewise.
29421 (vsx_splat_<mode>_mem): Likewise.
29423 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29425 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29427 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29429 PR middle-end/80809
29430 * omp-low.c (finish_taskreg_remap): New function.
29431 (finish_taskreg_scan): If unit size of ctx->record_type
29432 is non-constant, unshare the size expression and replace
29433 decls in it with possible outer var refs.
29435 PR middle-end/80809
29436 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29437 GOVD_SHARED rather than GOVD_PRIVATE with it.
29438 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29439 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29441 PR middle-end/80853
29442 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29443 as last argument to build_outer_var_ref for pointer bases of array
29444 section reductions.
29446 2017-05-19 Martin Sebor <msebor@redhat.com>
29448 * print-tree.c (print_node): Print DECL_READ_P flag.
29450 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29452 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29453 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29454 * cgraph.c: Likewise.
29455 * cgraphunit.c: Likewise.
29456 * gengtype.c: Likewise.
29457 * ipa-cp.c: Likewise.
29458 * ipa-devirt.c: Likewise.
29459 * ipa-icf.c: Likewise.
29460 * ipa-predicate.c: Likewise.
29461 * ipa-profile.c: Likewise.
29462 * ipa-prop.c: Likewise.
29463 * ipa-split.c: Likewise.
29465 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29466 edge_predicate_pool, dump_inline_hints,
29467 inline_summary::account_size_time, redirect_to_unreachable,
29468 edge_set_predicate, set_hint_predicate,
29469 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29470 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29471 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29472 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29473 ipa_call_summary_t::remove, initialize_growth_caches,
29474 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29475 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29476 mark_modified, unmodified_parm_1, unmodified_parm,
29477 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29478 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29479 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29480 will_be_nonconstant_predicate, record_modified_bb_info,
29481 get_minimal_bb, record_modified, param_change_prob,
29482 phi_result_unknown_predicate, predicate_for_phi_result,
29483 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29484 estimate_function_body_sizes, compute_inline_parameters,
29485 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29486 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29487 inline_update_callee_summaries, remap_edge_change_prob,
29488 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29489 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29490 inline_analyze_function, inline_summary_t::insert,
29491 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29492 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29493 inline_free_summary): Move to ipa-fnsummary.h
29494 (predicate_t): Remove.
29495 * ipa-fnsummary.c: New file.
29496 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29497 (enum inline_hints_vals, inline_hints, agg_position_info,
29498 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29499 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29500 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29501 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29502 inline_read_summary, inline_write_summary, inline_free_summary,
29503 inline_analyze_function, initialize_inline_failed,
29504 inline_merge_summary, inline_update_overall_summary,
29505 compute_inline_parameters): Move to ipa-fnsummary.h
29506 * ipa-fnsummary.h: New file.
29507 * ipa-inline-transform.h: Include ipa-inline.h.
29508 * ipa-inline.c: LIkewise.
29510 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29512 * ipa-inline.c (edge_badness): Use inlined_time instead of
29513 inline_summaries->get.
29515 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29517 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29519 2017-05-22 Nathan Sidwell <nathan@acm.org>
29521 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29522 (fdump-lang): Document 'raw' option.
29523 * dumpfile.h (TDI_tu): Delete.
29524 * dumpfile.c (dump_files): Remove translation-unit.
29525 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29527 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29529 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29530 command option from $(AWK) call.
29531 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29532 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29533 [FORMAT]: Remove handling of variable.
29534 * config/avr/t-multilib: Regenerate.
29536 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29538 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29540 (dump_inline_summary): Do not print self_time.
29541 (estimate_function_body_sizes): Do not set self_time.
29542 (compute_inline_parameters): Likewise.
29543 (inline_read_section, inline_write_summary): Do not stream self_time.
29544 * ipa-inline.h (inline_summary): Drop self_time.
29546 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29548 * ipa-inline-analysis.c (account_size_time): Rename to ...
29549 (inline_summary::account_size_time): ... this one.
29550 (reset_ipa_call_summary): Turn to ...
29551 (ipa_call_summary::reset): ... this one.
29552 (reset_inline_summary): Turn to ...
29553 (inline_summary::reset): ... this one.
29554 (inline_summary_t::remove): Update.
29555 (inline_summary_t::duplicate): Update.
29556 (ipa_call_summary_t::remove): Update.
29557 (dump_inline_summary): Update.
29558 (estimate_function_body_sizes): Update.
29559 (compute_inline_parameters): Update.
29560 (estimate_node_size_and_time): Update.
29561 (inline_merge_summary): Update.
29562 (inline_update_overall_summary): Update.
29563 (inline_read_section): Update.
29564 (inline_write_summary): Update.
29565 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29566 add account_size_time and reset member functions.
29567 (ipa_call_summary): Add reset function.
29568 * ipa-predicate.h (predicate::operator &): Constify.
29570 2017-05-22 Richard Biener <rguenther@suse.de>
29572 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29574 2017-05-19 Jason Merrill <jason@redhat.com>
29576 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29578 2017-05-19 Marek Polacek <polacek@redhat.com>
29581 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29582 TYPE_OVERFLOW_WRAPS checks.
29584 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29586 * tree-core.h (enum omp_clause_default_kind): Add
29587 "OMP_CLAUSE_DEFAULT_PRESENT".
29588 * tree-pretty-print.c (dump_omp_clause): Handle it.
29589 * gimplify.c (enum gimplify_omp_var_data): Add
29590 "GOVD_MAP_FORCE_PRESENT".
29591 (gimplify_adjust_omp_clauses_1): Map it to
29592 "GOMP_MAP_FORCE_PRESENT".
29593 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29595 * gimplify.c (oacc_default_clause): Clarify.
29597 2017-05-19 Nathan Sidwell <nathan@acm.org>
29599 LANG_HOOK_REGISTER_DUMPS
29600 * toplev.c (general_init): Call register dump lang hook.
29601 * doc/invoke.texi: Document -fdump-lang option family.
29602 * dumpfile.c (dump_files): Remove class dump here.
29603 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29604 * dumpfile.h (tree_dump_index): Remove TDI_class.
29605 * langhooks-def.h (lhd_register_dumps): Declare.
29606 (LANG_HOOKS_REGISTER_DUMPS): Define.
29607 (LANG_HOOKS_INITIALIZER): Add it.
29608 * langhooks.c (lhd_register_dumps): Define.
29609 * langhooks.h (struct lang_hooks): Add register_dumps.
29611 2017-05-19 Nathan Sidwell <nathan@acm.org>
29613 * context.h (context::set_passes): New.
29614 * context.c (context::context): Do not create pass manager.
29615 * toplev.c (general_init): Create pass manager here.
29617 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29619 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29620 use this splitter if two add or or instructions would also work for
29621 the constant we want to generate.
29623 2017-05-19 Richard Biener <rguenther@suse.de>
29626 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29627 predicate evaluation.
29629 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29631 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29633 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29636 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29638 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29639 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29640 (inline_edge_summary_vec): Turn into ...
29641 (ipa_call_summaries): ... this one.
29642 (redirect_to_unreachable, edge_set_predicate,
29643 evaluate_properties_for_edge, inline_summary_alloc,
29644 reset_ipa_call_summary, reset_inline_summary,
29645 inline_summary_t::duplicate): Update.
29646 (inline_edge_duplication_hook): Turn to ...
29647 (ipa_call_summary_t::duplicate): ... this one.
29648 (inline_edge_removal_hook): Turn to ...
29649 (ipa_call_summary_t::remove): ... this one.
29650 (dump_inline_edge_summary): Turn to ...
29651 (dump_ipa_call_summary): ... this one.
29652 (estimate_function_body_sizes): Update.
29653 (inline_update_callee_summaries): Update.
29654 (remap_edge_change_prob): Update.
29655 (remap_edge_summaries): Update.
29656 (inline_merge_summary): Update.
29657 (do_estimate_edge_time): Update.
29658 (inline_generate_summary): Update.
29659 (inline_read_section): Update.
29660 (inline_read_summary): Update.
29661 (inline_free_summary): Update.
29662 * ipa-inline.c (can_inline_edge_p): Update.
29663 (compute_inlined_call_time): Update.
29664 (want_inline_small_function_p): Update.
29665 (edge_badness): Update.
29666 (early_inliner): Update.
29667 * ipa-inline.h (inline_edge_summary): Turn to ...
29668 (ipa_call_summary): ... this one.
29669 (ipa_call_summary_t): New class.
29670 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29671 (ipa_call_summaries): New.
29672 (inline_edge_summary): Remove.
29673 (estimate_edge_growth): Update.
29674 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29675 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29676 * ipa-split.c (execute_split_functions): Update.
29677 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29679 2017-05-19 Richard Biener <rguenther@suse.de>
29681 PR middle-end/80764
29682 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29684 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29686 * config/rs6000/rs6000.c (struct machine_function): Add field
29687 fpr_is_wrapped_separately.
29688 (rs6000_get_separate_components): Use 64 components. Handle the
29689 new FPR components.
29690 (rs6000_components_for_bb): Handle the FPR components.
29691 (rs6000_emit_prologue_components): Handle the FPR components.
29692 (rs6000_emit_epilogue_components): Handle the FPR components.
29693 (rs6000_set_handled_components): Handle the FPR components.
29694 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29695 that are already separately shrink-wrapped.
29696 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29697 that are already separately shrink-wrapped.
29699 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29702 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29705 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29706 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29707 cases where the register allocator uses a move and an offsettable
29708 memory operation to/from a FPR register on ISA 2.06/2.07.
29709 (define_peephole2 for Altivec d-form store): Likewise.
29711 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29714 * config/i386/mmx.md (*mov<mode>_internal): Enable
29715 alternatives 11, 12, 13 and 14 also for 32bit targets.
29716 Remove alternatives 15, 16, 17 and 18.
29717 * config/i386/sse.md (vec_concatv2di): Change
29718 alternative (!x, *y) to (x, ?!*Yn).
29720 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29722 * dumpfile.h (enum dump_kind): Remove stray comma.
29724 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29726 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29727 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29728 predicate::num_conditions
29729 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29730 (CHANGED): turn into predicate::changed.
29731 (agg_position_info): Move to ipa-predicate.h
29732 (add_condition, predicate::add_clause, predicate::operator &=,
29733 predicate::or_with, predicate::evaluate, predicate::probability,
29734 dump_condition, dump_clause, predicate::dump,
29735 predicate::remap_after_duplication, predicate::remap_after_inlining,
29736 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29737 (evaluate_conditions_for_known_args): Update.
29738 (set_cond_stmt_execution_predicate): Update.
29739 * ipa-inline.h: Include ipa-predicate.h
29740 (condition, inline_param_summary, conditions, agg_position_info,
29741 predicate): Move to ipa-predicate.h
29742 * ipa-predicate.c: New file.
29743 * ipa-predicate.h: New file.
29745 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29747 * final.c (leaf_function_p): Check we are not in a sequence.
29749 2017-05-18 Martin Liska <mliska@suse.cz>
29751 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29752 * dumpfile.c (dump_register): Use new enum dump_kind.
29753 (get_dump_file_name): Likewise.
29754 (dump_enable_all): Likewise.
29755 (dump_switch_p_1): Likewise.
29756 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29757 * dumpfile.h (enum dump_kind): New enum type.
29758 (struct dump_file_info): Create constructor and
29759 format fields and comments.
29760 * passes.c (pass_manager::register_one_dump_file):
29762 * statistics.c (statistics_early_init): Likewise.
29763 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29764 TDF_TREE with TDF_SLIM.
29765 (gather_memory_references_ref): Likewise.
29767 2017-05-18 Martin Liska <mliska@suse.cz>
29769 * vec.h (struct vnull): Use it.
29771 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29773 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29774 (true_predicate, false_predicate, true_predicate_p,
29775 false_predicate_p): Remove.
29776 (single_cond_predicate, not_inlined_predicate): Turn to member function
29778 (add_condition): Update.
29779 (add_clause): Turn to...
29780 (predicate::add_clause): ... this one; update; allow passing NULL
29782 (and_predicates): Turn to ...
29783 (predicate::operator &=): ... this one.
29784 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29785 (or_predicates): Turn to ...
29786 (predicate::or_with): ... this one.
29787 (evaluate_predicate): Turn to ...
29788 (predicate::evaluate): ... this one.
29789 (predicate_probability): Turn to ...
29790 (predicate::probability): ... this one.
29791 (dump_condition): Update.
29792 (dump_predicate): Turn to ...
29793 (predicate::dump): ... this one.
29794 (account_size_time): Update.
29795 (edge_set_predicate): Update.
29796 (set_hint_predicate): UPdate.
29797 (evaluate_conditions_for_known_args): Update.
29798 (evaluate_properties_for_edge): Update.
29799 (remap_predicate_after_duplication): Turn to...
29800 (predicate::remap_after_duplication): ... this one.
29801 (remap_hint_predicate_after_duplication): Update.
29802 (inline_summary_t::duplicate): UPdate.
29803 (dump_inline_edge_summary): Update.
29804 (dump_inline_summary): Update.
29805 (set_cond_stmt_execution_predicate): Update.
29806 (set_switch_stmt_execution_predicate): Update.
29807 (compute_bb_predicates): Update.
29808 (will_be_nonconstant_expr_predicate): Update.
29809 (will_be_nonconstant_predicate): Update.
29810 (phi_result_unknown_predicate): Update.
29811 (predicate_for_phi_result): Update.
29812 (array_index_predicate): Update.
29813 (estimate_function_body_sizes): Update.
29814 (estimate_node_size_and_time): Update.
29815 (estimate_ipcp_clone_size_and_time): Update.
29816 (remap_predicate): Rename to ...
29817 (predicate::remap_after_inlining): ... this one.
29818 (remap_hint_predicate): Update.
29819 (inline_merge_summary): Update.
29820 (inline_update_overall_summary): Update.
29821 (estimate_size_after_inlining): Update.
29822 (read_predicate): Rename to ...
29823 (predicate::stream_in): ... this one.
29824 (read_inline_edge_summary): Update.
29825 (write_predicate): Rename to ...
29826 (predicate::stream_out): ... this one.
29827 (write_inline_edge_summary): Update.
29828 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29829 (clause_t): Turn to uint32_t
29830 (predicate): Turn to class; implement constructor and operators
29832 (size_time_entry): Update.
29833 (inline_summary): Update.
29834 (inline_edge_summary): Update.
29836 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29838 * fold-const.c (fold_binary_loc): Move transformation...
29839 * match.pd (C - X CMP X): ... here.
29841 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29843 * config/sparc/sparc.c (sparc_option_override): Set function
29844 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29845 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29847 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29849 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29851 2017-05-18 Marek Polacek <polacek@redhat.com>
29854 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29855 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29856 gimple_assign_load_p.
29858 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29860 PR middle-end/80692
29861 * real.c (do_compare): Give decimal_do_compare preference over
29862 comparing just the signs.
29864 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29866 * doc/md.texi (Canonicalization of Instructions): Describe the
29867 canonical form of instructions that inherently set a condition
29870 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29872 PR middle-end/80775
29873 * tree-cfg.c: Move deletion of unreachable case statements to after
29874 the merging of consecutive case labels.
29876 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29878 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29879 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29880 restoring of callee-saved registers.
29882 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29884 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29885 * config/visium/visium.c (single_set_and_flags): Likewise.
29886 * config/visium/visium.md (Substitutions): Likewise.
29888 2017-05-17 Martin Liska <mliska@suse.cz>
29890 * cfg.c: Introduce dump_flags_t type and
29891 use it instead of int type.
29893 * cfghooks.c: Likewise.
29894 * cfghooks.h (struct cfg_hooks): Likewise.
29895 * cfgrtl.c: Likewise.
29896 * cfgrtl.h: Likewise.
29897 * cgraph.c (cgraph_node::get_body): Likewise.
29898 * coretypes.h: Likewise.
29899 * domwalk.c: Likewise.
29900 * domwalk.h: Likewise.
29901 * dumpfile.c (struct dump_option_value_info): Likewise.
29902 (dump_enable_all): Likewise.
29903 (dump_switch_p_1): Likewise.
29904 (opt_info_switch_p): Likewise.
29905 * dumpfile.h (enum tree_dump_index): Likewise.
29906 (struct dump_file_info): Likewise.
29907 * genemit.c: Likewise.
29908 * generic-match-head.c: Likewise.
29909 * gengtype.c (open_base_files): Likewise.
29910 * gimple-pretty-print.c: Likewise.
29911 * gimple-pretty-print.h: Likewise.
29912 * graph.c (print_graph_cfg): Likewise.
29913 * graphite-scop-detection.c (dot_all_sese): Likewise.
29914 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29915 * loop-unroll.c (report_unroll): Likewise.
29916 * passes.c (pass_manager::register_one_dump_file): Likewise.
29917 * print-tree.c: Likewise.
29918 * statistics.c: Likewise.
29919 * tree-cfg.c: Likewise.
29920 * tree-cfg.h: Likewise.
29921 * tree-dfa.c: Likewise.
29922 * tree-dfa.h: Likewise.
29923 * tree-dump.c (dump_function): Likewise.
29924 * tree-dump.h (struct dump_info): Likewise.
29925 * tree-pretty-print.c: Likewise.
29926 * tree-pretty-print.h: Likewise.
29927 * tree-ssa-live.c: Likewise.
29928 * tree-ssa-live.h: Likewise.
29929 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29930 * tree-vect-loop.c: Likewise.
29931 * tree-vect-slp.c: Likewise.
29933 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29934 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29936 PR tree-optimization/80457
29937 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29938 of all arguments to a statement as scalar_to_vec operations.
29939 (vectorizable_call): Adjust call to vect_model_simple_cost for
29941 (vectorizable_conversion): Likewise.
29942 (vectorizable_assignment): Likewise.
29943 (vectorizable_shift): Likewise.
29944 (vectorizable_operation): Likewise.
29945 (vectorizable_comparison): Likewise.
29946 (vect_is_simple_cond): Record the def types for operands.
29947 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29948 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29949 for statement argument count.
29951 2017-05-16 Carl Love <cel@us.ibm.com>
29953 * config/rs6000/rs6000-c: Add support for built-in functions
29954 vector unsigned long long vec_bperm (vector unsigned long long,
29955 vector unsigned char)
29956 vector signed long long vec_mule (vector signed int,
29958 vector unsigned long long vec_mule (vector unsigned int,
29959 vector unsigned int)
29960 vector signed long long vec_mulo (vector signed int,
29962 vector unsigned long long vec_mulo (vector unsigned int,
29963 vector unsigned int)
29964 vector signed char vec_sldw (vector signed char,
29965 vector signed char,
29967 vector unsigned char vec_sldw (vector unsigned char,
29968 vector unsigned char,
29970 vector signed short vec_sldw (vector signed short,
29971 vector signed short,
29973 vector unsigned short vec_sldw (vector unsigned short,
29974 vector unsigned short,
29976 vector signed int vec_sldw (vector signed int,
29979 vector unsigned int vec_sldw (vector unsigned int,
29980 vector unsigned int,
29982 vector signed long long vec_sldw (vector signed long long,
29983 vector signed long long,
29985 vector unsigned long long vec_sldw (vector unsigned long long,
29986 vector unsigned long long,
29988 * config/rs6000/rs6000-c: Add support for built-in functions
29989 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29990 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29991 * doc/extend.texi: Update the built-in documentation for the
29992 new built-in functions.
29994 2017-05-16 Marek Polacek <polacek@redhat.com>
29998 * tree.c (save_expr): Don't fold the expression.
30000 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30002 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30003 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30004 and (?*y,m). Update insn attributes.
30006 2017-05-16 Martin Liska <mliska@suse.cz>
30008 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30009 flags argument of print_gimple_stmt, print_gimple_expr,
30010 print_generic_stmt and print_generic_expr.
30011 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30012 * coretypes.h: Likewise.
30013 * except.c (dump_eh_tree): Likewise.
30014 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30015 * gimple-pretty-print.h: Likewise.
30016 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30017 (backprop::push_to_worklist): Likewise.
30018 (backprop::pop_from_worklist): Likewise.
30019 (backprop::process_use): Likewise.
30020 (backprop::intersect_uses): Likewise.
30021 (note_replacement): Likewise.
30022 * gimple-ssa-store-merging.c
30023 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30024 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30025 (pass_store_merging::execute): Likewise.
30026 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30027 (ssa_base_cand_dump_callback): Likewise.
30028 (dump_incr_vec): Likewise.
30029 (replace_refs): Likewise.
30030 (replace_mult_candidate): Likewise.
30031 (create_add_on_incoming_edge): Likewise.
30032 (create_phi_basis): Likewise.
30033 (insert_initializers): Likewise.
30034 (all_phi_incrs_profitable): Likewise.
30035 (introduce_cast_before_cand): Likewise.
30036 (replace_one_candidate): Likewise.
30037 * gimplify.c (gimplify_expr): Likewise.
30038 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30039 (set_rename): Likewise.
30040 (rename_uses): Likewise.
30041 (copy_loop_phi_nodes): Likewise.
30042 (add_close_phis_to_merge_points): Likewise.
30043 (copy_loop_close_phi_args): Likewise.
30044 (copy_cond_phi_args): Likewise.
30045 (graphite_copy_stmts_from_block): Likewise.
30046 (translate_pending_phi_nodes): Likewise.
30047 * graphite-poly.c (print_pdr): Likewise.
30048 (dump_gbb_cases): Likewise.
30049 (dump_gbb_conditions): Likewise.
30050 (print_scop_params): Likewise.
30051 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30052 (build_cross_bb_scalars_use): Likewise.
30053 (gather_bbs::before_dom_children): Likewise.
30054 * hsa-dump.c (dump_hsa_immed): Likewise.
30055 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30056 (get_replacement_map): Likewise.
30057 * ipa-inline-analysis.c (dump_condition): Likewise.
30058 (estimate_function_body_sizes): Likewise.
30059 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30060 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30061 * ipa-prop.c (ipa_dump_param): Likewise.
30062 (ipa_print_node_jump_functions_for_edge): Likewise.
30063 (ipa_modify_call_arguments): Likewise.
30064 (ipa_modify_expr): Likewise.
30065 (ipa_dump_param_adjustments): Likewise.
30066 (ipa_dump_agg_replacement_values): Likewise.
30067 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30068 * ipa-pure-const.c (check_stmt): Likewise.
30069 (pass_nothrow::execute): Likewise.
30070 * ipa-split.c (execute_split_functions): Likewise.
30071 * omp-offload.c (dump_oacc_loop_part): Likewise.
30072 (dump_oacc_loop): Likewise.
30073 * trans-mem.c (tm_log_emit): Likewise.
30074 (tm_memopt_accumulate_memops): Likewise.
30075 (dump_tm_memopt_set): Likewise.
30076 (dump_tm_memopt_transform): Likewise.
30077 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30078 (print_loop): Likewise.
30079 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30080 (chkp_gather_checks_info): Likewise.
30081 (chkp_get_check_result): Likewise.
30082 (chkp_remove_check_if_pass): Likewise.
30083 (chkp_use_outer_bounds_if_possible): Likewise.
30084 (chkp_reduce_bounds_lifetime): Likewise.
30085 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30086 (chkp_mark_completed_bounds): Likewise.
30087 (chkp_register_incomplete_bounds): Likewise.
30088 (chkp_mark_invalid_bounds): Likewise.
30089 (chkp_maybe_copy_and_register_bounds): Likewise.
30090 (chkp_build_returned_bound): Likewise.
30091 (chkp_get_bound_for_parm): Likewise.
30092 (chkp_build_bndldx): Likewise.
30093 (chkp_get_bounds_by_definition): Likewise.
30094 (chkp_generate_extern_var_bounds): Likewise.
30095 (chkp_get_bounds_for_decl_addr): Likewise.
30096 * tree-chrec.c (chrec_apply): Likewise.
30097 * tree-data-ref.c (dump_data_reference): Likewise.
30098 (dump_subscript): Likewise.
30099 (dump_data_dependence_relation): Likewise.
30100 (analyze_overlapping_iterations): Likewise.
30101 * tree-inline.c (expand_call_inline): Likewise.
30102 (tree_function_versioning): Likewise.
30103 * tree-into-ssa.c (dump_defs_stack): Likewise.
30104 (dump_currdefs): Likewise.
30105 (dump_names_replaced_by): Likewise.
30106 (dump_update_ssa): Likewise.
30107 (update_ssa): Likewise.
30108 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30109 * tree-parloops.c (build_new_reduction): Likewise.
30110 (try_create_reduction_list): Likewise.
30111 (ref_conflicts_with_region): Likewise.
30112 (oacc_entry_exit_ok_1): Likewise.
30113 (oacc_entry_exit_single_gang): Likewise.
30114 * tree-pretty-print.h: Likewise.
30115 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30116 (get_scalar_evolution): Likewise.
30117 (add_to_evolution): Likewise.
30118 (get_loop_exit_condition): Likewise.
30119 (analyze_evolution_in_loop): Likewise.
30120 (analyze_initial_condition): Likewise.
30121 (analyze_scalar_evolution): Likewise.
30122 (instantiate_scev): Likewise.
30123 (number_of_latch_executions): Likewise.
30124 (gather_chrec_stats): Likewise.
30125 (final_value_replacement_loop): Likewise.
30126 (scev_const_prop): Likewise.
30127 * tree-sra.c (dump_access): Likewise.
30128 (disqualify_candidate): Likewise.
30129 (create_access): Likewise.
30130 (reject): Likewise.
30131 (maybe_add_sra_candidate): Likewise.
30132 (create_access_replacement): Likewise.
30133 (analyze_access_subtree): Likewise.
30134 (analyze_all_variable_accesses): Likewise.
30135 (sra_modify_assign): Likewise.
30136 (initialize_constant_pool_replacements): Likewise.
30137 (find_param_candidates): Likewise.
30138 (decide_one_param_reduction): Likewise.
30139 (replace_removed_params_ssa_names): Likewise.
30140 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30141 * tree-ssa-copy.c (dump_copy_of): Likewise.
30142 (copy_prop_visit_cond_stmt): Likewise.
30143 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30144 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30145 (record_equivalences_from_stmt): Likewise.
30146 * tree-ssa-dse.c (compute_trims): Likewise.
30147 (delete_dead_call): Likewise.
30148 (delete_dead_assignment): Likewise.
30149 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30150 (forward_propagate_into_cond): Likewise.
30151 (pass_forwprop::execute): Likewise.
30152 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30153 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30155 (move_computations_worker): Likewise.
30156 (execute_sm): Likewise.
30157 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30158 (remove_exits_and_undefined_stmts): Likewise.
30159 (remove_redundant_iv_tests): Likewise.
30160 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30161 (adjust_iv_update_pos): Likewise.
30162 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30163 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30164 (value_replacement): Likewise.
30165 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30166 * tree-ssa-pre.c (print_pre_expr): Likewise.
30167 (get_representative_for): Likewise.
30168 (create_expression_by_pieces): Likewise.
30169 (insert_into_preds_of_block): Likewise.
30170 (eliminate_insert): Likewise.
30171 (eliminate_dom_walker::before_dom_children): Likewise.
30172 (eliminate): Likewise.
30173 (remove_dead_inserted_code): Likewise.
30174 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30175 * tree-ssa-reassoc.c (get_rank): Likewise.
30176 (eliminate_duplicate_pair): Likewise.
30177 (eliminate_plus_minus_pair): Likewise.
30178 (eliminate_not_pairs): Likewise.
30179 (undistribute_ops_list): Likewise.
30180 (eliminate_redundant_comparison): Likewise.
30181 (update_range_test): Likewise.
30182 (optimize_range_tests_var_bound): Likewise.
30183 (optimize_vec_cond_expr): Likewise.
30184 (rewrite_expr_tree): Likewise.
30185 (rewrite_expr_tree_parallel): Likewise.
30186 (linearize_expr): Likewise.
30187 (break_up_subtract): Likewise.
30188 (linearize_expr_tree): Likewise.
30189 (attempt_builtin_powi): Likewise.
30190 (attempt_builtin_copysign): Likewise.
30191 (transform_stmt_to_copy): Likewise.
30192 (transform_stmt_to_multiply): Likewise.
30193 (dump_ops_vector): Likewise.
30194 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30195 (print_scc): Likewise.
30196 (set_ssa_val_to): Likewise.
30197 (visit_reference_op_store): Likewise.
30198 (visit_use): Likewise.
30199 (sccvn_dom_walker::before_dom_children): Likewise.
30200 (run_scc_vn): Likewise.
30201 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30203 (expr_hash_elt::print): Likewise.
30204 (const_and_copies::pop_to_marker): Likewise.
30205 (const_and_copies::record_const_or_copy_raw): Likewise.
30206 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30207 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30208 (dump_predicates): Likewise.
30209 (find_uninit_use): Likewise.
30210 (warn_uninitialized_phi): Likewise.
30211 (pass_late_warn_uninitialized::execute): Likewise.
30212 * tree-ssa.c (verify_vssa): Likewise.
30213 (verify_ssa): Likewise.
30214 (maybe_optimize_var): Likewise.
30215 * tree-vrp.c (dump_value_range): Likewise.
30216 (dump_all_value_ranges): Likewise.
30217 (dump_asserts_for): Likewise.
30218 (register_edge_assert_for_2): Likewise.
30219 (vrp_visit_cond_stmt): Likewise.
30220 (vrp_visit_switch_stmt): Likewise.
30221 (vrp_visit_stmt): Likewise.
30222 (vrp_visit_phi_node): Likewise.
30223 (simplify_cond_using_ranges_1): Likewise.
30224 (fold_predicate_in): Likewise.
30225 (evrp_dom_walker::before_dom_children): Likewise.
30226 (evrp_dom_walker::push_value_range): Likewise.
30227 (evrp_dom_walker::pop_value_range): Likewise.
30228 (execute_early_vrp): Likewise.
30230 2017-05-16 Richard Biener <rguenther@suse.de>
30232 * dwarf2out.c (loc_list_from_tree_1): Do not create
30233 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30235 2017-05-16 Richard Biener <rguenther@suse.de>
30237 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30239 (note_variable_value_in_expr): If we resolved the decl ref
30240 do not push to the stack.
30242 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30244 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30245 operations in fast-math mode.
30246 (vaddq_f16): Likewise.
30247 (vmul_f16): Likewise.
30248 (vmulq_f16): Likewise.
30249 (vsub_f16): Likewise.
30250 (vsubq_f16): Likewise.
30251 * config/arm/neon.md (add<mode>3): New.
30253 (fma:<VH:mode>3): New. Also remove outdated comment.
30256 2017-05-16 Martin Liska <mliska@suse.cz>
30260 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30261 (odr_types_equivalent_p): Likewise.
30263 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30265 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30267 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30270 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30271 non-interunit SSE move alternatives with '?'.
30272 (zero-extendsidi peephole2): New peephole to skip intermediate
30273 general register in SSE zero-extend sequence.
30275 2017-05-15 Jeff Law <law@redhat.com>
30277 * reorg.c (relax_delay_slots): Create a new variable to hold
30278 the temporary target rather than clobbering TARGET_LABEL.
30280 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30281 missing argument to extract_bit_field call.
30282 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30284 2017-05-15 Martin Liska <mliska@suse.cz>
30287 * gcc.c (process_command): Do not allow empty argument of -o option.
30289 2017-05-15 Renlin Li <renlin.li@arm.com>
30291 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30292 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30293 * config/aarch64/constraints.md (Usf): Add long call check.
30294 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30295 (call_value): Likewise.
30296 (sibcall): Likewise.
30297 (sibcall_value): Likewise.
30299 (call_value_insn): New.
30300 (sibcall_insn): Update rtx pattern.
30301 (sibcall_value_insn): Likewise.
30302 (call_internal): Remove.
30303 (call_value_internal): Likewise.
30304 (sibcall_internal): Likewise.
30305 (sibcall_value_internal): Likewise.
30306 (call_reg): Likewise.
30307 (call_symbol): Likewise.
30308 (call_value_reg): Likewise.
30309 (call_value_symbol): Likewise.
30311 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30314 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30316 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30318 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30319 compatible with CCGOCmode and with CCZmode.
30321 2017-05-14 Martin Sebor <msebor@redhat.com>
30323 PR middle-end/77671
30324 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30325 (gimple_fold_builtin_snprintf): Same.
30326 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30327 (gimple_fold_builtin_snprintf): Same.
30328 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30329 of character types.
30330 (is_call_safe): New function.
30331 (try_substitute_return_value): Call it.
30332 (try_simplify_call): New function.
30333 (pass_sprintf_length::handle_gimple_call): Call it.
30335 2017-05-14 Martin Sebor <msebor@redhat.com>
30337 PR middle-end/80669
30338 * builtins.c (expand_builtin_stpncpy): Simplify.
30340 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30342 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30343 * config/i386/i386.h
30344 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30345 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30346 (struct machine_function): Add new members call_ms2sysv,
30347 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30348 (struct machine_frame_state): New fields sp_realigned and
30349 sp_realigned_offset.
30350 * config/i386/i386.c
30351 (enum xlogue_stub): New enum.
30352 (enum xlogue_stub_sets): New enum.
30353 (class xlogue_layout): New class.
30354 (struct ix86_frame): New fields stack_realign_allocate_offset,
30355 stack_realign_offset and outlined_save_offset. Modify comments to
30356 detail stack layout when using out-of-line stubs.
30357 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30358 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30359 -mcall-ms2sysv-xlogues.
30360 (stub_managed_regs): New static variable.
30361 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30362 registers managed by out-of-line stub.
30363 (disable_call_ms2sysv_xlogues): New function.
30364 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30365 m->call_ms2sysv when appropriate and compute frame layout for
30367 (sp_valid_at, fp_valid_at): New inline functions.
30368 (choose_basereg): New function.
30369 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30371 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30372 Use align parameter of choose_baseaddr to generated aligned SSE movs
30374 (pro_epilogue_adjust_stack): Modify to track
30375 machine_frame_state::sp_realigned.
30376 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30377 (ix86_nsaved_sseregs): Likewise.
30378 (ix86_emit_save_regs): Likewise.
30379 (ix86_emit_save_regs_using_mov): Likewise.
30380 (ix86_emit_save_sse_regs_using_mov): Likewise.
30381 (get_scratch_register_on_entry): Likewise.
30382 (gen_frame_set): New function.
30383 (gen_frame_load): Likewise.
30384 (gen_frame_store): Likewise.
30385 (emit_outlined_ms2sysv_save): Likewise.
30386 (emit_outlined_ms2sysv_restore): Likewise.
30387 (ix86_expand_prologue): Modify stack re-alignment code and call
30388 emit_outlined_ms2sysv_save when appropriate.
30389 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30390 parameter rtx_insn *insn, which allows the function to be used to only
30391 generate the notes.
30392 (ix86_expand_epilogue): Modify validity checks of frame and stack
30393 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30394 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30395 * config/i386/predicates.md
30396 (save_multiple): New predicate.
30397 (restore_multiple): Likewise.
30398 * config/i386/sse.md
30399 (save_multiple<mode>): New pattern.
30400 (save_multiple_realign<mode>): Likewise.
30401 (restore_multiple<mode>): Likewise.
30402 (restore_multiple_and_return<mode>): Likewise.
30403 (restore_multiple_leave_return<mode>): Likewise.
30404 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30406 2017-05-14 Julia Koval <julia.koval@intel.com>
30408 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30409 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30410 (__builtin_ia32_xsetbv): New builtins.
30411 * config/i386/i386.c (ix86_expand_special_args_builtin):
30413 (ix86_expand_builtin): Special expand for new intrinsics.
30414 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30415 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30416 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30418 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30420 * cfganal.c (inverted_post_order_compute): Change argument type
30422 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30423 * df-core.c (rest_of_handle_df_initialize): Adjust.
30424 (rest_of_handle_df_finish): Likewise.
30425 (df_analyze_1): Likewise.
30426 (df_analyze): Likewise.
30427 (loop_inverted_post_order_compute): Change argument to be a vec *.
30428 (df_analyze_loop): Adjust.
30429 (df_get_n_blocks): Likewise.
30430 (df_get_postorder): Likewise.
30431 * df.h (struct df_d): Change field to be a vec.
30432 * lcm.c (compute_laterin): Adjust.
30433 (compute_available): Likewise.
30434 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30435 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30436 * tree-ssa-pre.c (compute_antic): Likewise.
30438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30440 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30441 (depth_first_search::depth_first_search): Change structure init
30442 function to this constructor.
30443 (depth_first_search::add_bb): Rename function to this member.
30444 (depth_first_search::execute): Likewise.
30445 (flow_dfs_compute_reverse_finish): Adjust.
30447 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30449 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30450 (longest_simple_path): Likewise.
30451 * shrink-wrap.c (spread_components): Likewise.
30452 (disqualify_problematic_components): Likewise.
30453 (emit_common_heads_for_components): Likewise.
30454 (emit_common_tails_for_components): Likewise.
30455 (insert_prologue_epilogue_for_components): Likewise.
30457 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30459 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30462 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30464 * df-core.c (df_set_blocks): Start using auto_bitmap.
30465 (df_compact_blocks): Likewise.
30466 * df-problems.c (df_rd_confluence_n): Likewise.
30467 * df-scan.c (df_insn_rescan_all): Likewise.
30468 (df_process_deferred_rescans): Likewise.
30469 (df_update_entry_block_defs): Likewise.
30470 (df_update_exit_block_uses): Likewise.
30471 (df_entry_block_bitmap_verify): Likewise.
30472 (df_exit_block_bitmap_verify): Likewise.
30473 (df_scan_verify): Likewise.
30474 * lra-constraints.c (lra_constraints): Likewise.
30475 (undo_optional_reloads): Likewise.
30476 (lra_undo_inheritance): Likewise.
30477 * lra-remat.c (calculate_gen_cands): Likewise.
30478 (do_remat): Likewise.
30479 * lra-spills.c (assign_spill_hard_regs): Likewise.
30480 (spill_pseudos): Likewise.
30481 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30482 (bitmap_set_subtract_values): Likewise.
30484 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30486 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30487 management with auto_bitmap.
30488 (fix_inter_tick): Likewise.
30489 (fix_recovery_deps): Likewise.
30490 * ira.c (add_store_equivs): Likewise.
30491 (find_moveable_pseudos): Likewise.
30492 (split_live_ranges_for_shrink_wrap): Likewise.
30493 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30494 (rtx_reuse_manager::seen_def_p): Likewise.
30495 (rtx_reuse_manager::set_seen_def): Likewise.
30496 * print-rtl.h (class rtx_reuse_manager): Likewise.
30498 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30500 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30502 (migrate_btr_def): Likewise.
30503 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30504 * df-core.c (loop_post_order_compute): Likewise.
30505 (loop_inverted_post_order_compute): Likewise.
30506 * hsa-common.h: Likewise.
30507 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30508 * init-regs.c (initialize_uninitialized_regs): Likewise.
30509 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30510 (inline_small_functions): Likewise.
30511 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30512 * ira.c (combine_and_move_insns): Likewise.
30513 (build_insn_chain): Likewise.
30514 * loop-invariant.c (find_invariants): Likewise.
30515 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30516 * predict.c (tree_predict_by_opcode): Likewise.
30517 (predict_paths_leading_to): Likewise.
30518 (predict_paths_leading_to_edge): Likewise.
30519 (estimate_loops_at_level): Likewise.
30520 (estimate_loops): Likewise.
30521 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30522 (spread_components): Likewise.
30523 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30524 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30525 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30526 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30527 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30528 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30529 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30530 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30531 (mark_threaded_blocks): Likewise.
30532 (thread_through_all_blocks): Likewise.
30533 * tree-ssa.c (verify_ssa): Likewise.
30534 (execute_update_addresses_taken): Likewise.
30535 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30537 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30539 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30541 (post_order_compute): Likewise.
30542 (inverted_post_order_compute): Likewise.
30543 (pre_and_rev_post_order_compute_fn): Likewise.
30545 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30547 * genrecog.c (int_set::int_set): Explicitly construct our
30548 auto_vec base class.
30549 * vec.h (auto_vec::auto_vec): New constructor.
30551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30553 * bitmap.h (class auto_bitmap): New constructor taking
30554 bitmap_obstack * argument.
30556 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30558 * bitmap.h (class auto_bitmap): Change type of m_bits to
30559 bitmap_head, and adjust ctor / dtor and member operators.
30561 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30563 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30564 when returned register mode doesn't match original mode.
30566 2017-05-12 Jeff Law <law@redhat.com>
30567 Jakub Jelinek <jakub@redhat.com>
30569 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30570 we look for cc setter after the compare-elim changes.
30571 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30572 within the vector to match what compare-elim now expects.
30573 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30574 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30576 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30577 after the compare-elim changes.
30578 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30579 the vector to match what compare-elim now expects.
30580 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30581 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30582 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30583 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30584 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30586 * config/visium/visium.c (single_set_and_flags): Fix where
30587 we look for cc setter after the compare-elim changes.
30588 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30589 with the vector to match what compare-elim now expects.
30590 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30591 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30592 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30593 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30594 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30595 (neg<mode>2_insn_set_overflow): Likewise.
30597 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30599 PR middle-end/79794
30600 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30601 maybe_expand_insn call, set ops[0].target. If still set after call,
30602 set alt_rtl. Add extra arg to recursive calls.
30603 (extract_bit_field): Add alt_rtl argument. Pass to
30605 * expmed.h (extract_bit_field): Fix prototype.
30606 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30607 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30608 to extract_bit_field_calls.
30609 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30610 Pass alt_rtl to extract_bit_field calls.
30611 * calls.c (store_unaligned_arguments_into_psuedos)
30612 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30613 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30615 * optabs.h (struct expand_operand): Add target bitfield.
30617 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30619 * compare-elim.c (try_eliminate_compare): Canonicalize
30620 operation with embedded compare to
30621 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30622 (set (reg) (operation)].
30624 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30626 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30629 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30630 cost of adding a carry flag for ADC instruction.
30631 [case MINUS]: Ignore the cost of subtracting a carry flag
30632 for SBB instruction.
30634 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30636 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30638 * config/rs6000/bmiintrin.h: New file.
30639 * config/rs6000/bmi2intrin.h: New file.
30640 * config/rs6000/x86intrin.h: New file.
30642 2017-05-12 Jeff Law <law@redhat.com>
30644 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30647 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30649 PR middle-end/80707
30650 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30652 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30654 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30655 early expansion of vector divide builtins.
30656 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30657 builtins identified as having unsigned arguments.
30659 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30661 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30662 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30663 expansion of vector logical operations (and, andc, or, xor,
30666 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30668 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30669 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30671 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30674 early GIMPLE expansion of vector multiplies.
30676 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30678 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30679 TARGET_HAVE_MOVT conditional.
30680 (movt splitter): Likewise.
30682 2017-05-12 Richard Biener <rguenther@suse.de>
30684 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30685 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30686 Fold all stmts not inplace.
30688 2017-05-12 Richard Biener <rguenther@suse.de>
30690 PR tree-optimization/80713
30691 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30692 inserted_exprs bit for not removed stmts.
30694 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30696 PR middle-end/69921
30697 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30698 parallelized" attribute for parallelized OpenACC kernels.
30699 * omp-offload.c (execute_oacc_device_lower): Use it.
30701 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30702 Set "oacc kernels" attribute.
30703 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30704 parameter. Adjust all users.
30705 (oacc_fn_attrib_kernels_p): Remove function.
30706 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30707 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30708 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30709 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30710 assert "oacc kernels" attribute is set.
30712 2017-05-11 Carl Love <cel@us.ibm.com>
30714 * config/rs6000/rs6000-c: Add support for built-in functions
30715 vector unsigned char vec_popcnt (vector signed char)
30716 vector unsigned char vec_popcnt (vector unsigned char)
30717 vector unsigned short vec_popcnt (vector signed short)
30718 vector unsigned short vec_popcnt (vector unsigned short)
30719 vector unsigned int vec_popcnt (vector signed int)
30720 vector unsigned int vec_popcnt (vector unsigned int)
30721 vector unsigned long long vec_popcnt (vector signed long long)
30722 vector unsigned long long vec_popcnt (vector unsigned long long)
30723 vector signed long long vec_slo (vector signed long long,
30724 vector signed char)
30725 vector signed long long vec_slo (vector signed long long,
30726 vector unsigned char)
30727 vector unsigned long long vec_slo (vector unsigned long long,
30728 vector signed char)
30729 vector unsigned long long vec_slo (vector unsigned long long,
30730 vector unsigned char)
30731 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30732 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30733 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30734 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30735 * doc/extend.texi: Update the built-in documentation file for the
30736 new built-in functions.
30738 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30740 * attribs.h (sorted_attr_string): Move machine independent
30741 functions for target clone support from the i386 port to common
30742 code. Rename ix86_function_versions to common_function_versions.
30743 Rename make_name to make_unique_name.
30744 (common_function_versions): Likewise.
30745 (make_unique_name): Likewise.
30746 (make_dispatcher_decl): Likewise.
30747 (is_function_default_version): Likewise.
30748 * attribs.c (attr_strcmp): Likewise.
30749 (sorted_attr_string): Likewise.
30750 (common_function_versions): Likewise.
30751 (make_unique_name): Likewise.
30752 (make_dispatcher_decl): Likewise.
30753 (is_function_default_version): Likewise.
30754 * config/i386/i386.c (attr_strcmp): Likewise.
30755 (sorted_attr_string): Likewise.
30756 (ix86_function_versions): Likewise.
30757 (make_name): Likewise.
30758 (make_dispatcher_decl): Likewise.
30759 (is_function_default_version): Likewise.
30760 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30762 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30765 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30766 Account for direct move costs for vec_construct of integer
30769 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30772 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30773 (UNSPEC_STX_ATOMIC): Ditto.
30774 (loaddi_via_sse): New insn.
30775 (storedi_via_sse): Ditto.
30776 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30777 Update corresponding peephole2 patterns.
30778 (atomic_storedi_fpu): Ditto.
30780 2017-05-11 Julia Koval <julia.koval@intel.com>
30782 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30783 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30785 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30786 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30787 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30789 2017-05-11 Nathan Sidwell <nathan@acm.org>
30791 * graphite-poly.c: Include dumpfile.h.
30793 * dumpfle.h (dump_function): Declare here ...
30794 * tree-dump.h (dump_function): ... not here.
30795 * dumpfile.c: #include tree-cfg.h.
30796 (dump_function): Move here from ...
30797 * tree-dump.c (dump_function): ... here.
30798 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30799 * graphite-poly.c: Don't include tree-dump.h.
30800 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30801 * print-tree.c: Likewise.
30802 * stor-layout.c: Likewise.
30803 * tree-nested.c: Likewise.
30805 * dumpfile.c (dump_start): Use TDF_FLAGS.
30806 (dump_enable_all): Fix TDF_KIND check thinko.
30808 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30810 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30811 array entries to represent two legal parameterizations of the
30812 overloaded __builtin_cmpb function, as represented by the
30813 P6_OV_BUILTIN_CMPB constant.
30814 (altivec_resolve_overloaded_builtin): Add special case handling
30815 for the __builtin_cmpb function, as represented by the
30816 P6_OV_BUILTIN_CMPB constant.
30817 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30818 (BU_P6_64BIT_2): New macro.
30819 (BU_P6_OVERLOAD_2): New macro
30820 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30821 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30822 (CMPB): Add overload support to represent both 32-bit and 64-bit
30823 compare-bytes function.
30824 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30825 support for TARGET_CMPB.
30826 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30827 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30828 documentation of the __builtin_cmpb overloaded built-in function.
30830 2017-05-11 Richard Biener <rguenther@suse.de>
30832 PR tree-optimization/80705
30833 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30834 bases are not vectorizable.
30836 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30838 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30839 when counting register pressure.
30841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30843 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30844 (iv_ca_more_deps): Renamed to ...
30845 (iv_ca_compare_deps): ... this.
30846 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30850 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30852 (determine_group_iv_costs): ... here.
30853 (find_inv_vars_cb): Record inv var if it's not recorded before.
30855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30857 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30858 (get_shiftadd_cost): Ditto.
30860 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30862 * tree-ssa-address.c: Include header file.
30863 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30865 (add_to_parts): Refactor.
30866 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30867 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30870 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30872 PR tree-optimization/53090
30873 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30875 (extract_cond_operands): Detect condition with IV on both sides
30876 and return COMP_IV_EXPR_2.
30877 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30878 (rewrite_use_compare): Simplify by removing call to function
30879 extract_cond_operands.
30881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30883 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30884 (extract_cond_operands): Detect condition comparing against non-
30885 invariant bound and return appropriate enum value.
30886 (find_interesting_uses_cond): Update use of extract_cond_operands.
30887 Handle its return value accordingly.
30888 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30892 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30893 nonlinear iv_use computation in loop invariant sensitive way.
30895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30897 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30898 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30902 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30903 (dump_cand): Support iv_cand.inv_exprs.
30904 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30906 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30909 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30911 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30913 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30914 as local function. Include necessary header files.
30915 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30917 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30919 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30921 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30923 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30924 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30925 RSHIFT_EXPR and BIT_NOT_EXPR.
30927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30929 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30930 (adjust_setup_cost): New parameter supporting round up adjustment.
30931 (struct address_cost_data): Delete.
30932 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30933 (split_address_cost, ptr_difference_cost): Delete.
30934 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30935 (struct ainc_cost_data): New struct.
30936 (get_address_cost_ainc): New function.
30937 (get_address_cost, get_computation_cost): Reimplement.
30938 (determine_group_iv_cost_address): Record inv_expr for all uses of
30940 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30941 (iv_ca_has_deps): Reimplemented to ...
30942 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30944 (iv_ca_extend): Call iv_ca_more_deps.
30946 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30948 * tree-ssa-address.c (struct mem_address): Move to header file.
30949 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30950 * tree-ssa-address.h (struct mem_address): Move from C file.
30951 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30955 * tree-affine.h (aff_combination_type): New interface.
30956 (aff_combination_zero_p): Remove static.
30957 (aff_combination_const_p): New interface.
30958 (aff_combination_singleton_var_p): New interfaces.
30960 2017-05-11 Richard Biener <rguenther@suse.de>
30962 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30963 Skip unreachable blocks and destinations.
30964 (eliminate): Move stmt removal and fixup ...
30965 (fini_eliminate): ... here. Skip inserted exprs.
30966 (pass_pre::execute): Move fini_pre after fini_eliminate.
30967 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30968 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30969 PRE to get rid of dead code that has invalid SSA form and
30970 split critical edges again.
30972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30974 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30976 2017-05-11 Richard Biener <rguenther@suse.de>
30978 * passes.c (execute_function_todo): Verify loops if they are
30979 said to be up-to-date.
30980 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30981 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30983 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30986 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30987 handle calling assemble_external ourself.
30990 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30991 modes with zero size. Enhance comment.
30993 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30995 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30996 built-ins for vec_xl and vec_xst with short and char pointer
30999 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31001 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31002 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31003 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31004 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31005 (_mm_maskz_min_round_ss): New intrinsics.
31006 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31007 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31008 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31009 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31010 (__builtin_ia32_minss_mask_round): New builtins.
31011 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31012 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31013 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31015 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31016 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31018 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31021 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31023 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31024 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31025 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31026 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31027 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31028 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31029 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31030 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31031 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31032 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31033 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31034 (__builtin_ia32_mulss_mask_round): New builtins.
31035 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31036 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31037 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31039 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31040 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31042 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31045 2017-05-10 Julia Koval <julia.koval@intel.com>
31047 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31048 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31049 (_mm256_setr_m128i): New intrinsics.
31051 2017-05-10 Julia Koval <julia.koval@intel.com>
31053 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31054 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31055 (_mm_maskz_rcp14_ss): New intrinsics.
31056 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31057 (__builtin_ia32_rcp14ss_mask): New builtins.
31058 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31060 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31062 PR tree-optimization/51513
31063 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31064 (assert_unreachable_fallthru_edge_p): Use it.
31065 (group_case_labels_stmt): Likewise.
31066 * tree-cfg.h: Prototype it.
31067 * stmt.c: Include cfghooks.h and tree-cfg.h.
31068 (emit_case_dispatch_table) <gap_label>: New local variable.
31069 Use it to fill dispatch table gaps.
31070 Test for default_label before updating probabilities.
31071 (expand_case) <default_label>: Remove unneeded initialization.
31072 Test for unreachable default case statement and remove its edge.
31073 Set default_label accordingly.
31074 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31076 2017-05-10 Carl Love <cel@us.ibm.com>
31078 * config/rs6000/rs6000-c: Add support for built-in functions
31079 vector signed char vec_neg (vector signed char)
31080 vector signed short int vec_neg (vector short int)
31081 vector signed int vec_neg (vector signed int)
31082 vector signed long long vec_neg (vector signed long long)
31083 vector float vec_neg (vector float)
31084 vector double vec_neg (vector double)
31085 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31087 * config/rs6000/altivec.h: Add define for vec_neg
31088 * doc/extend.texi: Update the built-in documentation for the
31089 new built-in functions.
31091 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31093 PR tree-optimization/77644
31094 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31096 2017-05-10 Nathan Sidwell <nathan@acm.org>
31098 * dumpfile.h (TDI_lang_all): New.
31099 (TDF_KIND): New. Renumber others
31100 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31102 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31104 (get_dump_file_name): Adjust suffix generation.
31105 (dump_enable_all): Use TDF_KIND.
31106 * doc/invoke.texi (-fdump-lang-all): Document.
31108 * dumpfile.h: Tabify.
31110 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31113 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31114 Move member access before delete.
31116 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31118 * tree-inline.c (expand_call_inline): Split block at stmt
31121 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31124 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31125 are now unused after splitting mov{sf,sd}_hardfloat.
31126 (f32_lr2): Likewise.
31127 (f32_lm): Likewise.
31128 (f32_lm2): Likewise.
31129 (f32_li): Likewise.
31130 (f32_li2): Likewise.
31131 (f32_lv): Likewise.
31132 (f32_sr): Likewise.
31133 (f32_sr2): Likewise.
31134 (f32_sm): Likewise.
31135 (f32_sm2): Likewise.
31136 (f32_si): Likewise.
31137 (f32_si2): Likewise.
31138 (f32_sv): Likewise.
31139 (f32_dm): Likewise.
31140 (f32_vsx): Likewise.
31141 (f32_av): Likewise.
31142 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31143 For movsf, order stores so the VSX stores occur before the GPR
31144 store which encourages the register allocator to use a traditional
31145 FPR instead of a GPR. For movsd, order the stores so that the GPR
31146 store comes before the VSX stores to allow the power6 to work.
31147 This is due to the power6 not having a 32-bit integer store
31148 instruction from a FPR.
31149 (movsf_hardfloat): Likewise.
31150 (movsd_hardfloat): Likewise.
31152 2017-05-09 Martin Sebor <msebor@redhat.com>
31154 PR translation/80280
31155 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31158 PR translation/80280
31159 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31160 data member added in r247778.
31161 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31163 2017-05-09 Nathan Sidwell <nathan@acm.org>
31165 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31167 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31170 2017-05-09 Marek Polacek <polacek@redhat.com>
31172 * doc/invoke.texi: Fix typo.
31174 2017-05-09 Richard Biener <rguenther@suse.de>
31176 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31177 (vrp_val_is_min): Likewise.
31178 (set_value_range_to_value): Likewise.
31179 (set_value_range_to_nonnegative): Likewise.
31180 (gimple_assign_nonzero_p): Likewise.
31181 (gimple_stmt_nonzero_p): Likewise.
31182 (vrp_int_const_binop): Likewise. Remove unreachable case.
31183 (adjust_range_with_scev): Adjust comments.
31184 (compare_range_with_value): Likewise.
31185 (extract_range_from_phi_node): Likewise.
31186 (test_for_singularity): Likewise.
31188 2017-05-09 Richard Biener <rguenther@suse.de>
31190 * tree-vrp.c (get_single_symbol): Add assert that we don't
31191 get overflowed constants as invariant part.
31192 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31193 checks. Use wi::cmp instead of recursing for integer constants.
31194 (compare_values): Just ignore whether we assumed undefined
31195 overflow instead of failing the compare.
31196 (extract_range_for_var_from_comparison_expr): Add comment before the
31197 TREE_NO_WARNING sets.
31198 (test_for_singularity): Likewise.
31199 (extract_range_from_comparison): Do not disable optimization
31200 when we assumed undefined overflow.
31201 (extract_range_basic): Remove init of unused var.
31203 2017-05-09 Richard Biener <rguenther@suse.de>
31205 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31206 (extract_range_from_multiplicative_op_1): Adjust.
31207 (extract_range_from_binary_expr_1): Use int_const_binop.
31209 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31212 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31213 rs6000_store_data_bypass_p in seven define_bypass directives and
31214 in several comments.
31215 * config/rs6000/rs6000-protos.h: Add prototype for
31216 rs6000_store_data_bypass_p function.
31217 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31218 function implements slightly different (rs6000-specific) semantics
31219 than store_data_bypass_p, returning false rather than aborting
31220 with assertion error when arguments do not satisfy the
31221 requirements of store data bypass.
31222 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31223 rs6000_store_data_bypass_p.
31225 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31227 * config/xtensa/xtensa-protos.h
31228 (xtensa_initial_elimination_offset): New declaration.
31229 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31230 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31231 macro definition, add case for FRAME_POINTER_REGNUM when
31232 FRAME_GROWS_DOWNWARD.
31233 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31234 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31235 xtensa_initial_elimination_offset.
31237 2017-05-08 Nathan Sidwell <nathan@acm.org>
31239 * doc/invoke.texi: Alphabetize -fdump options.
31241 2017-05-08 Martin Sebor <msebor@redhat.com>
31243 PR translation/80280
31244 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31246 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31248 * target.def (compute_frame_layout): New optional target hook.
31249 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31250 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31251 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31253 * reload1.c (verify_initial_elim_offsets): Likewise.
31254 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31255 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31256 (arm_get_frame_offsets): Split up into this ...
31257 (arm_compute_frame_layout): ... and this function.
31259 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31261 * config/aarch64/constraints.md (Usa): New constraint.
31262 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31264 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31266 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31267 with_multilib_list after it has been checked.
31269 2017-05-08 Richard Biener <rguenther@suse.de>
31271 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31272 (bitmap_set_subtract_values): Likewise.
31274 2017-05-08 Richard Biener <rguenther@suse.de>
31276 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31277 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31279 (gimple_stmt_nonzero_warnv_p): Rename to ...
31280 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31282 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31283 (extract_range_basic): Adjust, do not disable propagation on
31284 strict overflow sensitive simplification.
31285 (vrp_visit_cond_stmt): Likewise.
31287 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31289 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31290 body size unconditionally.
31292 2017-05-07 Jeff Law <law@redhat.com>
31295 2017-05-06 Jeff Law <law@redhat.com>
31296 PR tree-optimization/78496
31297 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31300 PR tree-optimization/78496
31301 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31302 (simplify_stmt_using_ranges): Call it.
31303 (vrp_dom_walker::before_dom_children): Extract equivalences
31304 from an ASSERT_EXPR with an equality comparison against a
31307 2017-05-06 Jeff Law <law@redhat.com>
31309 PR tree-optimization/78496
31310 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31313 PR tree-optimization/78496
31314 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31315 (simplify_stmt_using_ranges): Call it.
31316 (vrp_dom_walker::before_dom_children): Extract equivalences
31317 from an ASSERT_EXPR with an equality comparison against a
31320 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31322 * lra-constraints.c (lra_copy_reg_equiv): New function.
31323 (split_reg): Use it to copy equivalence information from the
31324 original register to the spill register.
31326 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31328 PR rtl-optimization/75964
31329 * simplify-rtx.c (simplify_const_relational_operation): Remove
31330 invalid handling of comparisons of integer ABS.
31332 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31334 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31335 initialize to zero.
31336 (init_regs): Remove declaration.
31337 (function_arg_advance_32): Initialize error_p as boolean variable.
31339 2017-05-05 Nathan Sidwell <nathan@acm.org>
31341 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31342 lines. Use for (;;).
31344 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31346 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31347 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31348 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31349 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31350 VF=2 that require versioning.
31352 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31354 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31357 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31359 * diagnostic.h (diagnostic_override_option_index): Convert from
31360 macro to inline function.
31362 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31364 * diagnostic.c (last_module_changed_p): New function.
31365 (set_last_module): New function.
31366 (diagnostic_report_current_module): Convert macro usage to
31367 the above functions.
31368 * diagnostic.h (diagnostic_context::last_module): Strengthen
31369 from const line_map * to const line_map_ordinary *.
31370 (diagnostic_last_module_changed): Delete macro.
31371 (diagnostic_set_last_module): Delete macro.
31373 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31375 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31376 with diagnostic_report_diagnostic.
31377 (diagnostic_n_impl_richloc): Likewise.
31378 * diagnostic.h (report_diagnostic): Delete macro.
31379 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31380 with diagnostic_report_diagnostic.
31381 * substring-locations.c (format_warning_va): Likewise.
31383 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31385 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31386 save/restor of format_spec. Move option-printing code to...
31387 (print_option_information): ...this new function, and
31388 reimplement by simply printing to the pretty_printer,
31389 rather than appending to the format string.
31391 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31393 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31394 handling logic into...
31395 (update_effective_level_from_pragmas): ...this new function.
31397 2017-05-04 Andrew Waterman <andrew@sifive.com>
31399 * config/riscv/riscv.opt (mstrict-align): New option.
31400 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31401 (SLOW_UNALIGNED_ACCESS): Define.
31402 (riscv_slow_unaligned_access): Declare.
31403 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31405 (riscv_slow_unaligned_access): New variable.
31406 (rocket_tune_info): Set slow_unaligned_access to true.
31407 (optimize_size_tune_info): Set slow_unaligned_access to false.
31408 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31409 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31410 (riscv_option_override): Set riscv_slow_unaligned_access.
31411 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31413 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31415 * config/riscv/riscv.md: Unify indentation.
31417 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31422 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31424 (extendsi<mode>2): Add support for doing sign extension via
31425 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31426 don't have ISA 3.0 instructions.
31427 (extendsi<mode>2 splitter): Likewise.
31428 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31429 generate the normal insns since SImode can now go in vector
31430 registers. Disallow the special UNSPECs needed for previous
31431 machines to hide SImode being used. Add new insns
31432 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31433 (fix_trunc<mode>si2_stfiwx): Likewise.
31434 (fix_trunc<mode>si2_internal): Likewise.
31435 (fixuns_trunc<mode>si2): Likewise.
31436 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31437 (fctiw<u>z_<mode>_smallint): Likewise.
31438 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31439 of floating point to 32-bit integer from doing a direct move to
31440 the GPR registers to do a store.
31441 (fctiwz_<mode>): Break long line.
31443 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31445 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31446 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31447 (addr_list, addr_offset_valid_p): New.
31448 (split_address_groups): Check offset validity with above function.
31449 (gt-tree-ssa-loop-ivopts.h): Include header file.
31451 2017-05-05 Nathan Sidwell <nathan@acm.org>
31453 * config.gcc (arm*-*-*): Add missing 'fi'.
31455 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31457 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31458 included in -fopt-info does not matter.
31459 * doc/optinfo.texi (-fopt-info): Fix description of default
31460 behavour. Explicitly say order of options included in -fopt-info
31463 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31465 * config.gcc: Allow combinations of aprofile and rmprofile values for
31466 --with-multilib-list.
31467 * config/arm/t-multilib: New file.
31468 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31469 variables. Remove setting of ISA and floating-point ABI in
31470 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31471 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31472 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31473 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31475 * config/arm/t-rmprofile: Likewise except for the matches changes.
31476 * doc/install.texi (--with-multilib-list): Document the combination of
31477 aprofile and rmprofile values and warn about pitfalls in doing that.
31479 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31481 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31482 (movdi_aarch64): Likewise.
31484 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31486 PR tree-optimization/80632
31487 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31489 (build_arrays): Initialize it for virtual phis.
31490 (fix_phi_nodes): Use it for virtual phis.
31492 PR tree-optimization/80558
31493 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31494 [x, y] op z into [x op, y op z] for op & or | if conditions
31497 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31498 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31501 * config/arm/arm.md (use_literal_pool): Remove.
31502 (64-bit immediate split): No longer takes cost into consideration
31503 if arm_disable_literal_pool is enabled.
31504 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31505 used when arm_disable_literal_pool is enabled.
31506 (arm_max_const_double_inline_cost): Remove use of
31507 arm_disable_literal_pool.
31508 (push_minipool_fix): Add assert.
31509 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31510 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31511 (no_literal_pool_sf_immediate): New.
31513 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31515 PR tree-optimization/80613
31516 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31517 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31519 2017-05-05 Richard Biener <rguenther@suse.de>
31521 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31523 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31525 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31526 of this flag from insn conditions due to removal from r247495.
31528 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31530 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31532 (arm_early_store_addr_dep_ptr): Likewise.
31533 * config/arm/aarch-common-protos.h
31534 (arm_early_load_addr_dep_ptr): Add prototype.
31535 (arm_early_store_addr_dep_ptr): Likewise.
31536 * config/arm/cortex-a53.md: Add new bypasses.
31538 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31540 * tree.c (next_type_uid): Change type to unsigned.
31541 (type_hash_canon): Decrement back next_type_uid if
31542 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31543 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31546 2017-05-04 Martin Sebor <msebor@redhat.com>
31548 * builtins.c: Fix a trivial typo in a comment.
31550 PR middle-end/79234
31551 * builtins.c (check_sizes): Adjust to handle reading past the end.
31552 Avoid printing excessive upper bound of ranges. Use %E to print
31553 tree nodes instead of converting them to %wu.
31554 (expand_builtin_memchr): New function.
31555 (compute_dest_size): Rename...
31556 (compute_objsize): ...to this.
31557 (expand_builtin_memcpy): Adjust.
31558 (expand_builtin_mempcpy): Adjust.
31559 (expand_builtin_strcat): Adjust.
31560 (expand_builtin_strcpy): Adjust.
31561 (check_strncat_sizes): Adjust.
31562 (expand_builtin_strncat): Adjust.
31563 (expand_builtin_strncpy): Adjust and simplify.
31564 (expand_builtin_memset): Adjust.
31565 (expand_builtin_bzero): Adjust.
31566 (expand_builtin_memcmp): Adjust.
31567 (expand_builtin): Handle memcmp.
31568 (maybe_emit_chk_warning): Check strncat just once.
31570 2017-05-04 Martin Sebor <msebor@redhat.com>
31572 PR preprocessor/79214
31573 PR middle-end/79222
31574 PR middle-end/79223
31575 * builtins.c (check_sizes): Add inlining context and issue
31576 warnings even when -Wno-system-headers is set.
31577 (check_strncat_sizes): Same.
31578 (expand_builtin_strncat): Same.
31579 (expand_builtin_memmove): New function.
31580 (expand_builtin_stpncpy): Same.
31581 (expand_builtin): Handle memmove and stpncpy.
31583 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31585 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31586 which is not used any more.
31588 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31590 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31592 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31594 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31595 (cortexa53_tunings): Likewise.
31596 (cortexa57_tunings): Likewise.
31597 (cortexa72_tunings): Likewise.
31598 (cortexa73_tunings): Likewise.
31600 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31602 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31603 Set loop alignment to 8.
31605 2017-05-04 Martin Sebor <msebor@redhat.com>
31607 PR translation/80280
31608 * builtins.c (expand_builtin_object_size): Add missing quoting to
31609 %D and like directives.
31610 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31611 (hsa_type_for_tree_type): Same.
31612 (verify_function_arguments): Same.
31613 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31614 * varasm.c (get_section): Same.
31617 2017-05-04 Martin Sebor <msebor@redhat.com>
31619 PR translation/80280
31620 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31622 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31624 * config/aarch64/aarch64.c (generic_addrcost_table):
31625 Change HI/TI mode setting.
31627 2017-05-04 Martin Jambor <mjambor@suse.cz>
31629 PR tree-optimization/80622
31630 * tree-sra.c (comes_initialized_p): New function.
31631 (build_accesses_from_assign): Only set write lazily when
31632 comes_initialized_p is false.
31633 (analyze_access_subtree): Use comes_initialized_p.
31634 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31635 instead of testing for PARM_DECL.
31637 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31639 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31640 constraint on operand 0 to allow more general addressing modes.
31641 Adjust output template.
31642 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31644 * config/aarch64/aarch64-protos.h
31645 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31646 * config/aarch64/constraints.md (Dp): New address constraint.
31647 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31650 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31652 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31653 update use of estimate_ipcp_clone_size_and_time.
31654 (estimate_local_effects): Update use of
31655 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31656 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31657 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31658 Return nonspecialized time.
31660 2017-05-04 Richard Biener <rguenther@suse.de>
31662 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31663 for the last VUSE which def dominates the PHI. Directly call
31665 (get_continuation_for_phi_1): Remove.
31667 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31669 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31670 to explain the use of truncating division. Cap the number of
31671 iterations to the maximum given by nb_iterations_upper_bound,
31674 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31676 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31677 * configure: Regenerate.
31678 * config.in: Regenerate.
31679 * config/i386/driver-mingw32.c: new file.
31680 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31681 * config.host: Link driver-mingw32.o on MinGW host.
31682 * doc/install.texi: Document new --enable-mingw-wildcard configure
31685 2017-05-04 Marek Polacek <polacek@redhat.com>
31687 PR tree-optimization/80612
31688 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31690 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31691 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31693 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31694 (movt splitter): Likewise.
31695 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31696 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31697 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31698 block for Thumb-1 with MOVT.
31699 (thumb2_legitimate_address_p): Move code block ...
31700 (can_avoid_literal_pool_for_label_p): ... into this new function.
31701 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31703 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31704 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31705 "M-profile targets with the MOVT instruction".
31707 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31709 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31710 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31711 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31713 2017-05-04 Martin Liska <mliska@suse.cz>
31715 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31716 variable cond_code.
31718 2017-05-04 Richard Biener <rguenther@suse.de>
31720 * tree.c (array_at_struct_end_p): Handle arrays at struct
31721 end with flexarrays more conservatively. Refactor and treat
31722 arrays of arrays or aggregates more strict. Fix
31723 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31724 * tree.c (array_at_struct_end_p): Adjust prototype.
31725 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31726 * gimple-fold.c (get_range_strlen): Likewise.
31727 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31729 2017-05-04 Richard Biener <rguenther@suse.de>
31731 PR tree-optimization/31130
31732 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31734 (supports_overflow_infinity): Likewise.
31735 (is_negative_overflow_infinity): Likewise.
31736 (is_positive_overflow_infinity): Likewise.
31737 (is_overflow_infinity): Likewise.
31738 (stmt_overflow_infinity): Likewise.
31739 (overflow_infinity_range_p): Likewise.
31740 (usable_range_p): Remove as always returning true.
31741 (make_overflow_infinity): Remove.
31742 (negative_overflow_infinity): Likewise.
31743 (positive_overflow_infinity): Likewise.
31744 (avoid_overflow_infinity): Likewise.
31745 (set_value_range): Adjust accordingly.
31746 (set_value_range_to_nonnegative): Likewise, remove now unused
31747 overflow_infinity arg.
31748 (vrp_operand_equal_p): Adjust.
31749 (update_value_range): Likewise.
31750 (range_int_cst_singleton_p): Likewise.
31751 (operand_less_p): Likewise.
31752 (compare_values_warnv): Likewise.
31753 (extract_range_for_var_from_comparison_expr): Likewise.
31754 (vrp_int_const_binop): Likewise.
31755 (zero_nonzero_bits_from_vr): Likewise.
31756 (extract_range_from_multiplicative_op_1): Likewise.
31757 (extract_range_from_binary_expr_1): Likewise.
31758 (extract_range_from_unary_expr): Likewise.
31759 (extract_range_from_comparison): Likewise.
31760 (extract_range_basic): Likewise.
31761 (adjust_range_with_scev): Likewise.
31762 (compare_ranges): Likewise.
31763 (compare_range_with_value): Likewise.
31764 (dump_value_range): Likewise.
31765 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31767 (simplify_cond_using_ranges): Adjust.
31769 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31771 * brig-builtins.def: Added a builtin for class_f64.
31772 * builtin-types.def: Added a builtin type needed by class_f64.
31774 2017-05-03 Jason Merrill <jason@redhat.com>
31776 * timevar.def: Add TV_CONSTEXPR.
31778 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31780 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31782 2017-05-03 Martin Jambor <mjambor@suse.cz>
31784 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31785 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31786 * ipa-cp.c (ipcp_propagate_stage): Do not call
31787 ipa_update_after_lto_read.
31788 * ipa-inline.c (ipa_inline): Likewise.
31790 2017-05-03 Martin Jambor <mjambor@suse.cz>
31792 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31793 tag. Added a default constructor and a destructor.
31794 (ipa_edge_args_sum_t): New class;
31795 (ipa_edge_args_sum): Declare.
31796 (ipa_edge_args_vector): Remove declaration.
31797 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31798 (ipa_free_edge_args_substructures): Remove declaration.
31799 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31800 (ipa_edge_args_info_available_for_edge_p): Likewise.
31801 * ipa-prop.c (ipa_edge_args_vector): Removed.
31802 (edge_removal_hook_holder): Likewise.
31803 (edge_duplication_hook_holder): Likewise.
31804 (ipa_edge_args_sum): New variable.
31805 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31806 ipa_edge_args_vector.
31807 (ipa_free_edge_args_substructures): Likewise.
31808 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31809 ipa_edge_args_vector.
31810 (ipa_edge_removal_hook): Turned into method
31811 ipa_edge_args_sum_t::remove.
31812 (ipa_edge_duplication_hook): Turned into method
31813 ipa_edge_args_sum_t::duplicate.
31814 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31815 registering edge hooks.
31816 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31817 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31818 ipa_edge_args_sum instead of ipa_edge_args_vector.
31819 * ipa-profile.c (ipa_profile): Likewise.
31821 2017-05-03 Martin Jambor <mjambor@suse.cz>
31823 * symbol-summary.h (function_summary): New method exists.
31824 (function_summary::symtab_removal): Deallocate through release.
31825 (call_summary): New class.
31826 (gt_ggc_mx): New overload.
31827 (gt_pch_nx): Likewise.
31828 (gt_pch_nx): Likewise.
31830 2017-05-03 Jeff Law <law@redhat.com>
31832 PR tree-optimization/78496
31833 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31834 from simplify_cond_using_ranges. Split off code to walk
31835 backwards through casts into ...
31836 (simplify_cond_using_ranges_2): New function.
31837 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31838 (execute_vrp): After identifying jump threads, call
31839 simplify_cond_using_ranges_2.
31841 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31844 * ipa-inline.h (inline_summary): Add ctor.
31845 (create_ggc): Do not use ggc_cleared_alloc.
31847 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31848 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31850 * gcc.c (handle_braces): Support escaping in switch matching
31852 * doc/invoke.texi (Spec Files): Document it.
31853 Remove superfluous @code markup in items.
31855 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31857 * diagnostic-show-locus.c (struct column_range): New struct.
31858 (get_affected_columns): New function.
31859 (get_printed_columns): New function.
31860 (struct correction): New struct.
31861 (correction::ensure_capacity): New function.
31862 (correction::ensure_terminated): New function.
31863 (struct line_corrections): New struct.
31864 (line_corrections::~line_corrections): New dtor.
31865 (line_corrections::add_hint): New function.
31866 (layout::print_trailing_fixits): Reimplement in terms of the new
31868 (selftest::test_overlapped_fixit_printing): New function.
31869 (selftest::diagnostic_show_locus_c_tests): Call it.
31871 2017-05-03 Nathan Sidwell <nathan@acm.org>
31873 Canonicalize canonical type hashing
31874 * tree.h (type_hash_canon_hash): Declare.
31875 * tree.c (type_hash_list, attribute_hash_list): Move into
31876 type_hash_canon_hash.
31877 (build_type_attribute_qual_variant): Break out hash code calc into
31878 type_hash_canon_hash.
31879 (type_hash_canon_hash): New. Generic type hash computation.
31880 (build_range_type_1, build_array_type_1, build_function_type,
31881 build_method_type_directly, build_offset_type, build_complex_type,
31882 make_vector_type): Call it.
31884 2017-05-03 Richard Biener <rguenther@suse.de>
31886 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31887 When all DRs have unknown misaligned do not always peel
31888 when there is a store but apply the same costing model as if
31889 there were only loads.
31891 2017-05-03 Richard Biener <rguenther@suse.de>
31894 PR tree-optimization/80492
31895 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31896 compare_base_decls returning dont-know properly.
31898 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31900 * config/arm/iterators.md (CCSI): New mode iterator.
31901 (arch): New mode attribute.
31902 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31903 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31904 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31905 code iterator for success result mode.
31906 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31907 the corresponding new insn generators.
31909 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31912 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31913 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31915 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31917 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31918 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31919 (DDR_A): Wrap DDR argument in brackets.
31920 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31921 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31922 (DDR_REVERSED_P): Likewise.
31924 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31926 PR tree-optimization/79472
31927 * tree-switch-conversion.c (struct switch_conv_info): Add
31928 contiguous_range and default_case_nonstandard fields.
31929 (collect_switch_conv_info): Compute contiguous_range and
31930 default_case_nonstandard fields, don't clear final_bb if
31931 contiguous_range and only the default case doesn't have the required
31933 (check_all_empty_except_final): Set default_case_nonstandard instead
31934 of failing if contiguous_range and the default case doesn't have empty
31936 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31937 and only the default case doesn't have the required constants. Skip
31939 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31940 if default_case_nonstandard.
31941 (build_constructors): Build constant 1 just once. Assert that default
31942 values aren't inserted in between cases if contiguous_range. Skip
31944 (build_arrays): Skip virtual phis.
31945 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31946 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31947 Handle virtual phis.
31948 (gen_inbound_check): Handle default_case_nonstandard case.
31949 (process_switch): Adjust check_final_bb caller. Call
31950 gather_default_values with the first non-default case instead of
31951 default case if default_case_nonstandard.
31953 2017-05-02 Nathan Sidwell <nathan@acm.org>
31955 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31956 check. Fix formatting.
31958 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31960 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31961 errors when comparing specialized and unspecialized times.
31963 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31965 * diagnostic-show-locus.c
31966 (layout::should_print_annotation_line_p): Make private.
31967 (layout::print_annotation_line): Make private.
31968 (layout::annotation_line_showed_range_p): Make private.
31969 (layout::show_ruler): Make private.
31970 (layout::print_source_line): Make private. Pass in line and
31971 line_width, rather than calling location_get_source_line. Drop
31973 (layout::print_leading_fixits): New method.
31974 (layout::print_any_fixits): Rename to...
31975 (layout::print_trailing_fixits): ...this, and make private.
31976 Don't print newline fixits.
31977 (diagnostic_show_locus): Move logic for printing one row into...
31978 (layout::print_line): ...this new function. Move the
31979 location_get_source_line call and error-handling from
31980 print_source_line to here. Call print_leading_fixits, and rename
31981 print_any_fixits to print_trailing_fixits.
31982 (selftest::test_fixit_insert_containing_newline): Update now that
31983 newlines are partially supported.
31984 (selftest::test_fixit_insert_containing_newline_2): New test.
31985 (selftest::test_fixit_replace_containing_newline): Update comments.
31986 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31987 * edit-context.c (class added_line): New class.
31988 (class edited_line): Describe newline handling in comment.
31989 (edited_line::actually_edited_p): New method.
31990 (edited_line::print_content): Delete redundant decl.
31991 (edited_line::m_predecessors): New field.
31992 (edited_file::print_content): Call edited_line::print_content.
31993 (edited_file::print_diff): Update to support newlines.
31994 (edited_file::print_diff_hunk): Likewise.
31995 (edited_file::print_run_of_changed_lines): New function.
31996 (edited_file::print_diff_line): Convert to...
31997 (print_diff_line): ...this.
31998 (edited_file::get_effective_line_count): New function.
31999 (edited_line::edited_line): Initialize new field m_predecessors.
32000 (edited_line::~edited_line): Clean up m_predecessors.
32001 (edited_line::apply_fixit): Handle newlines.
32002 (edited_line::get_effective_line_count): New function.
32003 (edited_line::print_content): New function.
32004 (edited_line::print_diff_lines): New function.
32005 (selftest::test_applying_fixits_insert_containing_newline): New
32007 (selftest::test_applying_fixits_replace_containing_newline): New
32009 (selftest::insert_line): New function.
32010 (selftest::test_applying_fixits_multiple_lines): Add example of
32012 (selftest::edit_context_c_tests): Call the new tests.
32014 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32016 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32017 parameter cand. Update dump information.
32018 (get_computation_cost): Update uses.
32020 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32022 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32023 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32024 (get_computation_at, rewrite_use_address): Update use of
32025 get_computation_aff.
32027 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32029 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32030 (get_computation): Delete.
32031 (get_computation_cost): Implement like get_computation_cost_at.
32032 Use get_computation_at.
32033 (get_computation_cost_at): Delete.
32034 (rewrite_use_nonlinear_expr): Use get_computation_at.
32035 (rewrite_use_compare, remove_unused_ivs): Ditto.
32037 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32039 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32041 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32043 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32044 (ivopts_global_cost_for_size): Rename parameter and update uses.
32045 (iv_ca_recount_cost): Update uses.
32046 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32047 candidates seperately in n_invs and n_cands.
32048 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32052 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32053 (find_inv_vars_cb): New.
32054 (find_depends): Renamed to ...
32055 (find_inv_vars): ... this.
32056 (add_candidate_1, force_var_cost): Call find_inv_vars.
32057 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32059 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32061 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32062 inv_vars. Add inv_exprs.
32063 (struct iv_cand): Rename depends_on to inv_vars.
32064 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32065 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32066 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32067 (dump_cand): Dump inv_vars.
32068 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32069 (record_invariant, find_depends, add_candidate_1): Ditto.
32070 (set_group_iv_cost, force_var_cost): Ditto.
32071 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32072 (get_computation_cost_at, get_computation_cost): Ditto.
32073 (determine_group_iv_cost_generic): Ditto.
32074 (determine_group_iv_cost_address): Ditto.
32075 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32076 (determine_group_iv_costs): Ditto.
32077 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32078 (iv_ca_set_remove_invariants): Renamed to ...
32079 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32080 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32081 (iv_ca_set_add_invariants): Renamed to ...
32082 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32083 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32084 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32085 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32086 (create_new_ivs): Remove useless dump.
32088 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32090 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32092 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32093 (iv_ca_set_no_cp, create_new_iv): Ditto.
32095 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32097 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32099 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32101 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32102 function tree_check2.
32104 2017-05-02 Martin Liska <mliska@suse.cz>
32106 * doc/gcov.texi: Add missing preposition.
32107 * gcov.c (function_info::function_info): Properly fill up
32108 all member variables.
32110 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32112 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32114 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32116 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32118 2017-05-02 Martin Liska <mliska@suse.cz>
32121 * lto-streamer-in.c (lto_read_tree_1): Remove
32122 LTO_STREAMER_DEBUG.
32123 * lto-streamer.c (struct tree_hash_entry): Likewise.
32124 (struct tree_entry_hasher): Likewise.
32125 (tree_entry_hasher::hash): Likewise.
32126 (tree_entry_hasher::equal): Likewise.
32127 (lto_streamer_init): Likewise.
32128 (lto_orig_address_map): Likewise.
32129 (lto_orig_address_get): Likewise.
32130 (lto_orig_address_remove): Likewise.
32131 * lto-streamer.h: Likewise.
32132 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32133 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32135 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32137 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32138 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32139 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32140 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32141 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32142 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32143 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32144 (mm_maskz_sub_ss): New intrinsics.
32145 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32146 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32147 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32148 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32149 (__builtin_ia32_subss_mask_round): New builtins.
32150 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32151 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32152 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32154 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32155 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32157 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32160 2017-05-02 Martin Jambor <mjambor@suse.cz>
32162 PR tree-optimization/78687
32163 * tree-sra.c (access): New field parent.
32164 (process_subtree_disqualification): New function.
32165 (disqualify_candidate): Call it.
32166 (build_accesses_from_assign): Reset write flag if creating an
32168 (build_access_subtree): Fill in parent field and also prpagate
32169 down grp_write flag.
32170 (create_artificial_child_access): New parameter set_grp_write, set
32171 grp_write to its value.
32172 (propagate_subaccesses_across_link): Also propagate grp_write flag
32174 (propagate_all_subaccesses): Push the closest parent back to work
32175 queue if add_access_to_work_queue returned true.
32177 2017-05-02 Richard Biener <rguenther@suse.de>
32179 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32180 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32181 -fstrict-overflow documentation.
32182 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32183 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32184 flag_strict_overflow.
32185 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32186 * lto-opts.c (lto_write_options): Do not stream it.
32187 * lto-wrapper.c (merge_and_complain): Do not handle it.
32188 * opts.c (default_options_table): Do not set -fstrict-overflow.
32189 (finish_options): Likewise do not clear it when sanitizing.
32190 * simplify-rtx.c (simplify_const_relational_operation): Do not
32191 test flag_strict_overflow.
32193 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32195 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32196 using enabled attribute.
32197 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32198 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32199 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32200 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32201 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32202 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32203 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32204 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32205 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32206 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32208 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32210 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32212 2017-05-02 Richard Biener <rguenther@suse.de>
32214 PR tree-optimization/80591
32216 2017-04-10 Richard Biener <rguenther@suse.de>
32218 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32221 2017-05-02 Richard Biener <rguenther@suse.de>
32223 PR tree-optimization/80549
32224 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32225 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32226 headers if they do not have a preheader.
32228 2017-05-02 Martin Liska <mliska@suse.cz>
32231 * common.opt: Fix typo.
32232 * doc/invoke.texi: Likewise.
32234 2017-05-01 Jan Beulich <jbeulich@suse.com>
32236 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32237 swapping, add (x,x,m,x,n) alternative.
32239 2017-05-01 Nathan Sidwell <nathan@acm.org>
32241 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32242 unnecessary unadjusted_alignment check.
32244 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32247 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32249 * gimplify.c (gimplify_cilk_detach): New function.
32250 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32251 * tree-core.h: Document EXPR_CILK_SPAWN.
32252 * tree.h (EXPR_CILK_SPAWN): Define.
32254 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32256 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32257 to use new fixit_hint representation, using the "replace" logic.
32258 (get_line_span_for_fixit_hint): Likewise.
32259 (layout::print_any_fixits): Likewise.
32260 (selftest::test_one_liner_many_fixits): Rename to...
32261 (selftest::test_one_liner_many_fixits_1): ...this, and update
32262 comment and expected output to reflect that the multiple fix-it
32263 hints are now consolidated into one insertion.
32264 (selftest::test_one_liner_many_fixits_2): New test.
32265 (selftest::test_diagnostic_show_locus_one_liner): Update for
32267 (selftest::test_fixit_consolidation): Update for fix-it API
32269 * diagnostic.c (print_parseable_fixits): Likewise.
32270 * edit-context.c (edited_line::m_line_events): Convert from
32271 auto_vec <line_event *> to auto_vec <line_event>.
32272 (class line_event): Convert from abstract base class to a concrete
32273 class, taking over the role of replace_event.
32274 (class insert_event): Delete.
32275 (class replace_event): Rename to class line_event. Convert to
32277 (edit_context::add_fixits): Reimplement.
32278 (edit_context::apply_insert): Delete.
32279 (edit_context::apply_replace): Rename to...
32280 (edit_context::apply_fixit): ...this. Convert to half-open range.
32281 (edited_file::apply_insert): Delete.
32282 (edited_file::apply_replace): Rename to...
32283 (edited_file::apply_fixit): ...this.
32284 (edited_line::~edited_line): Drop deletion of events.
32285 (edited_line::apply_insert): Delete.
32286 (edited_line::apply_replace): Rename to...
32287 (edited_line::apply_fixit): ...this. Convert to half-open range.
32288 Update for change to type of m_line_events.
32289 * edit-context.h (edit_context::apply_insert): Delete.
32290 (edit_context::apply_replace): Rename to...
32291 (edit_context::apply_fixit): ...this.
32293 2017-05-01 Martin Sebor <msebor@redhat.com>
32295 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32298 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32301 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32302 __get_cpuid_max returns 0.
32303 (__get_cpuid_count): Ditto.
32305 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32307 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32308 replacement expression is another instance of one of its arguments.
32310 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32313 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32314 check for stack push/pop autoinc.
32315 * config/i386/i386.c (ix86_agi_dependent): Return false
32316 if the only reason why modified_in_p returned true is that
32317 addr is SP based and set_insn is a push or pop.
32319 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32321 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32324 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32327 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32328 (account_size_time): Use two predicates - exec_pred and
32330 (evaluate_conditions_for_known_args): Compute both clause and
32332 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32333 (inline_summary_t::duplicate): Update.
32334 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32336 (compute_inline_parameters): Likewise.
32337 (estimate_edge_size_and_time): Update caluclation of time.
32338 (estimate_node_size_and_time): Compute both time and nonspecialized
32340 (estimate_ipcp_clone_size_and_time): Update.
32341 (inline_merge_summary): Update.
32342 (do_estimate_edge_time): Update.
32343 (do_estimate_edge_size): Update.
32344 (do_estimate_edge_hints): Update.
32345 (inline_read_section, inline_write_summary): Stream both new predicates.
32346 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32348 (compute_inlined_call_time): Cleanup.
32349 (big_speedup_p): Update.
32350 (edge_badness): Update.
32351 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32352 (size_time_entry): Replace predicate by exec_predicate and
32353 nonconst_predicate.
32354 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32355 (estimate_edge_time): Return also nonspec_time.
32356 (reset_edge_growth_cache): Update.
32358 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32360 PR rtl-optimization/80491
32361 * ifcvt.c (noce_process_if_block): When looking for x setter
32362 with missing else_bb, don't check only the insn right before
32363 cond_earliest, but look for the last insn that x is modified in
32364 within the same bb.
32366 PR rtl-optimization/80491
32367 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32369 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32371 PR tree-optimization/80487
32372 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32374 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32376 PR tree-optimization/79697
32377 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32378 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32379 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32381 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32382 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32384 2017-04-28 Martin Sebor <msebor@redhat.com>
32386 PR tree-optimization/80523
32387 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32388 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32390 (maybe_warn, format_directive, parse_directive): Use new functions.
32391 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32393 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32395 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32397 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32399 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32400 target_header_dir): Set correctly.
32401 * configure: Regenerated.
32402 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32403 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32404 instead of SYSTEM_HEADER_DIR.
32406 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32408 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32409 (estimate_local_effects): Likewise.
32410 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32411 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32412 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32413 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32414 do_estimate_edge_time, estimate_edge_time): Likewise.
32415 * ipa-inline-analysis.c (estimate_node_size_and_time,
32416 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32417 (estimate_time_after_inlining): Remove.
32419 2017-04-28 Martin Liska <mliska@suse.cz>
32421 * doc/gcov.texi: Enhance documentation of gcov.
32423 2017-04-28 Martin Liska <mliska@suse.cz>
32425 * doc/gcov.texi: Sort options in alphabetic order.
32426 * doc/gcov-dump.texi: Likewise.
32427 * doc/gcov-tool.texi: Likewise.
32428 * gcov.c (print_usage): Likewise.
32429 * gcov-dump.c (print_usage): Likewise.
32430 * gcov-tool.c (print_merge_usage_message): Likewise.
32431 (print_rewrite_usage_message): Likewise.
32432 (print_overlap_usage_message): Likewise.
32434 2017-04-28 Martin Liska <mliska@suse.cz>
32436 PR gcov-profile/53915
32437 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32439 2017-04-28 Martin Liska <mliska@suse.cz>
32441 PR gcov-profile/79891
32442 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32443 is marked by compiler as living on a line.
32444 (get_cycles_count): Remove usage of the union.
32445 (output_intermediate_file): Likewise.
32446 (find_source): Fix GNU coding style.
32447 (accumulate_line_counts): Remove old non-all block mode.
32448 (output_lines): Remove usage of the union.
32449 * profile.c (output_location): Include all BBs, even if
32450 belonging to a same line (and file) as a previous BB.
32452 2017-04-28 Martin Liska <mliska@suse.cz>
32454 * gcov.c (process_args): Handle new argument 'w'.
32455 (read_graph_file): Assign ID to BBs.
32456 (output_branch_count): Display BB # if verbose flag is set.
32457 (output_lines): Likewise for arcs.
32458 (print_usage): Add '--verbose' option help.
32459 * doc/gcov.texi: Document --verbose (-w) option.
32461 2017-04-28 Martin Liska <mliska@suse.cz>
32463 * gcov.c (struct block_location_info): New struct.
32464 (process_file): Fill up the new structure.
32465 (read_graph_file): Replace usage of encoding by the newly added
32467 (add_line_counts): Likewise.
32468 (accumulate_line_counts): Remove usage of the union.
32469 (function_info::function_info): New function.
32470 (function_info::~function_info): Likewise.
32471 (process_file): Call delete instead of release_function.
32472 (release_function): Release the function.
32473 (release_structures): Call delete instead of release_function.
32474 (solve_flow_graph): Replace usage of num_blocks.
32475 (find_exception_blocks): Likewise.
32476 (output_lines): Fix GNU coding style.
32478 2017-04-28 Martin Liska <mliska@suse.cz>
32481 * coverage.c (coverage_remove_note_file): New function.
32482 * coverage.h: Declare the function.
32483 * toplev.c (finalize): Clean if an error has been seen.
32485 2017-04-28 Martin Liska <mliska@suse.cz>
32487 PR gcov-profile/80031
32488 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32489 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32490 * gcov.c (read_graph_file): Read just number of blocks.
32491 * profile.c (branch_prob): Do not stream 0 flags per a basic
32494 2017-04-28 Martin Liska <mliska@suse.cz>
32496 * gcov-dump.c (tag_*): Add new argument to declarations.
32497 (dump_gcov_file): Likewise.
32498 (tag_blocks): Add and use new argument depth.
32499 (tag_arcs): Likewise.
32500 (tag_lines): Likewise.
32501 (tag_counters): Likewise.
32502 (tag_summary): Likewise.
32503 (dump_working_sets): Use depth to do a proper indentation.
32505 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32508 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32509 * symtab.c (symtab_node::debug_symtab): Move definition here.
32511 2017-04-28 Richard Biener <rguenther@suse.de>
32513 * lto-streamer.h (LTO_major_version): Bump to 7.
32515 2017-04-28 Richard Biener <rguenther@suse.de>
32517 * tree-vrp.c (assert_info): New struct.
32518 (add_assert_info): New helper.
32519 (register_edge_assert_for_2): Refactor to add asserts to a vector
32521 (register_edge_assert_for_1): Likewise.
32522 (register_edge_assert_for): Likewise.
32523 (finish_register_edge_assert_for): New helper actually registering
32524 asserts where live on edge.
32525 (find_conditional_asserts): Adjust.
32526 (find_switch_asserts): Likewise.
32527 (evrp_dom_walker::try_find_new_range): Generalize.
32528 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32530 2017-04-27 Marek Polacek <polacek@redhat.com>
32533 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32534 arg10 and arg11 to itype.
32536 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32538 * doc/extend.texi (Object Size Checking): Improve grammar.
32540 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32543 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32544 that the logic for permitting reciprocal estimates matches that
32547 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32550 * tree.c (type_cache_hasher::equal): Only compare
32551 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32552 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32553 non-aggregate element types.
32554 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32555 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32557 2017-04-27 Richard Biener <rguenther@suse.de>
32559 PR middle-end/80533
32560 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32561 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32562 keeping a reference to a trailing array.
32564 2017-04-27 Richard Biener <rguenther@suse.de>
32566 PR middle-end/80539
32567 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32568 being in loop-closed SSA form conservatively.
32569 (chrec_fold_multiply_poly_poly): Likewise.
32571 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32573 PR middle-end/79665
32574 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32575 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32577 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32580 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32581 (aarch64_function_arg_alignment): Return unsigned int again, but still
32582 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32583 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32584 Don't emit -Wpsabi note.
32585 (aarch64_function_arg_boundary): Likewise.
32586 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32589 2017-04-26 Nathan Sidwell <nathan@acm.org>
32591 * tree.h (crc32_unsigned_n): Declare.
32592 (crc32_unsigned, crc32_unsigned): Make inline.
32593 * tree.c (crc32_unsigned_bits): Replace with ...
32594 (crc32_unsigned_n): ... this.
32595 (crc32_unsigned, crc32_byte): Remove.
32596 (crc32_string): Remove unnecessary braces.
32598 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32600 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32601 * ipa-inline-analysis.c (MAX_TIME): Remove.
32602 (account_size_time): Use sreal for time.
32603 (dump_inline_summary): Update.
32604 (estimate_function_body_sizes): Update.
32605 (estimate_edge_size_and_time): Update.
32606 (estimate_calls_size_and_time): Update.
32607 (estimate_node_size_and_time): Update.
32608 (inline_merge_summary): Update.
32609 (inline_update_overall_summary): Update.
32610 (estimate_time_after_inlining): Update.
32611 (inline_read_section): Update.
32612 (inline_write_summary): Update.
32613 * ipa-inline.c (compute_uninlined_call_time): Update.
32614 (compute_inlined_call_time): Update.
32615 (recursive_inlining): Update.
32616 (inline_small_functions): Update.
32617 (dump_overall_stats): Update.
32618 * ipa-inline.h: Include sreal.h.
32619 (size_time_entry): Turn time to sreal.
32620 (inline_summary): Turn self_time nad time to sreal.
32622 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32624 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32626 (sreal::stream_out, sreal::stream_in): New.
32627 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32629 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32631 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32634 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32637 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32639 (dimode_scalar_chain::compute_convert_gain): Ditto.
32640 (dimode_scalar_chain::make_vector_copies): Ditto.
32641 (dimode_scalar_chain::convert_reg): Ditto.
32642 (dimode_scalar_chain::convert_insn): Ditto.
32643 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32644 (VI248_AVX512BW_1): New mode iterator.
32645 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32646 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32649 2017-04-25 Martin Sebor <msebor@redhat.com>
32651 PR tree-optimization/80497
32652 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32653 constants are representable in HOST_WIDE_INT.
32654 (parse_directive): Ditto.
32656 2017-04-25 Martin Sebor <msebor@redhat.com>
32659 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32660 (new_zero_array): Adjust signature.
32661 (dom_info::dom_init): Used unsigned rather that size_t.
32662 (dom_info::dom_info): Same.
32664 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32665 Jakub Jelinek <jakub@redhat.com>
32668 * config/arm/arm.c: Include gimple.h.
32669 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32670 returns negative, increment ncrn only if it returned positive.
32671 (arm_needs_doubleword_align): Return int instead of bool,
32672 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32673 members, but if there is any such non-FIELD_DECL
32674 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32675 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32676 returns negative, increment nregs only if it returned positive.
32677 (arm_setup_incoming_varargs): Likewise.
32678 (arm_function_arg_boundary): Emit -Wpsabi note if
32679 arm_needs_doubleword_align returns negative, return
32680 DOUBLEWORD_ALIGNMENT only if it returned positive.
32682 2017-04-25 Marek Polacek <polacek@redhat.com>
32685 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32686 first argument to type.
32688 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32691 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32692 type checks to test for compatibility instead of equality.
32694 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32695 Jakub Jelinek <jakub@redhat.com>
32698 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32700 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32701 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32702 the alignment computation, but return their maximum in warn_alignment.
32703 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32704 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32706 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32707 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32710 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32712 * config/arc/simdext.md (dmpyh): Fix typo.
32714 2017-04-25 Richard Biener <rguenther@suse.de>
32716 PR tree-optimization/80492
32717 * alias.c (compare_base_decls): Handle registers with asm
32718 specification conservatively.
32719 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32720 compare_base_decls returning dont-know properly.
32722 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32724 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32725 (legitimate_offset_address_p): New function.
32726 (arc_legitimate_address_p): Use above function.
32728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32730 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32732 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32734 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32735 ACCH registers whenever they are available.
32737 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32739 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32740 double regs fix when not used.
32742 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32744 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32746 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32747 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32749 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32751 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32752 when emitting short ADD instructions.
32754 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32756 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32758 (cmpsi_cc_c_insn): Likewise.
32759 (cbranchsi4_scratch): Compute proper instruction length using
32760 compact_hreg_operand.
32761 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32763 2017-04-25 Richard Biener <rguenther@suse.de>
32765 PR middle-end/80509
32766 * passes.c (pass_manager::pass_manager): Initialize
32767 m_name_to_pass_map.
32769 2017-04-25 Richard Biener <rguenther@suse.de>
32771 PR tree-optimization/79201
32772 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32774 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32777 * config/s390/vector.md: Split MEM->GPR vector moves for
32778 non-s_operand addresses.
32780 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32783 * config/s390/predicates.md (reload_const_wide_int_operand): New
32785 * config/s390/s390.md ("movti"): Remove d/P alternative.
32786 ("movti_bigconst"): New pattern definition.
32788 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32791 * s390-protos.h (s390_expand_cs_hqi): Removed.
32792 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32793 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32794 modes as well as CCZ1mode and CCZmode.
32795 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32796 signature of s390_emit_compare_and_swap.
32797 (s390_expand_cs_hqi): Likewise, make static.
32798 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32799 compare-and-swap, in some cases.
32800 (s390_expand_cs): Wrapper function.
32801 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32803 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32804 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32805 patterns for small and large integers. Forbid symref memory operands.
32806 Move expander to s390.c. Require cc register.
32807 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32808 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32809 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32810 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32811 symref memory operands. Remove CC mode and call s390_match_ccmode
32813 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32815 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32817 * config/s390/s390.md (define_peephole2): New peephole to help
32818 combining the load-and-test pattern with volatile memory.
32820 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32822 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32823 with CCZmode for TARGET_Z196.
32825 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32827 PR rtl-optimization/80501
32828 * combine.c (make_compound_operation_int): Set subreg_code to SET
32829 even for AND with mask of the sign bit of mode.
32831 PR rtl-optimization/80500
32832 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32833 sum's initial value.
32835 2017-04-25 Julian Brown <julian@codesourcery.com>
32836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32838 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32840 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32842 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32844 2017-04-25 Julian Brown <julian@codesourcery.com>
32845 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32847 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32848 (thunderx2t99_sha): New Reservation.
32850 2017-04-25 Julian Brown <julian@codesourcery.com>
32851 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32853 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32854 type for 1-element load.
32856 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32858 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32860 2017-04-24 Martin Jambor <mjambor@suse.cz>
32862 PR tree-optimization/80293
32863 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32864 char arrays not totally scalarizable if it is false.
32865 (analyze_all_variable_accesses): Pass correct value in the new
32866 parameter. Add a statistics counter.
32868 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32870 PR middle-end/79931
32871 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32873 2017-04-24 Richard Biener <rguenther@suse.de>
32875 PR tree-optimization/80494
32876 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32877 out for complex types.
32879 2017-04-24 Richard Biener <rguenther@suse.de>
32881 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32882 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32883 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32884 (DFS): Adjust and never fail.
32885 (sccvn_dom_walker::fail): Remove.
32886 (sccvn_dom_walker::before_dom_children): Adjust.
32887 (run_scc_vn): Likewise and never fail.
32888 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32889 (pass_fre::execute): Likewise.
32891 2017-04-24 Richard Biener <rguenther@suse.de>
32893 PR tree-optimization/79725
32894 * tree-ssa-sink.c (statement_sink_location): Return whether
32895 failure reason was zero uses. Move that check later.
32896 (sink_code_in_bb): Deal with zero uses by removing the stmt
32899 2017-04-24 Richard Biener <rguenther@suse.de>
32902 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32903 pointer-based references.
32905 2017-04-24 Richard Biener <rguenther@suse.de>
32908 * pass_manager.h (pass_manager::operator new): Remove.
32909 (pass_manager::operator delete): Likewise.
32910 * passes.c (pass_manager::operator new): Remove.
32911 (pass_manager::operator delete): Likewise.
32912 (pass_manager::pass_manager): Zero individual pass members.
32914 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32917 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32918 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32919 Check "XEXP (src, 1)" operand here.
32920 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32921 Check "XEXP (src, 1)" operand here.
32922 (dimode_scalar_chain::make_vector_copies): Detect count register
32923 of a shift instruction. Zero extend count register from QImode
32924 to DImode to satisfy vector shift pattern count operand predicate.
32925 Substitute vector shift count operand with a DImode copy.
32926 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32929 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32931 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32932 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32933 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32934 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32935 (UNSPEC_NOREX_MEM): Remove definition.
32937 2017-04-21 Richard Biener <rguenther@suse.de>
32939 PR tree-optimization/79547
32940 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32941 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32942 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32943 without any constraints.
32945 2017-04-21 Richard Biener <rguenther@suse.de>
32947 PR tree-optimization/78847
32948 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32950 2017-04-21 Richard Biener <rguenther@suse.de>
32952 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32953 (build_distinct_type_copy): Likewise.
32954 (build_variant_type_copy): Likewise.
32955 * tree.c (build_qualified_type): Pass down mem-stat info.
32956 (build_distinct_type_copy): Likewise.
32957 (build_variant_type_copy): Likewise.
32959 2017-04-21 Richard Biener <rguenther@suse.de>
32961 PR tree-optimization/80237
32962 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32964 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32965 for a simplified result.
32967 2016-04-21 Richard Biener <rguenther@suse.de>
32969 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32970 sth as strict as a simple_iv but a chrec without symbols and an
32971 operand defined in the loop we are peeling (and not some subloop).
32972 (propagate_constants_for_unrolling): Propagate all constants.
32974 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32977 * config/i386/i386.c (print_reg): Remove assert for disalowed
32978 regno values, call output_operand_lossage instead.
32980 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32983 * config/i386/constraints.md (Yc): New register constraint.
32984 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32985 Use Yc constraint for alternative 2 of operand 0. Remove
32986 preferred_for_speed attribute.
32988 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32990 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32991 lastprivate clauses in SIMT case.
32993 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32995 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32997 2017-04-20 Richard Biener <rguenther@suse.de>
32999 PR tree-optimization/57796
33000 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33001 as N scalar stores.
33002 (vect_model_load_cost): Cost gathers as N scalar loads.
33004 2017-04-20 Richard Biener <rguenther@suse.de>
33006 * ggc-page.c (ggc_allocated_p): Rename to ...
33007 (safe_lookup_page_table_entry): ... this and return the lookup
33009 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33011 2017-04-20 Richard Biener <rguenther@suse.de>
33013 PR tree-optimization/80453
33014 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33015 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33016 from the conditions.
33017 (vn_phi_eq): Pass them down.
33018 (vn_phi_lookup): Record them.
33019 (vn_phi_insert): Likewise.
33021 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33023 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33024 uninitialized variable warning to avoid buffer overrun.
33026 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33029 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33030 is suppressed for '{ 0 }' in C.
33032 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33034 * BASE-VER: Set to 8.0.0.
33036 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33038 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33039 priority .init_array and .fini_array section with SECTION_NOTYPE
33042 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33044 PR middle-end/80423
33045 * tree.h (build_array_type): Add typeless_storage default argument.
33046 * tree.c (type_cache_hasher::equal): Also compare
33047 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33048 (build_array_type): Add typeless_storage argument, set
33049 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33051 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33052 (build_array_type): Likewise. Add typeless_storage argument.
33054 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33055 Jakub Jelinek <jakub@redhat.com>
33057 PR tree-optimization/80426
33058 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33059 operation on symbolic operands, also compute the overflow for the
33060 invariant part when the operation degenerates into a negation.
33062 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33065 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33066 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33069 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33071 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33074 * config/avr/avr.c (tree.h): Include it.
33075 (cgraph.h): Include it.
33076 (avr_encode_section_info): Don't warn for uninitialized progmem
33077 variable if it's just an alias.
33079 2017-04-19 Richard Biener <rguenther@suse.de>
33082 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33083 when needed by AutoPGO.
33085 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33088 * doc/lto.texi: Remove an extra 'that'.
33090 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33092 PR rtl-optimization/80429
33093 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33094 are only used in debug insns.
33096 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33097 Vladimir Makarov <vmakarov@redhat.com>
33099 * config/sparc/predicates.md (input_operand): Add comment. Return
33100 true for any memory operand when LRA is in progress.
33101 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33103 2017-04-18 Jeff Law <law@redhat.com>
33106 * mips.md ({return,simple_return}_internal): Do not overwrite
33109 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33111 PR tree-optimization/80443
33112 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33113 instead of adding 1, subtract -1 and similarly instead of subtracting
33116 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33118 PR rtl-optimization/80357
33119 * haifa-sched.c (tmp_bitmap): New variable.
33120 (model_recompute): Handle duplicate use records.
33121 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33122 (free_global_sched_pressure_data): Free it.
33124 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33127 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33128 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33129 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33130 instead of SYSTEM_HEADER_DIR.
33132 2017-04-18 Jeff Law <law@redhat.com>
33134 PR middle-end/80422
33135 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33136 predecessors after walking up the insn chain.
33138 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33141 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33142 sizetype type into debug info.
33144 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33147 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33148 unneeded test for TARGET_UPPER_REGS_SF.
33149 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33151 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33154 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33155 instead of gsi_after_labels.
33157 2017-04-18 Jeff Law <law@redhat.com>
33159 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33163 2017-04-13 Jeff Law <law@redhat.com>
33164 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33165 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33167 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33170 * config/avr/avr.c (intl.h): Include it.
33171 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33173 2017-04-18 Martin Liska <mliska@suse.cz>
33175 PR gcov-profile/78783
33176 * gcov-tool.c (gcov_output_files): Validate that destination
33177 file is either removed by the tool or by a user.
33179 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33180 Guy Benyei <guybe@mellanox.com>
33182 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33183 block, and do not negate it, the stored id is already negative.
33185 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33187 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33189 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33192 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33193 masks of options that should be turned off if the VSX vector
33194 options are turned off.
33195 (OTHER_P8_VECTOR_MASKS): Likewise.
33196 (OTHER_VSX_VECTOR_MASKS): Likewise.
33197 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33198 rs6000_disable_incompatible_switches to validate no type switches
33200 (rs6000_incompatible_switch): New function to disallow turning on
33201 other vector options if -mno-vsx, -mno-power8-vector, or
33202 -mno-power9-vector are specified.
33204 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33206 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33208 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33210 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33211 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33212 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33213 (ARG_POINTER_CFA_OFFSET): Likewise.
33215 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33217 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33218 conditions to take advantage of various optimizations.
33220 2017-04-13 Jeff Law <law@redhat.com>
33222 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33223 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33224 (zero_extendsidi2_dext): Likewise.
33226 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33229 * fold-const.c (fold_ternary_loc): Revert
33230 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33233 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33235 PR rtl-optimization/80343
33236 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33237 new scratch pseudo.
33239 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33242 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33243 to ubsan_encode_value.
33245 2017-04-13 Jeff Law <law@redhat.com>
33247 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33248 appearing in DEBUG_INSNs.
33250 2017-04-13 Martin Liska <mliska@suse.cz>
33252 PR gcov-profile/80413
33253 * gcov-io.c (gcov_write_string): Copy to buffer just when
33254 allocated size is greater than zero.
33256 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33259 * dwarf2out.c (decls_for_scope): Ignore declarations of
33260 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33262 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33265 * ipa-visibility.c (non_local_p): Fix typos.
33266 (localize_node): When localizing symbol in same comdat group,
33267 dissolve the group only when we know external symbols are going
33269 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33271 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33273 PR tree-optimization/79390
33274 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33275 order does not result in usable sequence, retry with reversed operand
33281 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33282 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33283 op0 instead of fold_convert_loc (loc, type, arg0).
33285 2017-04-12 Jeff Law <law@redhat.com>
33287 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33288 has a delay slot in the generated code.
33290 * config/cris/cris.md (cris_preferred_reload_class): Return
33291 GENNONACR_REGS rather than GENERAL_REGS.
33293 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33296 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33297 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33298 signedness of the result type.
33300 2017-04-12 Richard Biener <rguenther@suse.de>
33301 Jeff Law <law@redhat.com>
33303 PR tree-optimization/80359
33304 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33305 trim stores to TARGET_MEM_REFs.
33307 2017-04-12 Richard Biener <rguenther@suse.de>
33309 PR tree-optimization/79390
33310 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33311 threading case even more.
33313 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33316 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33317 for quad_address_p for TImode, instead of just not indexed_address.
33319 2017-04-12 Richard Biener <rguenther@suse.de>
33320 Bernd Edlinger <bernd.edlinger@hotmail.de>
33322 PR middle-end/79671
33323 * alias.c (component_uses_parent_alias_set_from): Handle
33324 TYPE_TYPELESS_STORAGE.
33325 (get_alias_set): Likewise.
33326 * tree-core.h (tree_type_common): Add typeless_storage flag.
33327 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33328 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33329 for types containing members with TYPE_TYPELESS_STORAGE.
33330 (place_field): Likewise.
33331 (layout_type): Likewise for ARRAY_TYPE.
33332 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33333 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33334 TYPE_TYPELESS_STORAGE.
33335 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33337 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33340 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33341 first argument to type.
33343 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33347 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33348 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33349 (rs6000_expand_binop_builtin): Likewise.
33350 (rs6000_expand_ternop_builtin): Likewise; also add missing
33351 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33352 vshasigma built-ins.
33353 * doc/extend.texi: Document that vec_xxpermdi's third argument
33354 must be a constant.
33356 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33358 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33359 Use shift_const cost parameter when calculating gain of STV shifts.
33361 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33363 PR rtl-optimization/70478
33364 * lra-constraints.c (process_alt_operands): Check memory for
33365 disfavoring memory insn operand.
33367 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33369 PR middle-end/80100
33370 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33371 left shift in unsigned HOST_WIDE_INT type.
33373 PR rtl-optimization/80385
33374 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33375 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33378 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33379 if they have any depend clauses.
33381 2017-04-11 Martin Liska <mliska@suse.cz>
33384 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33385 * ipa-split.c (split_function): Create a local comdat symbol
33386 if caller is in a comdat group.
33388 2017-04-11 Martin Liska <mliska@suse.cz>
33391 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33394 2017-04-11 Martin Sebor <msebor@redhat.com>
33396 PR middle-end/80364
33397 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33398 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33400 (directive::set_width, directive::set_precision, format_character):
33402 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33405 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33408 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33409 conflict, set target->arch_name instead of target->cpu_name.
33411 2017-04-11 Richard Biener <rguenther@suse.de>
33413 PR tree-optimization/80374
33414 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33415 build_zero_cst, remove fold_convertible_p check again.
33417 2017-04-11 Martin Liska <mliska@suse.cz>
33420 * ubsan.c (instrument_object_size): Do not instrument register
33423 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33426 * config/i386/i386-builtin-types.def
33427 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33428 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33429 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33430 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33431 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33432 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33433 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33434 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33435 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33436 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33437 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33438 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33439 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33440 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33441 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33442 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33443 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33444 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33445 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33446 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33447 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33448 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33449 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33450 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33451 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33452 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33453 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33454 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33455 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33456 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33457 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33458 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33459 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33460 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33461 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33462 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33463 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33464 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33465 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33466 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33467 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33468 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33469 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33470 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33471 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33472 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33474 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33475 flag to second_arg_count, handle 4 argument function type _COUNT
33476 aliases, handle second_arg_count on second argument rather than last.
33478 2017-04-10 Jeff Law <law@redhat.com>
33480 PR tree-optimization/80374
33481 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33482 record anything if we can not convert integer_zero_node to the
33485 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33488 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33489 Enhance special handling given to the TARGET_P9_MINMAX option in
33490 relation to certain other options.
33492 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33494 PR tree-optimization/80153
33495 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33496 remove POINTER_PLUS_EXPR's base part directly, rather than through
33499 2017-04-10 Richard Biener <rguenther@suse.de>
33500 Bin Cheng <bin.cheng@arm.com>
33502 PR tree-optimization/80153
33503 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33504 the first element of pointer type aff_tree. Build result expr in
33506 (add_elt_to_tree): Convert to type unconditionally. Remove other
33507 fold_convert calls.
33508 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33509 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33511 2017-04-10 Richard Biener <rguenther@suse.de>
33513 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33516 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33518 PR rtl-optimization/70478
33519 * lra-constraints.c (curr_small_class_check): New.
33520 (update_and_check_small_class_inputs): New.
33521 (process_alt_operands): Update curr_small_class_check. Disfavor
33522 alternative insn memory operands. Check available regs for small
33525 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33528 * config/mips/mips.opt (-mvirt): Update description.
33529 * doc/invoke.texi (-mvirt): Likewise.
33531 2017-04-10 Richard Biener <rguenther@suse.de>
33533 PR middle-end/80362
33534 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33535 looking for NEGATE_EXPR in -A / -B to A / B folding.
33537 2017-04-10 Martin Liska <mliska@suse.cz>
33539 PR gcov-profile/80224
33540 * gcov.c (print_usage): Fix usage string.
33541 (get_gcov_intermediate_filename): Remove.
33542 (output_gcov_file): Use both for normal and intermediate format.
33543 (generate_results): Do not initialize special file for
33544 intermediate format.
33546 2017-04-10 Richard Biener <rguenther@suse.de>
33548 PR tree-optimization/80304
33549 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33552 2017-04-10 Nathan Sidwell <nathan@acm.org>
33555 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33556 (rs6000_init_builtins): Use it.
33558 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33560 * config/arm/arm.md (<mrc>): Add mode to SET source.
33561 (<mrrc>): Likewise.
33563 2017-04-10 Richard Biener <rguenther@suse.de>
33565 PR middle-end/80344
33566 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33568 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33571 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33572 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33573 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33574 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33575 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33576 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33577 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33578 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33579 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33580 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33581 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33582 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33583 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33584 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33585 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33586 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33587 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33588 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33589 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33590 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33591 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33592 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33593 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33595 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33597 PR rtl-optimization/70478
33598 * lra-constraints.c: Reverse the last patch.
33600 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33602 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33603 Add comment for WCHAR_T.
33605 2017-04-08 Martin Liska <mliska@suse.cz>
33608 2017-04-07 Martin Liska <mliska@suse.cz>
33611 * ipa-split.c (split_function): Add function part to a same comdat
33614 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33617 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33619 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33621 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33622 (vec_pairing): ...to this.
33623 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33624 (rs6000_sched_init): Adjust for name change.
33625 (struct rs6000_sched_context): Likewise.
33626 (rs6000_init_sched_context): Likewise.
33627 (rs6000_set_sched_context): Likewise.
33629 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33635 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33637 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33638 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33639 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33641 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33643 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33645 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33647 PR rtl-optimization/70703
33648 * ira-color.c (update_conflict_hard_regno_costs): Use
33649 int64_t instead of HOST_WIDE_INT.
33651 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33653 PR rtl-optimization/70478
33654 * lra-constraints.c (process_alt_operands): Disfavor alternative
33655 insn memory operands.
33657 2017-04-07 Jeff Law <law@redhat.com>
33659 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33660 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33662 2017-04-07 Martin Liska <mliska@suse.cz>
33665 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33666 Show error message instead of an ICE.
33668 2017-04-07 Martin Liska <mliska@suse.cz>
33671 * ipa-split.c (split_function): Add function part to a same comdat
33674 2017-04-07 Richard Biener <rguenther@suse.de>
33676 PR middle-end/80341
33677 * tree.c (get_unwidened): Also handle ! for_type case for
33679 * convert.c (do_narrow): Split out from ...
33680 (convert_to_integer_1): ... here. Do not pass final truncation
33681 type to get_unwidened for TRUNC_DIV_EXPR.
33683 2017-04-07 Richard Biener <rguenther@suse.de>
33685 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33687 (aff_combination_const): Adjust.
33688 (aff_combination_scale): Likewise.
33689 (aff_combination_add_elt): Likewise.
33690 (aff_combination_add_cst): Likewise.
33691 (aff_combination_convert): Likewise.
33692 (add_elt_to_tree): Likewise. Remove unused argument.
33693 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33695 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33697 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33699 * config/arm/arm.c (arm_default_short_enums): Use
33700 ARM_DEFAULT_SHORT_ENUMS.
33701 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33703 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33706 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33707 members with redundant out-of-class redeclaration.
33709 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33712 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33713 * config/i386/i386.md (*zero_extendsidi2):
33714 Add (?*x,*x) and (?*v,*v) alternatives.
33716 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33719 * config/i386/i386.c (ix86_expand_builtin)
33720 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33721 mode from insn data. Convert operands to insn operand mode.
33722 Copy operands that don't satisfy insn predicate to a register.
33724 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33726 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33729 2017-04-06 Richard Biener <rguenther@suse.de>
33731 PR tree-optimization/80334
33732 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33733 preserve alignment of accesses.
33735 2017-04-06 Richard Biener <rguenther@suse.de>
33737 PR tree-optimization/80262
33738 * tree-sra.c (build_ref_for_offset): Preserve address-space
33740 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33741 Drop useless address-space information on MEM_REF offsets.
33743 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33745 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33747 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33749 PR rtl-optimization/70703
33750 * ira-color.c (update_conflict_hard_regno_costs): Use
33751 HOST_WIDE_INT instead of long.
33753 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33756 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33757 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33759 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33760 for x86_64 target. Handle -m3dnowa option.
33762 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33764 PR rtl-optimization/70703
33765 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33766 (update_conflict_hard_regno_costs): Use long instead of unsigned
33767 arithmetic for cost calculation.
33769 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33770 Bernd Edlinger <bernd.edlinger@hotmail.de>
33773 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33776 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33779 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33780 ptr_mode with Pmode throughout.
33781 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33782 into probe_stack_range and use DImode.
33784 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33787 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33788 call_eh_return is true.
33790 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33792 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33793 Initialize last_match_fntype_index.
33795 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33798 * tree-nvr.c: Include internal-fn.h.
33799 (pass_return_slot::execute): Ignore internal calls without
33802 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33803 Richard Biener <rguenther@suse.de>
33806 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33807 captures used multiple times, except for the last use.
33808 * generic-match-head.c: Include gimplify.h.
33810 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33812 PR tree-optimization/79390
33813 * target.h (struct noce_if_info): Declare.
33814 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33815 * target.def (noce_conversion_profitable_p): New target hook.
33816 * ifcvt.h (struct noce_if_info): New type, moved from ...
33817 * ifcvt.c (struct noce_if_info): ... here.
33818 (noce_conversion_profitable_p): Renamed to ...
33819 (default_noce_conversion_profitable_p): ... this. No longer
33821 (noce_try_store_flag_constants, noce_try_addcc,
33822 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33823 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33824 instead of noce_conversion_profitable_p.
33825 * config/i386/i386.c: Include ifcvt.h.
33826 (ix86_option_override_internal): Don't override
33827 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33828 (ix86_noce_conversion_profitable_p): New function.
33829 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33830 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33831 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33832 * doc/tm.texi: Regenerated.
33834 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33836 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33839 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33842 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33843 instructions for small multiply cores.
33845 2017-04-04 Jeff Law <law@redhat.com>
33847 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33849 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33850 that are not set by the loop over the elements.
33852 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33855 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33856 int mode, convert_modes it to mode as unsigned, otherwise use
33857 lowpart_subreg to mode rather than SImode.
33858 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33859 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33860 Use DImode instead of SImode for the shift count operand.
33861 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33864 2017-04-04 Richard Biener <rguenther@suse.de>
33866 PR middle-end/80281
33867 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33868 arithmetic done for the negate or the plus. Simplify.
33869 (A - (-B) -> A + B): Likewise.
33870 * fold-const.c (split_tree): Make sure to not negate pointers.
33872 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33874 PR rtl-optimization/60818
33875 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33876 a compare of comparisons with the thing compared if this results
33877 in a different machine mode.
33879 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33881 * alias.c (base_alias_check): Fix typo in comment.
33882 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33883 * cgraphunit.c (symbol_table::compile): Likewise.
33884 * collect2.c (maybe_run_lto_and_relink): Likewise.
33885 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33886 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33887 * config/avr/avr.c (avr_map_op_t): Likewise.
33888 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33889 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33890 * config/epiphany/epiphany.md (movcc): Likewise.
33891 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33892 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33894 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33895 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33896 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33897 * config/sh/sh.c (sh_rtx_costs): Likewise.
33898 * fold-const.c (fold_truth_andor): Likewise.
33899 * genautomata.c (collapse_flag): Likewise.
33900 * gengtype.h (struct type::u::s): Likewise.
33901 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33902 * input.c (FORMAT_AMOUNT): Likewise.
33903 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33904 (known_aggs_to_agg_replacement_list): Likewise.
33905 * ipa-inline-analysis.c: Likewise.
33906 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33907 * ipa-polymorphic-call.c
33908 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33909 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33910 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33912 * modulo-sched.c (apply_reg_moves): Likewise.
33913 * omp-expand.c (build_omp_regions_1): Likewise.
33914 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33915 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33916 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33917 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33918 * value-prof.c: Likewise.
33919 * var-tracking.c (val_reset): Likewise.
33921 2017-04-03 Richard Biener <rguenther@suse.de>
33923 PR tree-optimization/80275
33924 * fold-const.c (split_address_to_core_and_offset): Handle
33927 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33929 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33930 descriptors is at least equal to that of functions.
33932 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33934 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33936 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33939 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33940 (mov<IMOD4:mode>): New expander.
33941 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33943 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33945 PR rtl-optimization/79405
33946 * fwprop.c (propagations_left): New variable.
33947 (forward_propagate_into): Decrement it.
33948 (fwprop_init): Initialize it.
33949 (fw_prop): If the variable has reached zero, stop propagating.
33950 (fwprop_addr): Ditto.
33952 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33955 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33956 a FUNCTION_DECL, pass it as decl instead of origin to
33959 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33961 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33964 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33967 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33968 TARGET_VSX_SMALL_INTEGER.
33970 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33972 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33973 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33975 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33977 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33978 extraction from odd-numbered MSA register.
33980 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33982 PR middle-end/80173
33983 * expmed.c (store_bit_field_1): Don't attempt to create
33984 a word subreg out of hard registers wider than word if they
33985 have HARD_REGNO_NREGS of 1 for their mode.
33987 PR middle-end/80163
33988 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33989 conversions to integer types wider than word and pointer.
33992 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33993 (rtx_equal_for_cselib_p): Pass 0 to it.
33994 * cselib.c (cselib_hasher::equal): Likewise.
33995 (rtx_equal_for_cselib_1): Add depth argument. If depth
33996 is 128, don't look up VALUE locs and punt. Increment
33997 depth in recursive calls when walking VALUE locs.
33999 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34001 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34002 (make_gcov_file_name): Use the canonical path name for generating
34004 (read_line): Fix handling of files with ascii null bytes.
34006 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34008 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34009 to initialise a vector register instead
34010 of using a const_int.
34012 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34014 PR translation/80189
34015 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34016 diagnostic messages.
34018 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34021 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34022 (dfp_diex_<mode>): Update mode of operand 1.
34023 * doc/extend.texi (dxex, dxexq): Document change to return type.
34024 (diex, diexq): Document change to argument type.
34026 2017-03-30 Martin Jambor <mjambor@suse.cz>
34029 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34030 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34031 it reflects the signature changes performed at the callee side.
34032 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34033 to cgraph_build_function_type_skip_args.
34034 (build_function_decl_skip_args): Adjust call to the above function.
34036 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34039 * config/i386/sse.md
34040 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34041 register as dest whenever it is a MEM not rtx_equal_p to the
34042 corresponding dup operand, and when forcing into reg move the
34043 reg into the memory afterwards.
34044 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34045 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34046 for the force_reg mode.
34047 (avx512vl_vextractf128<mode>): Use register as dest either
34048 always when a MEM, or when it is a MEM not rtx_equal_p to the
34049 corresponding dup operand, or even not when it is a CONST_VECTOR
34050 depending on the mode and lo vs. hi.
34051 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34053 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34054 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34055 Likewise. Require that operands[2] is even.
34056 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34057 Remove extraneous parens. Require that operands[2] is a multiple
34059 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34060 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34061 disallow memory then.
34063 2017-03-30 Richard Biener <rguenther@suse.de>
34065 PR tree-optimization/77498
34066 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34067 to non-constants over backedges.
34069 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34071 PR rtl-optimization/80233
34072 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34073 as last_combined_insn. Do not test for BARRIER_P separately.
34075 2017-03-29 Andreas Schwab <schwab@suse.de>
34078 * calls.c (prepare_call_address): Convert funexp to Pmode before
34079 copying to temp reg.
34081 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34083 PR tree-optimization/80158
34084 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34085 Handle possible future case of more than one alternate
34087 (replace_rhs_if_not_dup): Likewise.
34088 (replace_one_candidate): Likewise.
34090 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34092 PR rtl-optimization/80193
34093 * ira.c (ira): Do not check allocation for LRA.
34095 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34097 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34098 (nvptx_output_simt_exit): Declare.
34099 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34100 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34101 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34102 (nvptx_declare_function_name): ...here. Emit declaration of local
34103 memory space buffer for omp_simt_enter insn.
34104 (nvptx_output_unisimt_switch): New.
34105 (nvptx_output_softstack_switch): New.
34106 (nvptx_output_simt_enter): New.
34107 (nvptx_output_simt_exit): New.
34108 * config/nvptx/nvptx.h (struct machine_function): New fields
34109 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34110 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34111 (UNSPECV_SIMT_EXIT): Ditto.
34112 (omp_simt_enter_insn): New insn.
34113 (omp_simt_enter): New expansion.
34114 (omp_simt_exit): New insn.
34115 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34117 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34118 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34119 (expand_GOMP_SIMT_EXIT): New.
34120 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34121 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34122 (GOMP_SIMT_EXIT): Ditto.
34123 * target-insns.def (omp_simt_enter): New insn.
34124 (omp_simt_exit): Ditto.
34125 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34127 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34128 (lower_rec_input_clauses): Likewise.
34129 (lower_lastprivate_clauses): Handle SIMT privatization.
34131 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34132 (ompdevlow_adjust_simt_enter): New.
34133 (find_simtpriv_var_op): New.
34134 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34135 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34137 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34138 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34139 (copy_decl_for_dup_finish): Ditto.
34141 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34143 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34146 * config/i386/i386.c (ix86_option_override_internal): Always
34147 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34149 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34151 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34153 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34155 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34156 mark new edge's irreducible flag accordign to it.
34157 (vect_do_peeling): Check loop preheader edge's irreducible flag
34158 and pass it to function slpeel_add_loop_guard.
34160 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34162 PR tree-optimization/80218
34163 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34164 Update block frequencies and counts.
34166 2017-03-28 Richard Biener <rguenther@suse.de>
34168 PR tree-optimization/78644
34169 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34170 of a simplification result we may not use it at all.
34172 2017-03-28 Richard Biener <rguenther@suse.de>
34175 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34176 without arguments, generate default definition of a SSA name.
34178 2017-03-28 Richard Biener <rguenther@suse.de>
34180 PR middle-end/80222
34181 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34182 TYPE_REF_CAN_ALIAS_ALL references.
34183 * fold-const.c (fold_indirect_ref_1): Likewise.
34185 2017-03-28 Martin Liska <mliska@suse.cz>
34188 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34189 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34191 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34192 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34194 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34195 (EXTRA_SPECS): Define.
34196 (SUBTARGET_EXTRA_SPECS): Likewise.
34197 (SUBTARGET_CPP_SPEC): Likewise.
34198 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34199 SUBTARGET_EXTRA_SPECS.
34200 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34202 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34204 * config/arc/simdext.md (vst64_insn): Update pattern.
34205 (vld32wh_insn): Likewise.
34206 (vld32wl_insn): Likewise.
34207 (vld64_insn): Likewise.
34208 (vld32_insn): Likewise.
34210 2017-03-28 Marek Polacek <polacek@redhat.com>
34213 * fold-const.c (fold_comparison): Use protected_set_expr_location
34214 instead of SET_EXPR_LOCATION.
34216 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34218 * tree.c (add_expr): Avoid name lookup warning.
34220 2017-03-27 Jeff Law <law@redhat.com>
34222 PR tree-optimization/80216
34223 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34224 function name. Limit recursion depth.
34225 (record_temporary_equivalences): Corresponding changes.
34227 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34229 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34232 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34235 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34237 * cfgcleanup.c (reg_note_cfa_p): New array.
34238 (insns_have_identical_cfa_notes): New function.
34239 (old_insns_match_p): Don't cross-jump in between /f
34240 and non-/f instructions. If both i1 and i2 are frame related,
34241 verify all CFA notes, their order and content.
34243 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34246 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34247 HImode and SImode with zero extend to DImode to one insn.
34248 (bswap<mode>2_extenddi): Likewise.
34249 (bswapsi2_extenddi): Likewise.
34250 (bswaphi2_extendsi): Likewise.
34251 (bswaphi2): Combine bswap HImode and SImode into one insn.
34252 Separate memory insns from swapping register.
34253 (bswapsi2): Likewise.
34254 (bswap<mode>2): Likewise.
34255 (bswaphi2_internal): Delete, no longer used.
34256 (bswapsi2_internal): Likewise.
34257 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34258 store, and gpr<-gpr swap insns.
34259 (bswap<mode>2_store): Likewise.
34260 (bswaphi2_reg): Register only splitter, combine with the splitter.
34261 (bswaphi2 splitter): Likewise.
34262 (bswapsi2_reg): Likewise.
34263 (bswapsi2 splitter): Likewise.
34264 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34265 the insns into load, store, and register/register insns.
34266 (bswapdi2_ldbrx): Likewise.
34267 (bswapdi2_load): Likewise.
34268 (bswapdi2_store): Likewise.
34269 (bswapdi2_reg): Likewise.
34271 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34273 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34274 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34276 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34279 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34282 special handling for target option conflicts between dform
34283 options (-mpower9-dform, -mpower9-dform-vector,
34284 -mpower9-dform-scalar) and -mno-direct-move.
34286 2017-03-27 Richard Biener <rguenther@suse.de>
34288 PR tree-optimization/80181
34289 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34291 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34293 * config/arc/predicates.md (move_double_src_operand): Replace the
34294 call to move_double_src_operand with a call to address_operand.
34296 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34298 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34299 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34300 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34302 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34304 * config/arc/predicates.md (long_immediate_loadstore_operand):
34305 Consider scaled addresses cases.
34307 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34309 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34310 restored when in interrupt.
34311 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34312 doesn't have delay slot.
34314 2017-03-27 Richard Biener <rguenther@suse.de>
34317 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34318 inlined thunk clones.
34320 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34323 * asan.c (instrument_derefs): Copy over last operand from
34324 original COMPONENT_REF to the new COMPONENT_REF with
34325 DECL_BIT_FIELD_REPRESENTATIVE.
34326 * ubsan.c (instrument_object_size): Likewise.
34328 2017-03-27 Richard Biener <rguenther@suse.de>
34330 PR tree-optimization/80170
34331 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34332 sure DR/SCEV didnt fold in constants we do not see when looking
34333 at the reference base alignment.
34335 2017-03-27 Richard Biener <rguenther@suse.de>
34337 PR middle-end/80171
34338 * gimple-fold.c (fold_ctor_reference): Properly guard against
34339 NULL return value from canonicalize_constructor_val.
34341 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34344 * config/i386/i386.c (ix86_expand_builtin)
34345 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34346 flags reg setting and flags reg using instructions.
34347 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34348 clobbering instructions to zero extend op2.
34350 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34352 * doc/install.texi (Configuration) <--with-aix-soname>:
34353 Update link to AIX ld.
34355 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34357 PR rtl-optimization/80160
34358 PR rtl-optimization/80159
34359 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34360 reg_alternate_class into account.
34362 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34365 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34366 to consider in curr_insn_transform.
34368 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34370 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34371 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34372 and emit_mode_inner.
34374 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34376 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34377 argument to the overloaded builtin variants. Use the new flag to
34378 deprecate certain builtin variants.
34379 * config/s390/s390-builtin-types.def: Add new builtin types.
34380 * config/s390/s390-builtins.h: Support new flags field for
34381 overloaded builtins.
34382 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34383 (s390_macro_to_expand): Enable vector float data type.
34384 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34385 builtins by incrementing the __VEC__ version number.
34386 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34388 (s390_resolve_overloaded_builtin): Emit error messages depending
34389 on the builtin flags.
34390 * config/s390/s390.c (s390_expand_builtin): Support additional
34391 flags argument. Change error message to match the messages
34392 emitted in s390-c.c.
34393 * config/s390/s390.md: New UNSPEC_* constants.
34394 (op_type): Add new instruction types.
34395 * config/s390/vecintrin.h: Add new builtins and test data class
34397 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34398 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34399 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34400 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34401 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34402 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34404 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34405 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34406 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34407 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34409 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34410 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34411 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34412 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34414 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34415 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34416 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34417 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34418 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34419 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34420 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34422 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34423 ("vec_scatter_element<V_HW_4:mode>_DI")
34424 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34425 ("vec_fpint<mode>", "vflls")
34426 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34427 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34428 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34429 ("*vec_cmphe<mode>_cc"): ... these.
34431 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34432 mode constant instead of magic value.
34434 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34436 * config/s390/s390.c (s390_expand_vec_compare): Support other
34437 vector floating point modes than just V2DF.
34438 (s390_expand_vcond): Likewise.
34439 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34440 (s390_cannot_change_mode_class): Prevent mode changes between TF
34441 and V1TF in vector registers.
34442 * config/s390/s390.md (DF, SF): New mode attributes.
34443 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34444 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34445 SFmode support for VRs.
34446 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34448 (VFT, VF_HW): New mode iterators.
34449 (vw, sdx): New mode attributes.
34450 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34451 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34452 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34453 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34454 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34455 also the new vector floating point modes. Renaming to ...
34457 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34458 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34459 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34460 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34461 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34462 ("vec_unordered<mode>"): ... these.
34464 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34465 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34466 ("*vec_extendv2df"): New insn definitions.
34468 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34470 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34471 ("mulditi3_2", "*muldi3_sign"): New patterns.
34472 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34473 rename the pattern definition.
34475 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34477 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34479 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34481 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34483 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34484 instruction if possible.
34485 * config/s390/vector.md (vec_halfnumelts): New mode
34487 ("*vec_vllezlf<mode>"): New pattern.
34489 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34491 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34492 ("popcountv4si2", "popcountv2di2"): Rename to ...
34493 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34494 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34496 ("popcount<mode>2_vxe"): New pattern.
34498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34500 * common/config/s390/s390-common.c (processor_flags_table): Add
34502 * config.gcc: Add arch12.
34503 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34504 Default to arch12 for unknown CPU model numbers.
34505 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34506 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34507 PROCESSOR_max sanity check.
34508 * config/s390/s390-opts.h (enum processor_type): Add
34510 * config/s390/s390.c (processor_table): Add arch12.
34511 (s390_expand_builtin): Add check for B_VXE flag.
34512 (s390_issue_rate): Add PROCESSOR_ARCH12.
34513 (s390_get_sched_attrmask): Likewise.
34514 (s390_get_unit_mask): Likewise.
34515 (s390_sched_score): Enable z13 scheduling for arch12.
34516 (s390_sched_reorder): Likewise.
34517 (s390_sched_variable_issue): Likewise.
34518 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34520 (s390_tune_attr): Use z13 scheduling also for arch12.
34521 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34522 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34523 (TARGET_VXE_P): New macros.
34524 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34525 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34526 * config/s390/s390.opt: Add arch12 as processor_type.
34528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34530 * config/s390/s390.md
34531 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34532 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34533 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34535 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34536 Rename expanders to ...
34538 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34539 ("fixuns_truncdddi2_emu"): ... these.
34541 ("fixuns_trunc<mode>si2_emu"): New expander.
34543 ("*fixuns_truncdfdi2_z13"): Rename to ...
34544 ("*fixuns_truncdfdi2_vx"): ... this.
34546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34548 * config/s390/2964.md: Remove the single element vector compare
34549 instructions which are no longer used.
34550 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34552 (s390_canonicalize_comparison): Remove handling of DFmode
34554 (s390_expand_vec_compare_scalar): Remove function.
34555 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34556 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34558 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34560 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34562 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34564 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34565 will anyway by matched by mov<mode>_64dfp.
34567 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34569 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34570 vlef/vstef. Add missing operand to vleif.
34572 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34574 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34575 pair for all vector types with 64 bit elements.
34576 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34577 * config/s390/vector.md (V_HW_64): ... here.
34578 (V_128_NOSINGLE): New mode iterator.
34579 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34580 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34581 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34582 ("*vec_load_pairv2di"): Change to ...
34583 ("*vec_load_pair<mode>"): ... this one.
34585 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34587 * config/s390/constraints.md: Add comments.
34588 (jKK): Reject element sizes > 8 bytes.
34589 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34591 * config/s390/s390.md: Add the s_operand checks formerly in
34592 s390_split_ok_p to various splitters where they are still
34594 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34595 for 128 bit vectors. Plus two splitters.
34597 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34599 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34602 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34605 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34606 error if the boundary argument is not constant.
34608 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34610 PR rtl-optimization/80112
34611 * loop-doloop.c (doloop_condition_get): Don't check condition
34612 if cmp isn't SET with IF_THEN_ELSE src.
34614 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34616 PR tree-optimization/80158
34617 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34618 replacing a candidate statement, also replace it for the
34619 candidate's alternate interpretation.
34620 (replace_rhs_if_not_dup): Likewise.
34621 (replace_one_candidate): Likewise.
34623 2017-03-24 Richard Biener <rguenther@suse.de>
34625 PR tree-optimization/80167
34626 * graphite-isl-ast-to-gimple.c
34627 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34629 (translate_isl_ast_to_gimple::get_rename): Likewise.
34631 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34634 handling of certain combinations of target options, including the
34635 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34636 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34638 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34641 * config/arm/arm.md (*load_multiple): Add reload_completed to
34642 matching condition.
34644 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34645 Richard Biener <rguenth@suse.de>
34647 PR tree-optimization/79908
34648 PR tree-optimization/80136
34649 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34650 been cast away, gimplify_and_add suffices.
34652 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34654 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34656 2017-03-23 Richard Biener <rguenther@suse.de>
34658 PR tree-optimization/80032
34659 * gimplify.c (gimple_push_cleanup): Forced unconditional
34660 cleanups still have to go to the conditional_cleanups
34663 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34665 PR tree-optimization/80072
34666 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34668 (next_operand_entry_id): Change type to unsigned int.
34669 (sort_by_operand_rank): Make sure to return the right return value
34670 even if unsigned fields are bigger than INT_MAX.
34671 (struct oecount): Change cnt and id type to unsigned int.
34672 (oecount_hasher::equal): Formatting fix.
34673 (oecount_cmp): Make sure to return the right return value
34674 even if unsigned fields are bigger than INT_MAX.
34675 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34678 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34679 TREE_READONLY on result if writing it more than once.
34682 * doc/invoke.texi (-fsanitize=thread): Document that with
34683 -fnon-call-exceptions atomics are not able to throw
34687 * tsan.c: Include tree-eh.h.
34688 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34689 maybe_clean_or_replace_eh_stmt where needed.
34690 (instrument_memory_accesses): Add cfg_changed argument.
34691 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34692 if it returned true.
34693 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34695 PR rtl-optimization/63191
34696 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34697 wrapper function, moved the whole old content into ...
34698 (ix86_delegitimize_address_1): ... this. New inline function.
34699 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34700 true as last argument instead of ix86_delegitimize_address.
34702 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34704 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34705 cortexa57_branch_cost.
34707 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34709 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34711 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34714 * doc/md.texi (Constraints): Document wA constraint.
34715 * config/rs6000/constraints.md (wA): New.
34716 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34717 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34718 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34719 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34721 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34724 * gimplify.c (is_oacc_declared): New function.
34725 (oacc_default_clause): Use it to set default flags for acc declared
34726 variables inside parallel regions.
34727 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34728 declared variables.
34729 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34730 declare attribute to any decl as necessary.
34732 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34735 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34736 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34737 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34738 (arm_arch_lpae): This.
34739 * config/arm/arm.c (arm_arch7ve): Rename into ...
34740 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34741 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34744 2017-03-22 Martin Liska <mliska@suse.cz>
34747 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34748 error message instead of an ICE.
34750 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34752 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34754 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34756 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34761 2017-03-21 Martin Sebor <msebor@redhat.com>
34763 * doc/extend.texi: Use "cannot" instead of "can't."
34764 * doc/hostconfig.texi: Same.
34765 * doc/install.texi: Same.
34766 * doc/invoke.texi: Same.
34767 * doc/loop.texi: Same.
34768 * doc/md.texi: Same.
34769 * doc/objc.texi: Same.
34770 * doc/rtl.texi: Same.
34771 * doc/tm.texi: Same.
34772 * doc/tm.texi.in: Same.
34773 * doc/trouble.texi: Same.
34775 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34778 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34779 (collect_checksum_attributes): Set it.
34780 (die_checksum_ordered): Use it.
34782 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34784 PR tree-optimization/79908
34785 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34786 change: For a VA_ARG whose LHS has been cast away, use
34787 force_gimple_operand to construct the side effects.
34789 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34791 PR translation/80001
34792 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34793 more amenable to translation.
34794 (oacc_loop_auto_partitions): Likewise.
34796 2017-03-21 Marek Polacek <polacek@redhat.com>
34797 Martin Sebor <msebor@redhat.com>
34799 PR tree-optimization/80109
34800 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34801 on INTEGRAL_TYPE_P.
34803 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34804 Segher Boessenkool <segher@kernel.crashing.org>
34807 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34808 check reg_used_between_p between insn and one of succ or succ2
34809 depending on if succ is artificial insn not inserted into insn
34812 2017-03-21 Martin Liska <mliska@suse.cz>
34814 PR gcov-profile/80081
34815 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34816 * doc/gcc.texi: Include gcov-dump stuff.
34817 * doc/gcov-dump.texi: New file.
34819 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34821 PR rtl-optimization/79150
34822 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34823 conditional jump, if the jump is the last insn of the loop.
34825 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34826 Richard Biener <rguenth@suse.de>
34828 PR tree-optimization/79908
34829 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34830 been cast away, use force_gimple_operand to construct the side
34833 2017-03-21 Martin Liska <mliska@suse.cz>
34835 PR libfortran/79956
34836 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34839 2017-03-21 Brad Spengler <spender@grsecurity.net>
34842 * plugin.c (htab_hash_plugin): New function.
34843 (add_new_plugin): Use it and adjust.
34844 (parse_plugin_arg_opt): Adjust.
34845 (init_one_plugin): Likewise.
34847 2017-03-21 Richard Biener <rguenther@suse.de>
34849 PR tree-optimization/80032
34850 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34851 if set force the cleanup to happen unconditionally.
34852 (gimplify_target_expr): Push inserted clobbers with force_uncond
34853 to avoid them being removed by control-dependent DCE.
34855 2017-03-21 Richard Biener <rguenther@suse.de>
34857 PR tree-optimization/80122
34858 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34859 va_arg_pack_len when the inlined call stmt requires pack
34861 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34863 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34866 * tsan.c (instrument_builtin_call): If the memory model argument
34867 is not a constant, assume it is valid.
34870 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34873 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34875 PR rtl-optimization/79910
34876 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34877 if its dest is used by an insn before I2 (other than the combined
34878 insns themselves, which are properly handled already).
34880 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34883 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34885 * combine.c (record_used_regs): New static function.
34886 (try_combine): Handle situations where there is an additional
34887 instruction between I2 and I3 which needs to have a LOG_LINK
34891 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34893 * combine.c (try_combine): Delete redundant i1 test. Call
34894 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34896 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34899 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34900 alternatives 13/14.
34902 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34904 PR tree-optimization/80054
34905 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34906 the optimization if a PHI or any of its arguments is not dominated
34907 by the candidate's basis. Use gphi* rather than gimple* as
34909 (replace_profitable_candidates): Clean up a gimple* variable that
34910 should be a gphi* variable.
34912 2017-03-20 Martin Sebor <msebor@redhat.com>
34915 * doc/extend.texi (attribute constructor): Document present limitation.
34917 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34920 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34921 __POWER9_VECTOR__ #ifdef control, change template definition to
34922 use Power9-specific built-in function.
34923 (vec_any_eq): Likewise.
34924 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34925 to control outcomes from this test.
34926 (vector_ae_<mode>p): For VEC_F modes, likewise.
34928 2017-03-20 Ian Lance Taylor <iant@google.com>
34930 * config/i386/i386.c (ix86_function_regparm): Save an extra
34931 register for -fsplit-stack with DECL_STATIC_CHAIN.
34933 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34936 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34937 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34939 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34941 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34943 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34946 2017-03-20 Marek Polacek <polacek@redhat.com>
34949 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34951 2017-03-20 Richard Biener <rguenther@suse.de>
34953 PR tree-optimization/80113
34954 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34955 allocate extra SSA name for PHI def.
34956 (add_close_phis_to_outer_loops): Likewise.
34957 (add_close_phis_to_merge_points): Likewise.
34958 (copy_loop_close_phi_args): Likewise.
34959 (copy_cond_phi_nodes): Likewise.
34961 2017-03-20 Martin Liska <mliska@suse.cz>
34963 PR middle-end/79753
34964 * tree-chkp.c (chkp_build_returned_bound): Do not build
34965 returned bounds for a LHS that's not a BOUNDED_P type.
34967 2017-03-20 Martin Liska <mliska@suse.cz>
34971 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34972 COMPLEX_CST and VECTOR_CST.
34974 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34977 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34978 target operand. A new splitter adds the clobber statement in case
34979 the target operand is dead anyway.
34981 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34983 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34984 to age-old versions of binutils and glibc.
34986 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34988 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34990 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34992 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34994 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34996 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34997 requirement for binutils 2.13.
34999 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35001 * combine.c (try_combine): Delete redundant i1 test. Call
35002 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35004 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35006 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35007 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35009 <riscv64-*-elf>: Re-arrange section
35010 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35011 <riscv32-*-linux>: Likewise.
35012 <riscv64-*-elf>: Likewise
35013 <riscv64-*-linux>: Likewise.
35015 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35018 * aarch64.opt(verbose-cost-dump): Fix typo.
35020 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35023 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35024 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35026 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35028 * reload.c (find_reloads): When reloading a nonoffsettable address,
35029 use RELOAD_OTHER for it and its address reloads.
35031 PR rtl-optimization/79910
35032 * combine.c (record_used_regs): New static function.
35033 (try_combine): Handle situations where there is an additional
35034 instruction between I2 and I3 which needs to have a LOG_LINK
35037 2017-03-17 Jeff Law <law@redhat.com>
35039 PR tree-optimization/71437
35040 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35041 conditional in the hash table first.
35042 (vrp_dom_walker::before_dom_children): Extract condition from
35043 ASSERT_EXPR. Record condition, its inverion and any implied
35044 conditions as well.
35046 2017-03-17 Marek Polacek <polacek@redhat.com>
35047 Markus Trippelsdorf <markus@trippelsdorf.de>
35049 PR tree-optimization/80079
35050 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35053 2017-03-17 Richard Biener <rguenther@suse.de>
35055 PR middle-end/80075
35056 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35057 Properly verify the LHS before the RHS possibly claims to be
35059 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35062 2017-03-17 Martin Jambor <mjambor@suse.cz>
35064 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35065 (List of -O2 options): Likewise.
35066 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35069 2017-03-17 Tom de Vries <tom@codesourcery.com>
35071 * gcov-dump.c (print_usage): Print bug_report_url.
35073 2017-03-17 Richard Biener <rguenther@suse.de>
35075 PR middle-end/80050
35076 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35077 (parser::peek): Likewise.
35079 2017-03-17 Richard Biener <rguenther@suse.de>
35081 PR tree-optimization/80048
35082 * sese.c (free_sese_info): Properly release rename_map and
35083 copied_bb_map elements.
35085 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35087 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35088 Add linked-list forward and backlinks. Insert on
35089 construction, remove on destruction.
35090 (class pass_store_merging): Add m_stores_head field.
35091 (pass_store_merging::terminate_and_process_all_chains):
35092 Iterate over m_stores_head list.
35093 (pass_store_merging::terminate_all_aliasing_chains):
35095 (pass_store_merging::execute): Check for debug stmts first.
35096 Push new chains onto the m_stores_head stack.
35098 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35101 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35102 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35103 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35105 2017-03-16 Jeff Law <law@redhat.com>
35107 PR tree-optimization/71437
35108 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35109 member function. Implementation moved into after_dom_children
35110 member function and into the threader's thread_outgoing_edges
35112 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35113 some code into new thread_outgoing_edges.
35114 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35115 definition. Simplify marker handling (do it here). Assume we always
35116 have the available expression and the const/copies tables.
35117 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35119 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35120 * tree-vrp.c (equiv_stack): No longer file scoped.
35121 (vrp_dom_walker): New class.
35122 (vrp_dom_walker::before_dom_children): New member function.
35123 (vrp_dom_walker::after_dom_children): Likewise.
35124 (identify_jump_threads): Setup domwalker. Use it rather than
35125 walking edges in a random order by hand. Simplify setup/finalization.
35126 (finalize_jump_threads): Remove.
35127 (vrp_finalize): Do not call identify_jump_threads here.
35128 (execute_vrp): Do it here instead and call thread_through_all_blocks
35131 PR tree-optimization/71437
35132 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35134 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35136 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35137 (dom_opt_dom_walker::thread_across_edge): Remove
35138 handle_dominating_asserts argument. All callers changed.
35139 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35140 changes. Remove calls to lhs_of_dominating_assert. Other
35141 uses of handle_dominating_asserts turn into unconditional code
35142 (simplify_control_stmt_condition_1): Likewise.
35143 (simplify_control_stmt_condition): Likewise.
35144 (thread_through_normal_block, thread_across_edge): Likewise.
35145 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35146 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35147 object if it is not an SSA_NAME.
35148 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35149 before calling into the VRP specific simplifiers.
35150 (identify_jump_threads): Remove handle_dominating_asserts
35153 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35156 * tree-diagnostic.c (default_tree_printer): No longer static.
35157 * tree-diagnostic.h (default_tree_printer): New prototype.
35159 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35161 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35162 Change ins into fmov.
35164 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35166 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35167 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35168 Use h_con constraint for operand 1.
35169 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35170 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35172 2017-03-15 Jeff Law <law@redhat.com>
35174 PR tree-optimization/71437
35175 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35176 (record_temporary_equivalences): Use it.
35178 PR tree-optimization/71437
35179 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35180 tree-ssa-scopedtables.
35181 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35182 (record_conditions, record_cond, vuse_eq): Likewise.
35183 (record_edge_info): Adjust to API tweak of record_conditions.
35184 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35185 (record_temporary_equivalences, optimize_stmt): Likewise.
35186 (eliminate_redundant_computations): Likewise.
35187 (record_equivalences_from_stmt): Likewise.
35188 * tree-ssa-scopedtables.c: Include options.h and params.h.
35189 (vuse_eq): New function, moved from tree-ssa-dom.c
35190 (build_and_record_new_cond): Likewise.
35191 (record_conditions): Likewise. Accept vector of conditions rather
35192 than edge_equivalence structure for first argument.
35193 for the first argument.
35194 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35195 from tree-ssa-dom.c.
35196 (avail_exprs_stack::record_cond): Likewise.
35197 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35198 from tree-ssa-dom.c.
35199 (avail_exprs_stack): Add new member functions lookup_avail_expr
35201 (record_conditions): Declare.
35203 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35206 * lra-constraints.c (process_alt_operands): Increase reject for
35207 reloading an input/output operand.
35209 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35212 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35213 insns to convert from signed/unsigned char/short to IEEE 128-bit
35215 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35217 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35220 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35221 subreg of inner mode for values already in registers.
35223 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35225 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35226 iteration reg is used after the loop.
35228 2017-03-14 Martin Sebor <msebor@redhat.com>
35230 PR tree-optimization/79800
35231 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35232 precision in negative-positive range.
35233 (format_floating): Call non-const overload with adjusted precision.
35235 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35238 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35241 2017-03-14 Martin Sebor <msebor@redhat.com>
35243 PR middle-end/80020
35244 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35245 * builtins.def (aligned_alloc): Use it.
35248 * Makefile.in (GTFILES): Add calls.c.
35249 * calls.c: Include "gt-calls.h".
35251 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35253 PR rtl-optimization/79728
35254 * regs.h (struct target_regs): New field
35255 x_contains_allocatable_regs_of_mode.
35256 (contains_allocatable_regs_of_mode): New macro.
35257 * reginfo.c (init_reg_sets_1): Initialize it, and change
35258 contains_reg_of_mode so it includes global regs as well.
35259 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35260 rather than contains_regs_of_mode.
35262 2017-03-14 Martin Liska <mliska@suse.cz>
35264 * doc/invoke.texi: Document options that can't be combined with
35265 -fcheck-pointer-bounds.
35267 2017-03-14 Martin Liska <mliska@suse.cz>
35269 PR middle-end/79831
35270 * doc/invoke.texi (-Wchkp): Document the option.
35272 2017-03-14 Martin Liska <mliska@suse.cz>
35274 * Makefile.in: Install gcov-dump.
35276 2017-03-14 Martin Liska <mliska@suse.cz>
35278 * multiple_target.c (expand_target_clones): Bail out for
35279 an invalid attribute.
35281 2017-03-14 Richard Biener <rguenther@suse.de>
35283 * alias.c (struct alias_set_entry): Pack properly.
35284 * cfgloop.h (struct loop): Likewise.
35285 * cse.c (struct set): Likewise.
35286 * ipa-utils.c (struct searchc_env): Likewise.
35287 * loop-invariant.c (struct invariant): Likewise.
35288 * lra-remat.c (struct cand): Likewise.
35289 * recog.c (struct change_t): Likewise.
35290 * rtl.h (struct address_info): Likewise.
35291 * symbol-summary.h (function_summary): Likewise.
35292 * tree-loop-distribution.c (struct partition): Likewise.
35293 * tree-object-size.c (struct object_size_info): Likewise.
35294 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35295 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35296 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35297 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35298 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35299 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35300 (struct _stmt_vec_info): Likewise.
35302 2017-03-14 Martin Liska <mliska@suse.cz>
35305 * multiple_target.c (create_dispatcher_calls): Check that
35306 a target can create a function dispatcher.
35308 2017-03-14 Martin Liska <mliska@suse.cz>
35311 * multiple_target.c (expand_target_clones): Drop local.local
35312 flag for default implementation.
35314 2017-03-14 Richard Biener <rguenther@suse.de>
35316 PR tree-optimization/80030
35317 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35319 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35321 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35322 gcc_fallthrough() instead of __attribute__((fallthrough));
35324 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35326 * doc/gcc.texi: Remove "up" link to (DIR).
35327 * doc/gccint.texi: Ditto.
35329 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35331 * doc/install.texi (Specific) <avr>: Remove reference to
35334 2017-03-13 Jeff Law <law@redhat.com>
35336 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35337 attribute rather than comments.
35339 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35340 match_scratch operand is highest.
35342 2017-03-13 Martin Liska <mliska@suse.cz>
35344 PR middle-end/78339
35345 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35346 is a CHKP clone, use original declaration.
35348 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35350 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35351 (arc_conditional_register_usage): Use a different allocation order
35352 when optimizing for size.
35353 * common/config/arc/arc-common.c (arc_option_optimization_table):
35354 Section anchors default on when optimizing for size.
35356 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35358 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35360 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35362 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35363 * config/arc/arc.md (cpu_facility): Add cd variant.
35364 (*movqi_insn): Add code density variant.
35365 (*movhi_insn): Likewise.
35366 (*movqi_insn): Likewise.
35367 (*addsi3_mixed): Likewise.
35368 (subsi3_insn): Likewise.
35370 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35372 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35374 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35376 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35377 expressions with MINUS and UNARY ops.
35379 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35382 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35384 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35385 between vec_select and vector argument.
35386 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35387 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35388 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35389 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35390 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35391 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35393 2017-03-13 Richard Biener <rguenther@suse.de>
35396 * params.def (vect-max-peeling-for-alignment): Fix typo.
35398 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35400 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35401 issue that only occurred with binutils below 2.18.
35403 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35405 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35406 refer to binutils 2.11/2.12 minimum.
35408 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35410 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35411 ftp.kernel.org and simplify binutils requirement.
35413 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35415 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35417 (Optimize Options): Ditto. Also remove redundancy.
35419 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35421 PR translation/79848
35422 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35424 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35425 to G_ to avoid double translation.
35427 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35429 PR translation/79923
35430 * auto-profile.c (get_combined_location): Convert leading
35431 character of diagnostics to lower case and remove trailing period.
35432 (read_profile): Likewise for various diagnostics.
35433 * config/arm/arm.c (arm_option_override): Remove trailing period
35434 from various diagnostics.
35435 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35436 (msp430_expand_delay_cycles): Likewise.
35438 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35441 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35442 full command-line argument, rather than just "str".
35443 (aarch64_validate_march): Likewise.
35444 (aarch64_validate_mtune): Likewise.
35446 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35448 PR rtl-optimization/78911
35449 * lra-assigns.c (must_not_spill_p): New function.
35450 (spill_for): Use it.
35452 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35454 PR tree-optimization/79981
35455 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35456 ATOMIC_COMPARE_EXCHANGE ifn result.
35457 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35458 IFN_ATOMIC_COMPARE_EXCHANGE.
35460 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35463 * opts.c (parse_sanitizer_options): Add missing question mark to
35464 "did you mean" message.
35466 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35468 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35470 (VMULEUH_UNS): Likewise.
35471 (VMULOUB_UNS): Likewise.
35472 (VMULOUH_UNS): Likewise.
35473 * config/rs6000/rs6000.c (builtin_function_type): Remove
35474 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35476 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35479 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35480 x with result of extra_parsing_for_operand_code_0.
35481 (function_reader::extra_parsing_for_operand_code_0): Convert
35482 return type from void to rtx, returning x. When reading
35483 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35484 larger size containing struct block_symbol.
35486 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35488 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35489 -mfloat128-hardware without -m64.
35491 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35494 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35495 entries to the case statement that marks unsigned arguments to
35496 overloaded functions.
35498 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35500 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35501 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35503 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35506 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35507 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35509 2017-03-10 Martin Liska <mliska@suse.cz>
35513 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35514 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35517 2017-03-10 Olivier Hainque <hainque@adacore.com>
35519 * tree-switch-conversion (array_value_type): Start by resetting
35520 candidate type to it's main variant.
35522 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35524 PR rtl-optimization/79909
35525 * combine.c (try_combine): Use simplify_replace_rtx on individual
35526 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35527 of the whole CALL_INSN_FUNCTION_USAGE.
35529 PR tree-optimization/79972
35530 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35531 get_range_info on SSA_NAMEs. Formatting fixes.
35533 2017-03-10 Richard Biener <rguenther@suse.de>
35534 Jakub Jelinek <jakub@redhat.com>
35536 PR tree-optimization/77975
35537 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35538 edge to be constant.
35539 (get_val_for): For constant x return it. Formatting fix.
35540 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35541 would use the same bases as the current one.
35543 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35545 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35546 instead of vec_select for V1TImode.
35547 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35549 (VSX_LE_128): Add V1TI to this mode iterator.
35550 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35551 (*vsx_le_perm_store_<mode>): Likewise.
35552 (pre-reload splitter for VSX stores): Likewise.
35553 (post-reload splitter for VSX stores): Likewise.
35554 (*vsx_xxpermdi2_le_<mode>): Likewise.
35555 (*vsx_lxvd2x2_le_<mode>): Likewise.
35556 (*vsx_stxvd2x2_le_<mode>): Likewise.
35558 2017-03-09 Michael Eager <eager@eagercon.com>
35560 Correct failures with --enable-checking=yes,rtl.
35562 * config/microblaze/microblaze.c (microblaze_expand_shift):
35563 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35564 test for const0_rtx.
35565 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35566 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35568 2017-03-09 Richard Biener <rguenther@suse.de>
35570 PR tree-optimization/79977
35571 * graphite-scop-detection.c (scop_detection::merge_sese):
35572 Handle the case of extra exits to blocks dominating the entry.
35574 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35576 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35579 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35581 PR rtl-optimization/79949
35582 * lra-constraints.c (process_alt_operands): Check memory when
35583 trying to predict a cycle. Print about the overall increase.
35585 2017-03-09 Richard Biener <rguenther@suse.de>
35587 PR middle-end/79971
35588 * gimple-expr.c (useless_type_conversion_p): Preserve
35589 TYPE_SATURATING for fixed-point types.
35591 2017-03-09 Richard Biener <rguenther@suse.de>
35594 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35595 alignment of BLKmode params.
35597 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35600 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35601 (VALL_NO_V2Q): Likewise.
35603 * config/aarch64/aarch64-simd.md
35604 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35606 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35607 VALL_NO_V2Q mode iterator.
35608 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35610 2017-03-09 Martin Liska <mliska@suse.cz>
35612 PR tree-optimization/79631
35613 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35614 tree_int_cst_sign_bit just for INTEGER constants.
35616 2017-03-09 Martin Liska <mliska@suse.cz>
35620 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35623 2017-03-09 Marek Polacek <polacek@redhat.com>
35626 * tree.c (inchash::add_expr): Handle TREE_VEC.
35628 2017-03-09 Martin Liska <mliska@suse.cz>
35631 (chkp_narrow_size_and_offset): New function.
35632 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35633 (void chkp_parse_bit_field_ref): New function.
35634 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35635 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35637 2017-03-09 Martin Liska <mliska@suse.cz>
35640 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35641 (chkp_find_bounds_1): Remove gcc_unreachable.
35643 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35646 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35647 BUILT_IN_SYNC*, determine the access type from the size suffix and
35648 always build a MEM_REF with that type. Handle forgotten
35649 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35652 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35653 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35654 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35655 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35656 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35657 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35658 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35659 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35660 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35661 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35662 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35663 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35664 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35665 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35666 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35667 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35668 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35669 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35670 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35671 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35672 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35673 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35674 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35675 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35676 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35677 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35678 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35679 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35680 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35681 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35682 definitions outside of __OPTIMIZE__ guarded section.
35685 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35686 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35687 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35688 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35691 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35693 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35694 ("vfenez<mode>"): Add missing constraints.
35696 2017-03-08 Martin Sebor <msebor@redhat.com>
35699 * config/nds32/nds32.c (nds32_option_override):
35700 Fix misspelled diagnostic.
35702 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35705 * gimplify.c (gimplify_omp_for): Replace index var in outer
35706 taskloop statement with an artificial variable and add
35707 OMP_CLAUSE_PRIVATE clause for it.
35709 2017-03-08 Richard Biener <rguenther@suse.de>
35711 PR tree-optimization/79955
35712 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35713 for accesses that are completely outside of the variable.
35715 2017-03-08 Andrew Haley <aph@redhat.com>
35717 PR tree-optimization/79943
35718 * tree-ssa-loop-split.c (compute_new_first_bound): When
35719 calculating the new upper bound, (END-BEG) should be added, not
35722 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35724 * config/avr/avr.md (setmemhi): Make sure match_dup
35725 operand number comes before match_scratch.
35727 2017-03-08 Richard Biener <rguenther@suse.de>
35729 PR tree-optimization/79920
35730 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35731 with ncopies == 1 to ...
35732 (vect_transform_slp_perm_load): ... here. Properly compute
35733 all element loads by iterating VF times over the group. Do
35734 not handle ncopies (computed in a broken way) in
35735 vect_create_mask_and_perm.
35737 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35740 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35741 is a uniform vector, use uniform_vector_p return value instead of
35742 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35744 2017-03-07 Marek Polacek <polacek@redhat.com>
35746 PR middle-end/79809
35747 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35748 (alloca_call_type): Likewise.
35750 2017-03-07 Martin Liska <mliska@suse.cz>
35752 * gcov.c (process_args): Put comment to correct location.
35754 2017-03-07 Martin Liska <mliska@suse.cz>
35756 PR middle-end/68270
35757 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35758 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35759 (chkp_narrow_bounds_for_field): Likewise.
35760 (chkp_parse_array_and_component_ref): Pass one more argument to
35763 2017-03-07 Richard Biener <rguenther@suse.de>
35765 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35768 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35770 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35771 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35773 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35776 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35777 to end of description.
35778 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35780 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35782 PR rtl-optimization/79901
35783 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35785 (*avx512f_<code><mode>3<mask_name>): ... this.
35786 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35787 iterator instead of VI8_AVX2_AVX512BW.
35789 PR rtl-optimization/79901
35790 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35791 min/max expander, expand it using expand_vec_cond_expr.
35794 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35797 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35800 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35801 to void * for PCH reasons.
35802 * dwarf2out.c (output_loc_operands, output_die): Cast
35803 v.val_vec.array to unsigned char *.
35805 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35808 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35811 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35813 PR rtl-optimization/79571
35814 * lra-constraints.c (process_alt_operands): Calculate static
35815 reject and subtract it from overall when only addresses will be
35818 2017-03-06 Julia Koval <julia.koval@intel.com>
35821 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35822 incoming stack boundary to 128 for 64-bit targets.
35824 2017-03-06 Richard Biener <rguenther@suse.de>
35826 PR tree-optimization/79894
35827 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35828 to NULL after folding it.
35830 2017-03-06 Richard Biener <rguenther@suse.de>
35832 PR tree-optimization/79824
35833 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35834 check disabling peeling for gaps.
35836 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35838 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35839 attributes): Document gettimeofday.
35841 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35843 * config/s390/s390.c (s390_option_override_internal): Set
35844 PARAM_MIN_VECT_LOOP_BOUND
35846 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35848 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35849 * config/s390/s390.md: Likewise.
35851 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35854 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35855 (<avx2_avx512>_perm<mode>): Rename to ...
35856 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35858 (<avx512>_perm<mode>_mask): Rename to ...
35859 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35861 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35862 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35863 instead of VI8F_256_512.
35864 (avx512f_perm<mode>): New define_expand.
35865 (avx512f_perm<mode>_mask): Likewise.
35866 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35867 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35869 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35871 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35872 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35874 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35876 2017-03-06 Martin Liska <mliska@suse.cz>
35879 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35880 when having a SSA NAME w/o VAR_DECL assigned to it.
35882 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35884 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35885 msa_dpsub_<su>_d): Fix MODE for vec_select.
35887 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35889 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35891 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35893 2017-03-06 Richard Biener <rguenther@suse.de>
35895 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35896 * plugin.c (register_plugin_info): Likewise.
35897 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35899 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35901 * config/i386/sse.md (sse_storehps, sse_storelps,
35902 avx_<castmode><avxsizesuffix>_<castmode>,
35903 avx512f_<castmode><avxsizesuffix>_<castmode>,
35904 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35905 in condition that at least one operand is not a MEM.
35907 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35909 PR middle-end/79805
35910 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35911 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35913 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35914 gimple_call_nothrow_p flag based on whether original builtin can throw.
35915 If it can, emit following stmts on the fallthrough edge.
35916 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35917 don't create new bb if inserting just debug stmts on the edge, try to
35918 insert them on the fallthru bb or just reset debug stmts.
35920 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35923 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35924 restore recog_data (including the operand rtxes inside it) around
35925 the call to get_insn_template.
35927 2017-03-03 Martin Sebor <msebor@redhat.com>
35929 PR tree-optimization/79699
35930 * context.c (context::~context): Free MPFR caches to avoid
35931 a memory leak on program exit.
35933 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35935 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35936 Use wide_int::ulow () instead of .elt (0).
35938 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35940 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35941 (*pushxf): Limit oF constraint to 32bit targets and add oC
35942 constraint for 64bit targets.
35943 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35944 (*pushdf): Change rmF constraint to rmC.
35946 2017-03-03 Martin Liska <mliska@suse.cz>
35948 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35949 Remove unused variable.
35951 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35954 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35955 is a memory operand, increase num_memory.
35956 (ix86_expand_args_builtin): Likewise.
35958 2017-03-03 Jan Hubicka <jh@suse.cz>
35961 * ipa-devirt.c (maybe_record_node): Properly handle
35962 __cxa_pure_virtual visibility.
35964 2017-03-03 Martin Liska <mliska@suse.cz>
35966 PR tree-optimization/79803
35967 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35969 (pass_loop_prefetch::execute): Disabled optimization if an
35970 assumption about L1 cache size is not met.
35972 2017-03-03 Martin Liska <mliska@suse.cz>
35974 PR rtl-optimization/79574
35975 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35976 (hash_scan_set): Likewise.
35977 (dump_hash_table): Likewise.
35978 (hoist_code): Likewise.
35980 2017-03-03 Richard Biener <rguenther@suse.de>
35982 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35983 in place of uhigh ().
35984 (fixed_convert_from_real): Likewise.
35986 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35989 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35991 2017-03-03 Richard Biener <rguenther@suse.de>
35993 PR middle-end/79818
35994 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35995 TYPE_OVERFLOW_UNDEFINED check.
35997 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35999 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36001 (vector_ae_<mode>_p): Likewise.
36002 (vector_nez_<mode>_p): Likewise.
36003 (vector_ne_v2di_p): Likewise.
36004 (vector_ae_v2di_p): Likewise.
36005 (vector_ne_<mode>_p): Likewise.
36006 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36008 (vsx_tsqrt<mode>2_fe): Likewise.
36010 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36013 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36015 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36017 PR rtl-optimization/79780
36018 * cprop.c (one_cprop_pass): When second and further conditional trap
36019 in a single basic block is turned into an unconditional trap, turn it
36020 into a deleted note to avoid RTL verification failures.
36022 2017-03-02 Richard Biener <rguenther@suse.de>
36024 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36026 2017-03-02 Richard Biener <rguenther@suse.de>
36028 PR tree-optimization/79345
36030 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36031 param and abort the walk, returning -1 if it is hit.
36032 (walk_aliased_vdefs): Take a limit param and pass it on.
36033 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36034 defaulting to 0 and return a signed int.
36035 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36036 (check_defs): New helper.
36037 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36038 about uninitialized memory.
36039 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36040 bogus uninitialized warning.
36041 (fixed_convert_from_real): Likewise.
36043 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36045 PR tree-optimization/66768
36046 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36047 iv_use if base object can't be determined.
36049 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36051 PR tree-optimization/79345
36052 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36053 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36054 (get_pattern_stats): Initialize it.
36055 * genemit.c (gen_expand): Verify match_scratch numbers come after
36056 match_operand/match_dup numbers.
36057 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36058 match_scratch numbers.
36059 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36061 * config/s390/s390.md (trunctdsd2): Likewise.
36063 2017-03-02 Richard Biener <rguenther@suse.de>
36065 * wide-int.h (wide_int_storage::operator=): Implement in terms
36068 2017-03-02 Richard Biener <rguenther@suse.de>
36070 PR tree-optimization/79777
36071 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36072 the to insert expression to sth existing.
36074 2017-03-01 Martin Sebor <msebor@redhat.com>
36076 PR middle-end/79692
36077 * gimple-ssa-sprintf.c
36078 (directive::known_width_and_precision): New function.
36079 (format_integer): Use it.
36080 (get_mpfr_format_length): Consider the full range of precision
36081 when computing %g output with the # flag. Set the likely byte
36082 count to 3 rather than 1 when precision is indeterminate.
36083 (format_floating): Correct the lower bound of precision.
36085 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36087 * doc/invoke.texi: Document default code model for 64-bit Linux.
36089 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36092 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36093 udiv rather than div since input pattern is unsigned.
36095 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36097 * config/i386/i386.c (print_reg): Warn for values of
36098 unsupported size in integer register.
36100 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36103 * config/rs6000/predicates.md (current_file_function_operand): Do
36104 not allow self calls to be local if the function is replaceable.
36106 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36109 * config/rs6000/altivec.h (vec_ctz and others): Change the
36110 preprocessor macro that controls conditional compilation from
36111 _ARCH_PWR9 to __POWER9_VECTOR__.
36112 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36113 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36114 control (instead of _ARCH_PWR9 control) so that template
36115 definition uses power9-specific function.
36116 (vec_any_eq): Likewise.
36117 (vec_all_ne): Change macro definition to use a power9-specific
36118 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36119 _ARCH_PWR9 control).
36120 (vec_any_eq) Likewise.
36121 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36122 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36123 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36124 support for xvcmpnedp instruction.
36125 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36126 macro expansion so that Power9 implementation of vec_all_ne does
36127 not use the AltiVec predicate framework.
36128 (VCMPNEH_P): Likewise.
36129 (VCMPNEW_P): Likewise.
36130 (VCMPNED_P): Likewise.
36131 (VCMPNEFP_P): Likewise.
36132 (VCMPNEDP_P): Likewise.
36133 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36134 implementation of vec_any_eq to not use AltiVec predicate
36136 (VCMPAEH_P): Likewise.
36137 (VCMPAEW_P): Likewise.
36138 (VCMPAED_P): Likewise.
36139 (VCMPAEFP_P): Likewise.
36140 (VCMPAEDP_P): Likewise.
36141 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36142 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36143 not use the AltiVec predicate framework.
36144 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36145 of vec_any_eq to not use AltiVec predicate framework.
36146 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36147 support for predefined __POWER9_VECTOR__ macro to indicate that
36148 Power9 instruction selection is enabled.
36149 (altivec_overloaded_builtins): Remove extraneous
36150 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36151 function argument types RS6000_BTI_bool_V16QI and
36152 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36153 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36154 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36155 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36156 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36157 Power9 for implementations of vec_cmpne. Change the signature for
36158 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36159 (representing vec_all_ne) to remove the previously described first
36160 argument of type RS6000_BTI_INTSI, as this was an artifact of
36161 reliance on the AltiVec predicate framework, which is no longer
36162 used in the implementation of these functions. Add
36163 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36164 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36165 since, unlike the AltiVec predicate framework implementation, we
36166 do not share function descriptors between vec_alle and vec_anyeq.
36167 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36168 set of modes that receive special treatment even when
36169 TARGET_P9_VECTOR is true. The special treatment emits code that
36170 does not depend on Power9 instructions.
36171 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36172 define_expand to not rely on AltiVec predicate framework.
36173 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36175 (vector_ne_v2di_p): Change this define_expand to not rely on
36176 AltiVec predicate framework.
36177 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36179 (vector_ne_<mode>_p): Change this define_expand to not rely on
36180 AltiVec predicate framework.
36181 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36183 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36184 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36185 define_insn pattern.
36186 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36187 define_insn pattern because the xvcmpne<VSs>. instruction is not
36189 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36190 instruction is not supported.
36192 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36194 * config/nvptx/nvptx.c: Include intl.h.
36196 2017-03-01 Martin Jambor <mjambor@suse.cz>
36199 * ipa-prop.h (ipa_bits): Removed field known.
36200 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36201 to pointers. Adjusted their comments to warn about their sharing.
36202 (ipcp_transformation_summary): Change bits to a vector of pointers.
36203 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36204 (ipa_get_ipa_bits_for_value): Declare.
36205 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36206 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36207 (ipa_bits_hash_table): Likewise.
36208 (ipa_vr_ggc_hash_traits): Likewise.
36209 (ipa_vr_hash_table): Likewise.
36210 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36211 being pointers and vr_known being removed.
36212 (ipa_set_jf_unknown): Likewise.
36213 (ipa_get_ipa_bits_for_value): New function.
36214 (ipa_set_jfunc_bits): Likewise.
36215 (ipa_get_value_range): New overloaded functions.
36216 (ipa_set_jfunc_vr): Likewise.
36217 (ipa_compute_jump_functions_for_edge): Use the above functions to
36218 construct bits and vr parts of jump functions.
36219 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36220 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36222 (ipcp_grow_transformations_if_necessary): Also allocate
36223 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36225 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36226 them. Fix too long lines.
36227 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36228 vr_known being removed.
36229 (ipa_read_jump_function): Use new setter functions to construct bits
36230 and vr parts of jump functions or set them to NULL.
36231 (write_ipcp_transformation_info): Adjust for bits being pointers.
36232 (read_ipcp_transformation_info): Likewise.
36233 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36235 Include gt-ipa-prop.h.
36236 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36238 (ipcp_store_bits_results): Likewise.
36239 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36240 Do not write to existing jump functions but use a temporary instead.
36242 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36245 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36246 attempt to use its first operand as BIT_FIELD_REF base.
36248 2017-03-01 Richard Biener <rguenther@suse.de>
36250 PR middle-end/79721
36251 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36252 interpolating formula in wrapping arithmetic.
36253 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36255 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36257 PR tree-optimization/79734
36258 * tree-vect-generic.c (expand_vector_condition): Optimize
36259 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36260 Handle VEC_COND_EXPR where comparison has different inner width from
36261 type's inner width.
36263 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36265 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36266 markup, and similar issues. Remove @opindex entries for things
36267 that aren't options. Add missing -mmpy-option entries.
36269 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36271 PR tree-optimization/79737
36272 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36273 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36274 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36275 instead of byte_size. Formatting fix.
36276 (shift_bytes_in_array_right): Formatting fix.
36278 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36281 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36282 condition on optimize for the leaf function test.
36284 2017-02-28 Martin Liska <mliska@suse.cz>
36287 * read-rtl-function.c (function_reader::handle_unknown_directive):
36288 Bail out when one uses -flto.
36290 2017-02-28 Martin Liska <mliska@suse.cz>
36292 * common.opt: Replace space with tabular for options of <number>
36294 * config/i386/i386.opt: Show <number> value for
36295 -mlarge-data-threshold.
36296 * opts.c (print_filtered_help): Do not display number in hexadecimal
36299 2017-02-28 Martin Liska <mliska@suse.cz>
36301 * common.opt: Fix --help=option -Q for options which are of
36304 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36306 * config/i386/i386.c (print_reg): Error out for values
36307 of 8-bit size in invalid integer register.
36309 2017-02-28 Martin Sebor <msebor@redhat.com>
36311 PR tree-optimization/79691
36312 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36314 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36317 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36318 gcc_unreachable with output_operand_lossage.
36320 2017-02-28 Richard Biener <rguenther@suse.de>
36322 PR tree-optimization/79740
36323 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36325 (visit_nary_op): Insert the nary into the hashtable if we
36326 pattern-matched sth.
36327 * tree-ssa-pre.c (eliminate_insert): Robustify.
36329 2017-02-28 Richard Biener <rguenther@suse.de>
36331 PR middle-end/79731
36332 * fold-const.c (decode_field_reference): Reject out-of-bound
36335 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36337 * config/i386/i386.c: Include intl.h.
36338 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36339 instead of just cond ? "..." : "...".
36340 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36341 * coverage.c (read_counts_file): Likewise.
36342 * omp-offload.c: Include intl.h.
36343 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36344 of just cond ? "..." : "...".
36345 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36346 of just cond ? "..." : "...".
36348 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36351 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36353 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36354 'tune for' CPU name.
36355 * config/arm/arm-cpu-data.h: Regenerated.
36357 2017-02-28 Richard Biener <rguenther@suse.de>
36359 PR tree-optimization/79732
36360 * tree-inline.c (expand_call_inline): Do not shadow var.
36362 2017-02-28 Richard Biener <rguenther@suse.de>
36364 PR tree-optimization/79723
36365 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36366 address-space properly.
36368 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36370 * doc/optinfo.texi (Optimization groups): Fix option used for
36372 * doc/invoke.texi (-fopt-info): Document "omp".
36373 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36374 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36375 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36376 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36377 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36379 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36381 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36382 OPTGROUP_OMP with "omp".
36384 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36387 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36388 for arithmetic shift of unsigned V2DI.
36390 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36392 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36393 arc/linux.h headers.
36394 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36395 (LINK_SPEC): Likewise.
36396 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36397 (EXTRA_SPECS): Likewise.
36398 (STARTFILE_SPEC): Likewise.
36399 (ENDFILE_SPEC): Likewise.
36400 (LIB_SPEC): Likewise.
36401 (TARGET_SDATA_DEFAULT): Likewise.
36402 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36403 (MULTILIB_DEFAULTS): Likewise.
36404 (DWARF2_UNWIND_INFO): Likewise.
36405 * config/arc/big.h: New file.
36406 * config/arc/elf.h: Likewise.
36407 * config/arc/linux.h: Likewise.
36408 * config/arc/t-uClibc: Remove.
36410 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36412 PR tree-optimization/77536
36413 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36414 (tree_transform_and_unroll_loop): Use above function to compute the
36415 estimated niter of unrolled loop and use it when scaling profile.
36416 Also use count info rather than frequency if it's non-zero.
36417 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36418 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36419 (vect_transform_loop): Call above function.
36421 2017-02-27 Richard Biener <rguenther@suse.de>
36423 PR tree-optimization/45397
36424 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36425 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36426 (visit_nary_op): Add pattern matching for CSEing sign-changed
36427 or truncated operations with wider ones.
36429 2017-02-27 Richard Biener <rguenther@suse.de>
36431 PR tree-optimization/79690
36432 * tree-vect-stmts.c (vectorizable_store): Use vector type
36433 built from the DR with address-space.
36435 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36437 * doc/invoke.texi (Optimize Options): Refine the description
36438 of asan-use-after-return.
36440 2017-02-25 Alan Modra <amodra@gmail.com>
36442 PR rtl-optimization/79584
36443 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36444 base, not ad->base_term, the reg within base. Remove assertion
36445 that ad->base == ad->base_term. Replace gen_int_mode using
36446 bogus mode with const0_rtx.
36448 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36450 PR middle-end/79396
36451 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36452 FMA_EXPR like tcc_binary or tcc_unary.
36454 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36457 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36459 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36460 (output_loc_operands): Handle DW_OP_call_ref and
36461 DW_OP_GNU_variable_value.
36462 (struct variable_value_struct): New type.
36463 (struct variable_value_hasher): Likewise.
36464 (variable_value_hash): New variable.
36465 (string_types): Remove.
36466 (copy_loc_descr): New function.
36467 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36468 (prepend_loc_descr_to_each): New function.
36469 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36470 instead of add_loc_descr_to_each if the first argument is single
36471 location list and the second has multiple.
36472 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36473 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36474 when looking for variable value which doesn't have other location info.
36475 (loc_list_from_tree): Formatting fix.
36476 (gen_array_type_die): Simplify DW_AT_string_length handling.
36477 (adjust_string_types): Remove.
36478 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36479 string_types. Call resolve_variable_values.
36480 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36481 (resolve_addr_in_expr): Likewise. Add A argument.
36482 (copy_deref_exprloc): Remove deref argument. Adjust for the
36483 original expression being DW_OP_GNU_variable_value with optionally
36484 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36485 optionally after it.
36486 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36487 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36488 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36489 (variable_value_hasher::hash, variable_value_hasher::equal): New
36491 (resolve_variable_value_in_expr, resolve_variable_value,
36492 resolve_variable_values, note_variable_value_in_expr,
36493 note_variable_value): New functions.
36494 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36497 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36500 * opts.h (handle_generated_option): Add GENERATED_P argument.
36501 * opts-common.c (handle_option): Adjust function comment.
36502 (handle_generated_option): Add GENERATED_P argument, pass it to
36504 (control_warning_option): Pass false to handle_generated_option
36506 * opts.c (maybe_default_option): Pass true to handle_generated_option
36508 * optc-gen.awk: Likewise.
36510 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36512 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36513 a REG, look at the REG it is a SUBREG of.
36514 (splitter for cmpeqsi_t): Ditto.
36516 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36518 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36519 the special USEs with the pattern of the insn, not the insn itself.
36521 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36524 * doc/invoke.texi: Document -mload-store-pairs.
36526 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36527 Sandra Loosemore <sandra@codesourcery.com>
36529 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36530 argument isn't a CONST_INT.
36531 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36532 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36533 (nios2_validate_compare): Bypass alternate compare logic if *op2
36534 is not a CONST_INT.
36535 (ldstwm_operation_p): Return false if first_base is not a REG or
36536 if first_offset is not a CONST_INT.
36538 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36540 * config/cris/cris.md: Use correct operand in a define_peephole2.
36542 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36544 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36546 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36548 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36549 this_insn if it is an INSN or JUMP_INSN.
36550 (force_offsettable): Look at base, not at addr.
36551 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36552 on things that aren't necessarily CONST_INTs.
36554 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36556 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36557 -mfpmath=sse is the default also for x86-32 targets with SSE2
36558 instruction set when @option{-ffast-math} is enabled
36560 2017-02-24 Jeff Law <law@redhat.com>
36562 PR rtl-optimizatoin/79286
36563 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36566 2017-02-24 Richard Biener <rguenther@suse.de>
36568 PR tree-optimization/79389
36569 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36572 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36574 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36575 function comment to reflect reality.
36576 (loop_exits_before_overflow): Fix typo in function description.
36578 2017-02-24 Richard Biener <rguenther@suse.de>
36580 PR tree-optimization/79389
36581 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36582 properly that a threading opportunity exists. Detect conditional
36583 copy/constant propagation opportunities.
36585 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36587 * config/visium/visium.md (type): Add trap.
36588 (b): New mode attribute.
36589 (*btst): Rename into...
36590 (*btst<mode>): ...this and adjust.
36591 (*cbranchsi4_btst_insn): Rename into...
36592 (*cbranch<mode>4_btst_insn): ...this and adjust.
36593 (trap): New define_insn.
36595 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36597 PR tree-optimization/79389
36598 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36599 (noce_reversed_cond_code): New function.
36600 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36601 reversed_comparison_code. Formatting fix.
36602 (noce_try_store_flag): Test rev_cond != NULL in addition to
36603 reversed_comparison_code.
36604 (noce_try_store_flag_constants): Likewise.
36605 (noce_try_store_flag_mask): Likewise.
36606 (noce_try_addcc): Use rev_cond if non-NULL instead of
36607 reversed_comparison_code.
36608 (noce_try_cmove_arith): Likewise. Formatting fixes.
36609 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36610 (noce_find_if_block): Initialize rev_cond.
36611 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36612 instead of false as last argument never attempt to reverse it
36615 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36617 PR tree-optimization/79663
36618 * tree-predcom.c (combine_chains): Process refs in reverse order
36619 only for ZERO length chains, and add explaining comment.
36621 2017-02-23 Jeff Law <law@redhat.com>
36623 PR tree-optimization/79578
36624 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36625 in call to operand_equal_p.
36627 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36630 * config/i386/cpuid.h: Fix another undefined behavior.
36632 2017-02-23 Richard Biener <rguenther@suse.de>
36634 PR tree-optimization/79683
36635 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36636 vector types for data-refs.
36638 2017-02-23 Martin Liska <mliska@suse.cz>
36640 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36642 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36644 PR middle-end/79665
36645 * internal-fn.c (get_range_pos_neg): Moved to ...
36646 * tree.c (get_range_pos_neg): ... here. No longer static.
36647 * tree.h (get_range_pos_neg): New prototype.
36648 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36649 are known to be in between 0 and signed maximum inclusive, try to
36650 expand both unsigned and signed divmod and use the cheaper one from
36653 2017-02-22 Jeff Law <law@redhat.com>
36655 PR tree-optimization/79578
36656 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36657 to compare base operands.
36659 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36662 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36663 gpc_reg_operand instead of fpr_reg_operand.
36665 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36667 * config/mips/mips.c (mips_return_in_memory): Force FP
36668 vector types to be returned in memory for o32 ABI.
36670 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36672 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36673 instead of DW_TAG_member for static data member declarations and don't
36674 set no_linkage_name for static inline data members.
36675 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36678 2017-02-22 Martin Liska <mliska@suse.cz>
36680 * doc/invoke.texi: Replace inequality signs with square brackets
36683 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36686 * lra-constraints.c (simplify_operand_subreg): Handle
36687 WORD_REGISTER_OPERATIONS targets.
36689 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36692 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36693 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36694 elimination by swapping fld*.
36696 2017-02-22 Richard Biener <rguenther@suse.de>
36698 PR tree-optimization/79673
36699 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36700 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36701 irrelevant address-space qualifiers and avoiding a
36702 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36704 2017-02-22 Richard Biener <rguenther@suse.de>
36706 PR tree-optimization/79666
36707 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36708 to not symbolically negate if that may introduce undefined
36711 2017-02-22 Martin Liska <mliska@suse.cz>
36714 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36715 * data-streamer-out.c (streamer_write_gcov_count_stream):
36717 * value-prof.c (stream_out_histogram_value): Make assert more
36718 precise based on type of counter.
36720 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36723 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36724 Use nonimmediate_operand instead of memory_operand for operand 1.
36725 (float-extend standard_x87sse_constant_load splitter): Ditto.
36727 2017-02-21 Jeff Law <law@redhat.com>
36729 PR tree-optimization/79621
36730 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36731 blocks with edges to themselves.
36733 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36736 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36737 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36738 Use gimple_call_builtin_p.
36741 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36742 on temporarily removed DEBUG_INSNs.
36744 PR tree-optimization/79649
36745 * tree-loop-distribution.c (classify_partition): Give up on
36746 non-generic address space loads/stores.
36748 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36750 * doc/loop.texi (Loop manipulation): Remove nonexistent
36751 tree_ssa_loop_version from the documentation.
36752 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36754 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36757 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36758 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36759 * config/rs6000/rs6000.c: Include except.h.
36760 (rs6000_expand_split_stack_prologue): Call
36761 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36763 2017-02-21 Martin Jambor <mjambor@suse.cz>
36766 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36767 have been analyzed.
36769 2017-02-21 Martin Jambor <mjambor@suse.cz>
36771 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36772 for backward compatibility only.
36773 * doc/invoke.texi (Option Summary): Remove all references to
36774 -fipa-cp-alignment.
36776 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36780 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36782 * lra-constraints.c (curr_insn_transform): Handle
36783 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36785 2017-02-21 Martin Liska <mliska@suse.cz>
36787 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36789 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36791 PR translation/79638
36792 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36794 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36797 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36798 (arm_function_ok_for_sibcall): Return false for an indirect call by
36799 descriptor if all the argument registers are used.
36800 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36801 alignment of the function.
36803 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36805 PR tree-optimization/61441
36806 * simplify-rtx.c (simplify_const_unary_operation): For
36807 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36808 the sNaN unmodified.
36810 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36812 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36813 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36814 instead of SYSTEM_HEADER_DIR.
36816 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36817 Martin Liška <mliska@suse.cz>
36819 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36820 Fix typos and grammar, use active voice, and clarify.
36822 2017-02-20 Marek Polacek <polacek@redhat.com>
36824 PR middle-end/79537
36825 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36828 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36830 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36833 * config/i386/i386.c (ix86_expand_builtin): Handle
36834 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36835 ix86_builtins_isa[fcode].isa as a requirement of those
36836 flags and any other flag in the bitmask.
36837 (ix86_init_mmx_sse_builtins): Use 0 instead of
36838 ~OPTION_MASK_ISA_64BIT as mask.
36839 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36840 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36841 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36842 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36844 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36847 * lra-constraints.c (split_reg): Check requested split mode
36848 is supported by the register.
36850 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36852 * lra-constraints.c (simplify_operand_subreg): Remove early
36855 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36858 * lra-constraints.c (curr_insn_transform): Tighten condition
36859 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36861 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36864 * lra-constraints.c (curr_insn_transform): Handle
36865 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36867 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36870 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36872 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36874 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36877 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36880 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36882 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36885 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36887 * final.c (last_columnnum, override_columnnum): New variables.
36888 (final_start_function): Set last_columnnum, pass it to begin_prologue
36889 hook and pass 0 to dwarf2out_begin_prologue.
36890 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36891 to source_line debug hook.
36892 (notice_source_line): Compute last_columnnum and for debug_column_info
36893 return true on column changes.
36894 * debug.h (struct gcc_debug_hooks): Add column argument to
36895 source_line and begin_prologue hooks.
36896 (debug_nothing_int_charstar_int_bool): Remove prototype.
36897 (debug_nothing_int_int_charstar,
36898 debug_nothing_int_int_charstar_int_bool): New prototypes.
36899 (dwarf2out_begin_prologue): Add column argument.
36900 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36901 begin_prologue hooks.
36902 (debug_nothing_int_charstar_int_bool): Remove.
36903 (debug_nothing_int_int_charstar,
36904 debug_nothing_int_int_charstar_int_bool): New functions.
36905 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36906 through to dwarf2out_source_line.
36907 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36908 (dwarf2out_source_line): Add column argument, emit it if requested.
36909 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36911 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36912 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36913 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36914 through to dwarf2out_begin_prologue.
36915 (vmsdbgout_source_line): Add column argument, pass it through to
36916 dwarf2out_source_line.
36917 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36918 dbxout_source_line caller.
36919 (dbxout_source_line): Add column argument.
36921 * common.opt (gno-column-info, gcolumn-info): New options.
36922 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36923 (check_die): Also test for multiple DW_AT_decl_column attributes.
36924 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36925 DW_AT_decl_column if requested.
36926 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36928 (gen_variable_die): Likewise.
36929 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36930 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36933 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36935 (ix86_handle_option): Handle OPT_m3dnowa.
36936 * doc/invoke.texi (-m3dnowa): Document.
36937 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36938 -m3dnowa instead of -m3dnow -march=athlon.
36941 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36942 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36944 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36947 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36948 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36949 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36950 generator for vsx_xxpermdi_<mode>_be.
36951 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36952 force big-endian semantics.
36953 (vsx_xxpermdi_<mode>_be): New define_expand with same
36954 implementation as previous version of vsx_xxpermdi_<mode>.
36956 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36958 PR tree-optimization/79327
36959 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36960 variable, its initialization and use.
36962 2017-02-17 Julia Koval <julia.koval@intel.com>
36964 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36965 (OPTION_MASK_ISA_PKU_UNSET): New.
36966 (ix86_handle_option): Handle -mrdpid.
36967 * config/i386/cpuid.h (bit_RDPID): New.
36968 * config/i386/driver-i386.c (host_detect_local_cpu):
36969 Detect RDPID feature.
36970 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36971 * config/i386/i386-c.c (ix86_target_macros_internal):
36973 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36974 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36975 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36976 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36977 * config/i386/i386.md (define_insn "rdpid"): New.
36978 * config/i386/i386.opt Add -mrdpid.
36979 * config/i386/immintrin.h (_rdpid_u32): New.
36981 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36983 PR rtl-optimization/79541
36984 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36985 instead of transforming it into USE.
36987 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36989 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36990 If HONOR_SNANS (SFmode) force the input to a register.
36991 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36992 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36993 an frsp or similar insn.
36995 2017-02-17 Martin Liska <mliska@suse.cz>
36997 PR rtl-optimization/79577
36998 * params.def (selsched-max-sched-times): Increase minimum to 1.
37000 2017-02-17 Martin Liska <mliska@suse.cz>
37002 PR rtl-optimization/79574
37003 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37005 2017-02-17 Martin Liska <mliska@suse.cz>
37007 PR tree-optimization/79529
37008 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37009 ssa_defined_default_def_p to handle cases which are implicitly
37011 * tree-ssa.c (ssa_defined_default_def_p): New function.
37012 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37013 which are implicitly defined.
37014 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37016 2017-02-17 Richard Biener <rguenther@suse.de>
37018 PR middle-end/79576
37019 * params.def (max-ssa-name-query-depth): Limit to 10.
37021 2017-02-17 Richard Biener <rguenther@suse.de>
37023 PR tree-optimization/79552
37024 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37027 2017-02-17 Richard Biener <rguenther@suse.de>
37030 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37032 2017-02-17 Marek Polacek <polacek@redhat.com>
37034 PR middle-end/79536
37035 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37036 (fold_negate_expr): New wrapper.
37038 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37040 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37041 Correct terminology and de-emphasize pre-standard behavior.
37043 2017-02-16 Alan Modra <amodra@gmail.com>
37045 PR rtl-optimization/79286
37046 * ira.c (def_dominates_uses): New function.
37047 (update_equiv_regs): Don't create an equivalence for insns that
37048 may trap where the register def does not dominate the use.
37050 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37052 PR rtl-optimization/78127
37053 * lra.c (lra): Call lra_eliminate before finish the loop after
37056 2017-02-16 Richard Biener <rguenther@suse.de>
37058 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37060 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37061 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37062 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37063 (isl_val_int_from_wi): New function.
37064 (extract_affine_gmp): Rename to ...
37065 (extract_affine_wi): ... this, take a widest_int.
37066 (extract_affine_int): Just wrap extract_affine_wi.
37067 (add_param_constraints): Use isl_val_int_from_wi.
37068 (add_loop_constraints): Likewise, and extract_affine_wi.
37070 2017-02-15 Jeff Law <law@redhat.com>
37072 PR middle-end/79521
37073 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37074 ira_init_register_move_cost_if_necessary.
37076 2017-02-15 Martin Sebor <msebor@redhat.com>
37078 PR middle-end/32003
37079 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37080 removed in a prior commit.
37082 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37084 PR tree-optimization/79347
37085 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37086 counters during peeling.
37088 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37090 * Makefile.in (site.exp): Remove "set ISLVER".
37092 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37095 * real.c (real_from_integer): Call real_convert even for decimal.
37097 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37100 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37102 2017-02-14 Andrew Pinski <apinski@cavium.com>
37104 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37105 cores and change the partno/implementer to be correct.
37106 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37107 the 'B" as the implementer.
37108 * config/aarch64/aarch64-tune.md: Regenerate.
37110 2017-02-14 Carl Love <cel@us.ibm.com>
37112 * config/rs6000/rs6000.c: Add case statement entry to make the
37113 xvcvuxdsp built-in argument unsigned.
37114 * config/rs6000/vsx.md: Fix the source and return operand types so they
37115 match the instruction definitions from the ISA document. Fix typo
37116 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37119 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37122 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37123 member early_clobber_alts.
37124 * lra-lives.c (reg_early_clobber_p): New.
37125 (process_bb_lives): Use it.
37126 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37127 (debug_operand_data): Initialize early_clobber_alts.
37128 (setup_operand_alternative): Set up early_clobber_alts.
37129 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37130 alternatives to new_insn_reg.
37131 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37133 (lra_update_insn_regno_info): Pass the new arg.
37135 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37137 PR middle-end/79505
37138 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37139 (new_oacc_loop_raw): Don't clear already cleared fields.
37142 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37143 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37144 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37146 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37149 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37151 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37154 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37155 the extra instruction to the right place to store 128-bit constant
37158 2017-02-14 Martin Sebor <msebor@redhat.com>
37160 PR middle-end/79448
37161 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37162 warning for strings of unknown length.
37164 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37166 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37168 2017-02-14 Jeff Law <law@redhat.com>
37171 * ira-costs.c (scan_one_insn): Initialize register move costs
37172 for pseudos seen in USE/CLOBBER insns.
37174 PR tree-optimization/79095
37175 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37176 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37177 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37178 if the operands are known to be not equal, then the resulting range
37180 (intersect_ranges): If the new range is ~[0,0] and the old range is
37181 wide, then prefer ~[0,0].
37182 * tree-vrp.c (overflow_comparison_p_1): New function.
37183 (overflow_comparison_p): New function.
37184 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37185 if NAME is used in an overflow test.
37186 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37187 overflow check that can be expressed as an equality test, then adjust
37188 ops to be that equality test.
37190 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37192 * config/s390/s390-builtin-types.def: Remove flags argument.
37193 * config/s390/s390.c (s390_init_builtins): Likewise.
37195 2017-02-14 Martin Liska <mliska@suse.cz>
37197 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37198 vector. Fix trailing white spaces.
37200 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37202 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37205 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37207 PR rtl-optimization/68664
37208 * config/arm/arm.c (arm_sched_can_speculate_insn):
37209 New function. Declare prototype.
37210 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37212 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37214 PR rtl-optimization/68664
37215 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37217 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37219 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37221 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37222 max skip bytes for function, loop and jump.
37224 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37226 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37227 ABS_EXPR for gimple dump.
37229 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37232 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37234 PR tree-optimization/79408
37235 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37236 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37237 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37238 also if rhs1 is INTEGER_CST.
37240 2017-02-14 Richard Biener <rguenther@suse.de>
37242 PR middle-end/79432
37243 * tree-into-ssa.c (insert_phi_nodes): When the function can
37244 have abnormal edges rewrite SSA names with broken use-def
37245 dominance out of SSA and register them for PHI insertion.
37247 2017-02-13 Martin Sebor <msebor@redhat.com>
37249 PR middle-end/79496
37250 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37251 clearing info.nowrite flag when snprintf size argument is a range.
37253 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37255 * cprop.c (cprop_jump): Add missing space in string literal.
37256 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37257 (get_constraint_for_component_ref): Likewise.
37258 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37259 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37260 * lra-constraints.c (process_alt_operands): Likewise.
37261 * ipa-inline.c (inline_small_functions): Likewise.
37262 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37263 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37264 * trans-mem.c (diagnose_tm_1_op): Likewise.
37265 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37266 (grid_parallel_clauses_gridifiable): Likewise.
37268 * config/nvptx/mkoffload.c (process): Add space in between
37271 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37272 "MOD4_SSE_REGS" and "ALL_REGS".
37274 * spellcheck.c (test_data): Add , in between "foo" and "food".
37276 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37279 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37280 boundary crossing check and subsequent code generation agree.
37282 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37284 * config/aarch64/aarch64.c (has_memory_op): Delete.
37285 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37288 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37290 PR rtl-optimization/79388
37291 PR rtl-optimization/79450
37292 * combine.c (distribute_notes): When removing TEM_INSN for which
37293 corresponding dest has last value recorded, invalidate that last
37296 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37298 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37299 of explicit '@'. Add missing assembly comment marker on branch costs
37302 2017-02-13 Nathan Sidwell <nathan@acm.org>
37304 * gengtype-lex.l (<in_struct>): Add '/'.
37306 2017-02-13 Martin Liska <mliska@suse.cz>
37309 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37311 2017-02-13 Richard Biener <rguenther@suse.de>
37313 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37315 * configure: Re-generate.
37316 * config.in: Likewise.
37317 * graphite-dependences.c: Simplify as if
37318 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37319 * graphite-isl-ast-to-gimple.c: Likewise.
37320 * graphite-optimize-isl.c: Likewise.
37321 * graphite-poly.c: Likewise.
37322 * graphite-sese-to-poly.c: Likewise.
37323 * graphite.h: Likewise.
37324 * toplev.c: Include isl/version.h and use isl_version () for
37325 printing the ISL version.
37326 * doc/install.texi: Update ISL requirement.
37328 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37330 * doc/standards.texi (Standards): Update reference to
37333 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37335 * doc/extend.texi (Named Address Spaces): sourceware.org now
37337 * doc/install.texi (Binaries): Ditto.
37340 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37342 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37343 standard terminology "stringize"/"stringizing" throughout.
37344 * doc/cppinternals.texi: Likewise.
37346 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37348 * doc/extend.texi: Fix some spelling mistakes and typos.
37349 * doc/invoke.texi: Likewise.
37351 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37354 * params.def (inline-min-speedup) Change from 10 to 8.
37356 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37358 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37361 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37364 * ipa-inline-analysis.c (get_minimal_bb): New function.
37365 (record_modified): Use it.
37366 (remap_edge_change_prob): Handle also ancestor functions.
37368 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37370 * doc/contrib.texi (Contributors): Remove broken link into
37371 the Mauve CVS repository.
37373 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37375 PR middle-end/79454
37376 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37377 result computation whenever lhs doesn't have vector mode, not
37378 just when it has BLKmode.
37380 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37382 * doc/makefile.texi (profiledbootstrap): Refer to the
37383 installation instructions only in textual form.
37385 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37388 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37390 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37392 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37393 (Specific): Update mingw-w64 reference.
37395 (Specific): Remove broken link to Renesas RX processor.
37397 2017-02-10 Richard Biener <rguenther@suse.de>
37399 * toplev.c (process_options): Do not mention obsolete graphite
37400 options when printing sorry message about missing graphite support.
37401 Mention -floop-nest-optimize.
37403 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37405 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37406 (vtst_p16): Likewise.
37407 (vtstq_p8): Likewise.
37408 (vtstq_p16): Likewise.
37410 (vtstq_p64): Likewise.
37411 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37412 (vset_lane_p64): New.
37413 (vsetq_lane_p64): New.
37415 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37417 PR tree-optimization/79411
37418 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37419 stmt operands are SSA_NAMEs used in abnormal phis.
37420 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37423 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37426 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37429 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37431 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37433 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37435 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37439 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37440 not arbitrary TREE_CONSTANT.
37443 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37444 "omp declare target link" attribute unless is_global_var.
37445 * omp-offload.c (find_link_var_op): Likewise.
37447 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37448 Chung-Lin Tang <cltang@codesourcery.com>
37450 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37452 (gimplify_adjust_omp_clauses): Don't delete TILE.
37453 (gimplify_omp_for): Deal with TILE.
37454 * internal-fn.c (expand_GOACC_TILE): New function.
37455 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37457 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37458 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37460 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37461 avoid DIV for outermost collapse var.
37462 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37463 Remove out of date comments, fix whitespace.
37464 * omp-general.c (omp_extract_for_data): Deal with tiling.
37465 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37466 adjust OLF_DIM_BASE value.
37467 (struct omp_for_data): Add tiling field.
37468 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37469 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37470 for auto loops. Remove default auto determining, moved to
37471 oacc_loop_fixed_partitions.
37472 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37473 stmts, add e_mask field.
37474 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37475 (oacc_thread_numbers): Use oacc_dim_call.
37476 (oacc_xform_tile): New.
37477 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37478 (finish_oacc_loop): Adjust for ifns vector.
37479 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37480 add case for GOACC_TILE fns.
37481 (oacc_loop_xform_loop): Delete.
37482 (oacc_loop_process): Iterate over call list directly, and add
37483 handling for GOACC_TILE fns.
37484 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37486 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37487 vector partitioning to outer loops. Assign 2 partitions to loops
37488 when available. Add TILE handling.
37489 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37490 (execite_oacc_device_lower): Process GOACC_TILE fns,
37491 ignore unknown specs.
37492 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37493 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37494 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37496 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37498 * configure.ac (ACX_BUGURL): Update.
37499 * configure: Regenerate.
37501 2017-02-09 Richard Biener <rguenther@suse.de>
37503 PR tree-optimization/69823
37504 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37505 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37507 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37509 * config/arc/arc-c.def: Add __NPS400__ definition.
37510 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37511 (TARGET_NPS400): Define.
37513 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37515 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37517 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37518 pointer, arch_info.
37519 (arc_cpu_types): Fill the arch_info field with a pointer into the
37520 arc_arch_types table.
37521 (arc_selected_cpu): Declare.
37522 * config/arc/arc.c (arc_selected_cpu): Make global.
37523 (arc_selected_arch): Delete.
37524 (arc_base_cpu): Delete.
37525 (arc_override_options): Remove references to deleted variables,
37526 update access to arch information.
37527 (ARC_OPT): Update access to arch information.
37528 (ARC_OPTX): Likewise.
37529 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37530 (TARGET_ARC600): Update access to arch information.
37531 (TARGET_ARC601): Likewise.
37532 (TARGET_ARC700): Likewise.
37533 (TARGET_EM): Likewise.
37534 (TARGET_HS): Likewise.
37535 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37538 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37541 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37542 condition/operands for integer GE/LE/GEU/LEU operations.
37544 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37546 PR translation/79397
37547 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37550 2017-02-08 Martin Jambor <mjambor@suse.cz>
37553 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37554 whether allocation happened.
37555 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37556 nothing was allocated.
37558 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37560 PR tree-optimization/79408
37561 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37562 constant, but SSA_NAME with a known integer range, use the minimum
37563 of that range instead of op1 to determine if modulo can be replaced
37564 with its first operand.
37566 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37568 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37570 2017-02-08 Richard Biener <rguenther@suse.de>
37572 PR tree-optimization/71824
37573 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37574 Check all loops contained in the merged region.
37576 2017-02-07 Andrew Pinski <apinski@cavium.com>
37578 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37580 2017-02-07 Andrew Pinski <apinski@cavium.com>
37582 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37583 (thunderxt88): Likewise.
37584 (thunderxt81): Disable LSE and change v8.1 to v8.
37585 (thunderxt83): Likewise.
37587 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37588 Richard Biener <rguenther@suse.de>
37590 PR middle-end/79399
37591 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37592 type from int to size_t.
37593 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37595 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37597 PR rtl-optimization/79386
37598 * cprop.c (bypass_conditional_jumps): Initialize
37599 bypass_last_basic_block already before splitting bbs after
37600 unconditional traps...
37601 (bypass_conditional_jumps): ... rather than here.
37604 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37605 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37606 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37607 fix -masm=intel patterns.
37609 2017-02-07 Richard Biener <rguenther@suse.de>
37611 PR tree-optimization/79256
37612 PR middle-end/79278
37613 * builtins.c (get_object_alignment_2): Use min_align_of_type
37614 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37615 and ADJUST_FIELD_ALIGN.
37617 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37619 * doc/tm.texi: Regenerate.
37620 * stor-layout.c (layout_decl): Adjust.
37621 (update_alignment_for_field): Likewise.
37622 (place_field): Likewise.
37623 (min_align_of_type): Likewise.
37624 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37625 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37626 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37627 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37628 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37629 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37630 * config/i386/i386.c (x86_field_alignment): Likewise.
37631 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37632 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37633 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37634 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37635 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37636 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37640 2017-01-30 Richard Biener <rguenther@suse.de>
37642 PR tree-optimization/79256
37643 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37644 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37647 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37649 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37650 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37651 builtins to SImode and emit a zero-extend, if necessary.
37653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37655 * docs/invoke.texi (RISC-V Options): Alphabetize.
37657 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37659 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37662 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37664 * config/riscv/riscv.c: New file.
37665 * common/config/riscv/riscv-common.c: Likewise.
37666 * config.gcc: Likewise.
37667 * config/riscv/constraints.md: Likewise.
37668 * config/riscv/elf.h: Likewise.
37669 * config/riscv/generic.md: Likewise.
37670 * config/riscv/linux.h: Likewise.
37671 * config/riscv/multilib-generator: Likewise.
37672 * config/riscv/peephole.md: Likewise.
37673 * config/riscv/pic.md: Likewise.
37674 * config/riscv/predicates.md: Likewise.
37675 * config/riscv/riscv-builtins.c: Likewise.
37676 * config/riscv/riscv-c.c: Likewise.
37677 * config/riscv/riscv-ftypes.def: Likewise.
37678 * config/riscv/riscv-modes.def: Likewise.
37679 * config/riscv/riscv-opts.h: Likewise.
37680 * config/riscv/riscv-protos.h: Likewise.
37681 * config/riscv/riscv.h: Likewise.
37682 * config/riscv/riscv.md: Likewise.
37683 * config/riscv/riscv.opt: Likewise.
37684 * config/riscv/sync.md: Likewise.
37685 * config/riscv/t-elf-multilib: Likewise.
37686 * config/riscv/t-linux: Likewise.
37687 * config/riscv/t-linux-multilib: Likewise.
37688 * config/riscv/t-riscv: Likewise.
37689 * configure.ac: Likewise.
37690 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37691 Waterman as RISC-V maintainers.
37692 * doc/install.texi: Add RISC-V entries.
37693 * doc/invoke.texi: Add RISC-V options section.
37694 * doc/md.texi: Add RISC-V constraints section.
37695 * configure: Regenerated.
37697 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37700 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37701 false values to be constant vectors with all 0 or all 1 bits set.
37702 (vcondu<mode><mode>): Likewise.
37703 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37705 (fpmask_comparison_operator): Update comment.
37706 (vecint_comparison_operator): New predicate.
37707 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37708 vector conditionals when the true and false values are constant
37709 vectors with all 0 bits or all 1 bits set.
37711 2017-02-06 Martin Sebor <msebor@redhat.com>
37713 PR tree-optimization/79376
37714 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37716 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37718 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37719 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37720 to simplify split condition.
37722 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37724 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37725 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37728 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37730 PR rtl-optimization/68664
37731 * target.def (can_speculate_insn): New hook.
37732 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37733 * doc/tm.texi: Regenerate.
37734 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37735 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37736 (rs6000_sched_can_speculate_insn): New function.
37738 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37740 PR tree-optimization/79284
37741 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37742 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37743 vectorizable_mask_load_store, vectorizable_operation,
37744 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37745 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37746 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37747 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37748 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37749 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37750 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37751 is_gimple_assign (stmt). Replace another such test with
37752 is_gimple_assign (stmt).
37754 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37757 * config/avr/avr.c (rtl-iter.h): Include it.
37758 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37759 (avr_legitimate_combined_insn): ...and implementation.
37761 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37763 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37764 * config/s390/s390.c (s390_const_operand_ok)
37765 (s390_canonicalize_comparison, s390_extract_part)
37766 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37767 (s390_contiguous_bitmask_p, s390_rtx_costs)
37768 (legitimize_pic_address): Likewise.
37769 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37770 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37771 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37772 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37773 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37775 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37777 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37778 REGNO($0) == REGNO($1).
37780 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37782 * config/s390/linux.h(SIZE_TYPE): Add comment.
37784 2017-02-06 Julian Brown <julian@codesourcery.com>
37785 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37786 Virendra Pathak <virendra.pathak@broadcom.com>
37788 * config/aarch64/aarch64-cores.def: Change the scheduler
37790 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37791 * config/aarch64/thunderx2t99.md: New file.
37793 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37795 * doc/standards.texi (Go Language): Update link to language
37798 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37800 * tree-eh.c (lower_resx): Sanitize profile.
37801 (cleanup_empty_eh_move_lp): Likewise.
37803 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37806 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37808 * cfgloopmanip.h (loop_version): Update prototype.
37809 * modulo-sched.c (sms_schedule): Update call of loop_version.
37810 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37811 * tree-parloops.c (gen_parallel_loop): Likewise.
37812 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37813 * tree-ssa-loop-split.c (split_loop): Likewise.
37814 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37815 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37817 2017-02-05 Martin Liska <mliska@suse.cz>
37820 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37822 (print_operand_address): Initialize a struct to zero.
37824 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37826 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37827 garbage collector only in textual form.
37829 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37831 * doc/extend.texi (x86 specific memory model extensions for
37832 transactional memory): Simplify a phrase.
37834 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37837 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37838 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37839 (atomic_storedi_1): Likewise.
37841 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37843 PR tree-optimization/79338
37844 * tree-parloops.c (gather_scalar_reductions): Don't call
37845 vect_analyze_loop_form for loop->inner before destroying loop's
37848 2017-02-03 Martin Sebor <msebor@redhat.com>
37850 PR tree-optimization/79327
37851 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37852 when precision has resulted in leading zeros.
37853 (format_integer): Adjust the likely counter to assume an unknown
37854 argument that may be zero is non-zero.
37856 2017-02-03 Jason Merrill <jason@redhat.com>
37859 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37860 avoid copying non-taken branch.
37862 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37864 PR tree-optimization/79340
37865 * tree-vect-loop.c (vectorizable_reduction): Release
37866 vec_defs elements after safe_splicing them into other vectors.
37869 PR tree-optimization/79327
37870 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37871 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37873 (format_integer): Use wide_int_to_tree instead of build_int_cst
37874 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37875 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37876 of shortest and longest sequence.
37878 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37880 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37881 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37883 2017-02-03 Walter Lee <walt@tilera.com>
37886 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37887 after initial stackframe link reg save.
37888 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37890 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37893 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37894 wu for stxssp alternative.
37896 2017-02-03 Martin Sebor <msebor@redhat.com>
37898 PR tree-optimization/79352
37899 * gimple-fold.c (get_range_strlen): Add argument.
37900 (get_range_strlen): Change return type to bool.
37901 (get_maxval_strlen): Pass in a dummy argument.
37902 * gimple-fold.h (get_range_strlen): Change return type to bool.
37903 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37904 * tree.h (array_at_struct_end_p): Add argument.
37905 * tree.c (array_at_struct_end_p): Handle it.
37907 2017-02-03 Martin Liska <mliska@suse.cz>
37910 * multiple_target.c (create_dispatcher_calls): Redirect edge
37911 from a caller of a dispatcher.
37912 (expand_target_clones): Make the clones local.
37913 (ipa_target_clone): Do both target clones and resolvers.
37914 (ipa_dispatcher_calls): Remove the pass.
37915 (pass_dispatcher_calls::gate): Likewise.
37916 (make_pass_dispatcher_calls): Likewise.
37917 * passes.def (pass_target_clone): Put as very first IPA early
37920 2017-02-03 Martin Liska <mliska@suse.cz>
37922 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37923 in case of a function with ifunc attribute.
37925 2017-02-03 Martin Liska <mliska@suse.cz>
37927 * cgraph.c (cgraph_node::dump): Dump function version info.
37928 * symtab.c (symtab_node::dump_base): Add missing new line.
37930 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37933 (ifcombine_ifandif): Use it.
37935 2017-02-03 Martin Liska <mliska@suse.cz>
37937 * doc/invoke.texi: Document default value for
37938 use-after-scope-direct-emission-threshold.
37940 2017-02-03 Martin Liska <mliska@suse.cz>
37942 PR tree-optimization/79339
37943 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37944 (format_floating): Likewise.
37946 2017-02-03 Martin Liska <mliska@suse.cz>
37949 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37951 (ipa_node_params_t::remove): Likewise.
37952 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37953 initialization from removed ipa_node_params_t::insert.
37954 (ipa_node_params::~ipa_node_params): Move from removed
37955 ipa_node_params_t::release.
37956 * symbol-summary.h (symbol_summary::m_released): New member.
37957 Do not release a summary twice. Do not allow to call finalizer
37958 for types of a summary that live in GGC memory.
37960 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37962 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37965 2017-02-02 Martin Sebor <msebor@redhat.com>
37967 PR middle-end/79275
37968 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37969 (format_string): Tighten up the range of output for non-constant
37970 strings and correct the expected range for wide non-constant strings.
37972 2017-02-02 Martin Sebor <msebor@redhat.com>
37974 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37976 PR middle-end/32003
37977 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37979 (-fdump-tree-@var): Add to index and document how to come up
37980 with pass-specific option and dump file names.
37981 (-fdump-passes): Clarify where to look for output.
37983 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37985 PR middle-end/77445
37986 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37987 statistics of the analyzed path; allow threading for speed when
37988 any of BBs along the path are optimized for speed.
37990 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37992 PR middle-end/78468
37993 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37994 settings of the virtual registers.
37997 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37999 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38000 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38003 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38005 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38006 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38008 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38010 * config/s390/s390.md: Add missing comments with the expanded
38012 * config/s390/vector.md: Likewise.
38013 * config/s390/vx-builtins.md: Likewise.
38015 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38018 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38019 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38020 conditions on a single line.
38022 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38024 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38025 __S390_VX__ to __VX__.
38027 2017-02-01 Andrew Pinski <apinski@cavium.com>
38029 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38030 stmt_info to record_stmt_cost.
38031 (vect_get_known_peeling_cost): Pass stmt_info if known to
38033 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38034 cpu_vector_cost field into
38035 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38036 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38037 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38038 splitting of scalar_stmt_cost and vec_stmt_cost.
38039 (thunderx_vector_cost): Likewise.
38040 (cortexa57_vector_cost): LIkewise.
38041 (exynosm1_vector_cost): Likewise.
38042 (xgene1_vector_cost): Likewise.
38043 (thunderx2t99_vector_cost): Improve after the splitting of the two
38045 (aarch64_builtin_vectorization_cost): Update for the splitting of
38046 scalar_stmt_cost and vec_stmt_cost.
38048 2017-02-01 Torvald Riegel <triegel@redhat.com>
38049 Richard Henderson <rth@redhat.com>
38051 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38052 conditional on existance of a fast atomic load.
38053 * optabs-query.c (can_atomic_load_p): New function.
38054 * optabs-query.h (can_atomic_load_p): Declare it.
38055 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38056 no fast atomic load is available for the particular size of access.
38057 (expand_atomic_compare_and_swap): Likewise.
38058 (expand_atomic_load): Likewise.
38059 (expand_atomic_store): Likewise.
38060 (expand_atomic_fetch_op): Likewise.
38061 * testsuite/lib/target-supports.exp
38062 (check_effective_target_sync_int_128): Remove x86 because it provides
38063 no fast atomic load.
38064 (check_effective_target_sync_int_128_runtime): Likewise.
38066 2017-02-01 Richard Biener <rguenther@suse.de>
38068 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38069 (graphite_transform_loops): Provide opt-info for optimized nests.
38070 * tree-parloop.c (parallelize_loops): Provide opt-info for
38071 parallelized loops.
38073 2017-02-01 Richard Biener <rguenther@suse.de>
38075 PR middle-end/79315
38076 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38077 was not set before.
38079 2017-02-01 Richard Biener <rguenther@suse.de>
38081 PR tree-optimization/71824
38082 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38083 Verify the loops are valid in the merged SESE region.
38084 (scop_detection::can_represent_loop_1): Check analyzing the
38085 evolution of the number of iterations in the region succeeds.
38087 2017-01-31 Ian Lance Taylor <iant@golang.org>
38089 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38090 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38092 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38094 PR preprocessor/79210
38095 * input.c (get_substring_ranges_for_loc): Replace line_width
38096 assertion with error-handling.
38098 2017-01-31 Richard Biener <rguenther@suse.de>
38100 PR tree-optimization/77318
38101 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38102 (create_pw_aff_from_tree): Take loop parameter.
38103 (add_condition_to_pbb): Pass loop of the condition to
38104 create_pw_aff_from_tree.
38106 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38108 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38109 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38111 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38115 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38117 (convert_int_to_float128): Likewise.
38118 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38119 (convert_int_to_float128): Likewise.
38120 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38121 (UNSPEC_IEEE128_CONVERT): Likewise.
38122 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38123 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38124 Use local variables for IBM extended format.
38125 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38126 (fix_trunc<mode>si2_fprs): Likewise.
38127 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38128 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38129 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38130 to know that we can now have integers of all sizes in vector
38132 (fix<uns>_<mode>di2_hw): Likewise.
38133 (float<uns>_<mode>si2_hw): Likewise.
38134 (fix_<mode>si2_hw): Likewise.
38135 (fixuns_<mode>si2_hw): Likewise.
38136 (float<uns>_<mode>di2_hw): Likewise.
38137 (float_<mode>di2_hw): Likewise.
38138 (float_<mode>si2_hw): Likewise.
38139 (floatuns_<mode>di2_hw): Likewise.
38140 (floatuns_<mode>si2_hw): Likewise.
38141 (xscvqp<su>wz_<mode>): Delete, no longer used.
38142 (xscvqp<su>dz_<mode>): Likewise.
38143 (xscv<su>dqp_<mode>): Likewise.
38144 (ieee128_mfvsrd_64bit): Likewise.
38145 (ieee128_mfvsrd_32bit): Likewise.
38146 (ieee128_mfvsrwz): Likewise.
38147 (ieee128_mtvsrw): Likewise.
38148 (ieee128_mtvsrd_64bit): Likewise.
38149 (ieee128_mtvsrd_32bit): Likewise.
38151 2017-01-31 Martin Liska <mliska@suse.cz>
38154 * ipa-prop.c (ipa_free_all_node_params): Call release method
38155 instead of ~sumbol_summary to not to trigger double times
38158 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38160 PR tree-optimization/71691
38161 * bitmap.h (class auto_bitmap): New.
38162 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38163 is_maybe_undefined instead of ssa_undefined_value_p.
38165 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38167 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38168 __S390_ARCH_LEVEL__ to __ARCH__.
38170 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38172 PR tree-optimization/79267
38173 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38174 if should_remove_lhs_p is true.
38176 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38179 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38180 (add_alignment_attribute): New.
38181 (base_type_die): Add alignment attribute.
38182 (subrange_type_die): Likewise.
38183 (modified_type_die): Likewise.
38184 (gen_array_type_die): Likewise.
38185 (gen_descr_array_type_die: Likewise.
38186 (gen_enumeration_type_die): Likewise.
38187 (gen_subprogram_die): Likewise.
38188 (gen_variable_die): Likewise.
38189 (gen_field_die): Likewise.
38190 (gen_ptr_to_mbr_type_die): Likewise.
38191 (gen_struct_or_union_type_die): Likewise.
38192 (gen_subroutine_type_die): Likewise.
38193 (gen_typedef_die): Likewise.
38194 (base_type_cmp): Compare alignment attribute.
38196 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38199 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38200 (setb_unsigned) New pattern for setb with CCUNS.
38201 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38202 subfc./subfe sequence to avoid overflow problems. Generate a
38203 shorter sequence with cmpld/setb for power9.
38204 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38205 for generating subfc. instruction.
38206 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38207 now uses this instruction.
38209 2017-01-30 Ian Lance Taylor <iant@google.com>
38212 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38213 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38215 2017-01-30 Martin Sebor <msebor@redhat.com>
38217 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38218 Move constant to the right of a relational operator.
38219 (get_mpfr_format_length, format_character, format_string): Ditto.
38220 (should_warn_p, maybe_warn): Same.
38222 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38224 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38227 * asan.c (get_translation_unit_decl): Remove function.
38228 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38230 2017-01-30 Martin Liska <mliska@suse.cz>
38232 PR gcov-profile/79259
38233 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38234 -fprofile-generate.
38236 2017-01-30 Martin Liska <mliska@suse.cz>
38239 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38240 Initialize variables with NULL value.
38242 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38245 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38247 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38249 2017-01-30 Richard Biener <rguenther@suse.de>
38251 PR tree-optimization/79276
38252 * tree-vrp.c (process_assert_insertions): Properly adjust common
38253 when removing a duplicate.
38255 2017-01-30 Richard Biener <rguenther@suse.de>
38257 PR tree-optimization/79256
38258 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38259 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38261 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38263 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38266 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38267 ("*r<noxa>sbg_<mode>_sll_bitmask")
38268 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38269 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38270 Use contiguous_bitmask_nowrap_operand.
38272 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38275 * config/rs6000/altivec.h (vec_xl): Revise #define.
38276 (vec_xst): Likewise.
38278 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38280 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38282 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38284 PR rtl-optimization/79194
38285 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38286 traps before call to bypass_conditional_jumps.
38288 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38290 PR tree-optimization/71374
38291 * lra-constraints.c (check_conflict_input_operands): New.
38292 (match_reload): Use it.
38294 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38297 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38298 account to calculate conflict_set.
38300 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38302 PR rtl-optimization/78559
38303 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38304 other_insn in combine.
38306 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38308 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38309 uint16_type_node for BT_UINT16.
38311 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38313 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38314 "RTL Tests" to menu.
38315 (GIMPLE Tests): New node.
38316 (RTL Tests): New node.
38318 2017-01-27 Richard Biener <rguenther@suse.de>
38320 PR tree-optimization/79245
38321 * tree-loop-distribution.c (distribute_loop): Apply cost
38322 modeling also to detected patterns.
38324 2017-01-27 Richard Biener <rguenther@suse.de>
38326 PR tree-optimization/71433
38327 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38328 (compare_assert_loc): New function.
38329 (process_assert_insertions): Sort and optimize assert locations
38330 to remove duplicates and push down identical assertions on
38331 edges to their destination block.
38333 2017-01-27 Richard Biener <rguenther@suse.de>
38335 PR tree-optimization/79244
38336 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38337 out SSA names even if abnormal.
38339 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38341 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38342 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38343 instead of MPFR_RNDN.
38345 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38348 * arm.c (arm_option_override): Don't call build_target_option_node
38349 until after doing all option overrides.
38350 (arm_valid_target_attribute_tree): Likewise.
38352 2017-01-27 Martin Liska <mliska@suse.cz>
38354 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38355 for {cd}tors and C++ {cd}tors.
38357 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38359 * config/s390/s390.md ("*setmem_long_and")
38360 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38362 2017-01-26 Martin Sebor <msebor@redhat.com>
38364 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38367 2017-01-26 Martin Sebor <msebor@redhat.com>
38369 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38370 HAVE_DFmode before using XFmode or DFmode.
38371 (parse_directive): Avoid using the z length modifier to avoid
38372 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38374 PR middle-end/78703
38375 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38376 to accept adjustment as an array.
38377 (get_int_range): New function.
38378 (struct directive): Make width and prec arrays.
38379 (directive::set_width, directive::set_precision): Call get_int_range.
38380 (format_integer, format_floating): Handle width and precision ranges.
38381 (format_string, parse_directive): Same.
38383 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38386 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38387 comdat_type_p set, just clone them, but keep the children in the
38391 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38392 which have direct callers with -fvar-tracking-assignments enabled
38394 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38395 inside of type units.
38397 2017-01-26 Martin Sebor <msebor@redhat.com>
38399 PR middle-end/78703
38400 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38402 (struct format_result): Replace number_chars, number_chars_min,
38403 and number_chars_max with a single member of struct result_range.
38405 (format_result::operator+=): Adjust.
38406 (struct fmtresult): Remove bounded. Handle likely and unlikely
38408 (fmtresult::adjust_for_width_or_precision): New function.
38409 (fmtresult:type_max_digits): New function.
38410 (bytes_remaining): Handle likely and unlikely counters.
38411 (min_bytes_remaining): Remove.
38412 (format_percent): Simplify.
38413 (format_integer, format_floating): Set likely and unlikely counters.
38414 (get_string_length, format_character, format_string): Same.
38415 (format_plain, should_warn_p): New function.
38416 (maybe_warn): Call should_warn_p. Update diagnostic messages
38417 and handle those for all directives, including plain strings.
38418 (format_directive): Handle likely and unlikely counters.
38419 Remove unnecessary quoting from diagnostics. Add an informational
38421 (add_bytes): Remove.
38422 (pass_sprintf_length::compute_format_length): Simplify.
38423 (try_substitute_return_value): Handle likely and unlikely counters.
38425 2017-01-26 Carl Love <cel@us.ibm.com>
38427 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38428 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38430 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38433 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38434 endianess for subregs into account.
38435 * lra-constraints.c (lra_constraints): Do risky transformations
38436 always on the first iteration.
38437 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38438 last_call_used_reg_set.
38439 (process_bb_lives): Define and use last_call_used_reg_set.
38440 * lra.c (lra): Always continue after lra_constraints on the first
38443 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38445 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38447 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38449 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38451 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38452 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38453 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38454 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38455 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38456 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38457 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38458 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38459 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38461 2017-01-26 Marek Polacek <polacek@redhat.com>
38464 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38465 for the third operand.
38467 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38469 PR middle-end/79236
38470 * omp-low.c (struct omp_context): Add simt_stmt field.
38471 (scan_omp_for): Return omp_context *.
38472 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38473 context to the _simt_ SIMD stmt.
38474 (lower_omp_for): For combined SIMD with sibling _simt_
38475 SIMD, make sure to use the same decls in _looptemp_
38476 clauses as in the sibling.
38478 2017-01-26 David Sherwood <david.sherwood@arm.com>
38480 PR middle-end/79212
38481 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38484 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38487 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38488 emit fld b; fld a; if possible.
38490 * brig-builtins.def: Update copyright years.
38491 * config/arm/arm_acle_builtins.def: Update copyright years.
38493 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38496 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38497 constraint instead of o for the stxsd instruction.
38499 2017-01-25 Carl Love <cel@us.ibm.com>
38501 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38502 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38504 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38506 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38508 2017-01-25 Richard Biener <rguenther@suse.de>
38510 PR tree-optimization/69264
38511 * target.def (vector_alignment_reachable): Improve documentation.
38512 * doc/tm.texi: Regenerate.
38513 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38515 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38516 earlier changes with respect to TYPE_USER_ALIGN.
38517 (vector_alignment_reachable_p): Likewise. Improve dumping.
38519 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38522 * config/arm/arm.md (xordi3): Force constant operand into a register
38525 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38527 * doc/invoke.texi (-fstore-merging): Correct default optimization
38528 levels at which it is enabled.
38529 (-O): Move -fstore-merging from list to...
38532 2017-01-25 Richard Biener <rguenther@suse.de>
38535 * omp-expand.c: Include debug.h.
38536 (expand_omp_taskreg): Make sure to generate early debug before
38537 outlining anything from a function.
38538 (expand_omp_target): Likewise.
38539 (grid_expand_target_grid_body): Likewise.
38541 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38544 * asan.c (get_translation_unit_decl): New function.
38545 (asan_add_global): Extract modules file name from globals
38546 TRANSLATION_UNIT_DECL name.
38548 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38551 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38552 for long calls with APCS frame and VFP.
38554 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38556 * cfg.c (original_copy_tables_initialized_p): New function.
38557 * cfg.h (original_copy_tables_initialized_p): New decl.
38558 * cfgrtl.c (relink_block_chain): Guard the call to
38559 free_original_copy_tables with a call to
38560 original_copy_tables_initialized_p.
38561 * cgraph.h (symtab_node::native_rtl_p): New decl.
38562 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38563 (symtab_node::needed_p): Don't assert for early assembly output
38564 for __RTL functions.
38565 (cgraph_node::finalize_function): Set "force_output" for __RTL
38567 (cgraph_node::analyze): Bail out early for __RTL functions.
38568 (analyze_functions): Update assertion to support __RTL functions.
38569 (cgraph_node::expand): Bail out early for __RTL functions.
38570 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38572 * function.h (struct function): Update comment for field
38574 * gimple-expr.c: Include "tree-pass.h".
38575 (gimple_has_body_p): Return false for __RTL functions.
38576 * Makefile.in (OBJS): Add run-rtl-passes.o.
38577 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38579 (gcc::pass_manager::get_clean_slate): New accessor.
38580 * passes.c: Include "insn-addr.h".
38581 (should_skip_pass_p): Add logging. Update logic for running
38582 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38583 property-provider override so it is only done for gimple passes.
38585 (skip_pass): New function.
38586 (execute_one_pass): Call skip_pass when skipping passes.
38587 * read-md.c (md_reader::read_char): Support filtering
38588 the input to a subset of line numbers.
38589 (md_reader::md_reader): Initialize fields
38590 m_first_line and m_last_line.
38591 (md_reader::read_file_fragment): New function.
38592 * read-md.h (md_reader::read_file_fragment): New decl.
38593 (md_reader::m_first_line): New field.
38594 (md_reader::m_last_line): New field.
38595 * read-rtl-function.c (function_reader::create_function): Only
38596 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38597 curr_properties. Set DECL_INITIAL to a dummy block.
38598 (read_rtl_function_body_from_file_range): New function.
38599 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38601 * run-rtl-passes.c: New file.
38602 * run-rtl-passes.h: New file.
38604 2017-01-24 Jeff Law <law@redhat.com>
38606 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38609 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38611 PR tree-optimization/79159
38612 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38613 (record_nonwrapping_iv): Improve boundary using above function if no
38614 value range information.
38616 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38617 Martin Jambor <mjambor@suse.cz>
38619 * brig-builtins.def: New file.
38620 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38621 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38622 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38623 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38624 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38625 * builtin-types.def (BT_INT8): New.
38626 (BT_INT16): Likewise.
38627 (BT_UINT8): Likewise.
38628 (BT_UINT16): Likewise.
38629 (BT_FN_ULONG): Likewise.
38630 (BT_FN_UINT_INT): Likewise.
38631 (BT_FN_UINT_ULONG): Likewise.
38632 (BT_FN_UINT_LONG): Likewise.
38633 (BT_FN_UINT_PTR): Likewise.
38634 (BT_FN_ULONG_PTR): Likewise.
38635 (BT_FN_INT8_FLOAT): Likewise.
38636 (BT_FN_INT16_FLOAT): Likewise.
38637 (BT_FN_UINT32_FLOAT): Likewise.
38638 (BT_FN_UINT16_FLOAT): Likewise.
38639 (BT_FN_UINT8_FLOAT): Likewise.
38640 (BT_FN_UINT64_FLOAT): Likewise.
38641 (BT_FN_UINT16_UINT32): Likewise.
38642 (BT_FN_UINT32_UINT16): Likewise.
38643 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38644 (BT_FN_INT_PTR_INT): Likewise.
38645 (BT_FN_UINT_PTR_UINT): Likewise.
38646 (BT_FN_LONG_PTR_LONG): Likewise.
38647 (BT_FN_ULONG_PTR_ULONG): Likewise.
38648 (BT_FN_VOID_UINT64_UINT64): Likewise.
38649 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38650 (BT_FN_INT8_INT8_INT8): Likewise.
38651 (BT_FN_INT16_INT16_INT16): Likewise.
38652 (BT_FN_INT_INT_INT): Likewise.
38653 (BT_FN_UINT_FLOAT_UINT): Likewise.
38654 (BT_FN_FLOAT_UINT_UINT): Likewise.
38655 (BT_FN_ULONG_UINT_UINT): Likewise.
38656 (BT_FN_ULONG_UINT_PTR): Likewise.
38657 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38658 (BT_FN_UINT_UINT_UINT): Likewise.
38659 (BT_FN_VOID_UINT_PTR): Likewise.
38660 (BT_FN_UINT_UINT_PTR: Likewise.
38661 (BT_FN_UINT32_UINT64_PTR): Likewise.
38662 (BT_FN_INT_INT_UINT_UINT): Likewise.
38663 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38664 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38665 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38666 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38667 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38668 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38669 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38670 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38671 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38672 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38673 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38674 * doc/frontends.texi: List BRIG FE.
38675 * doc/install.texi (Testing): Add BRIG tesring requirements.
38676 * doc/invoke.texi (Overall Options): Mention BRIG.
38677 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38679 2017-01-24 Richard Biener <rguenther@suse.de>
38681 PR translation/79208
38682 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38684 2017-01-24 Martin Jambor <mjambor@suse.cz>
38687 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38688 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38689 and known_contexts.
38691 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38693 PR middle-end/79123
38694 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38695 casts from signed to unsigned really don't have a range.
38697 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38699 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38700 GMP_RNDx for compatiblity.
38702 2017-01-24 Martin Liska <mliska@suse.cz>
38705 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38706 that would prevent us to call alloca with -1 as argument.
38708 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38710 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38711 Avoid -Wformat-security warning.
38713 2017-01-23 Andrew Pinski <apinski@cavium.com>
38715 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38718 2017-01-23 Martin Sebor <msebor@redhat.com>
38720 PR middle-end/78703
38721 * gimple-ssa-sprintf.c (warn_level): New global.
38722 (format_integer): Use it here and throughout the rest of the file.
38723 Use the same switch to compute sign as base.
38724 (maybe_warn): New function.
38725 (format_directive): Factor out warnings into maybe_warn.
38726 Add debugging output. Use warn_level.
38727 (add_bytes): Use warn_level.
38728 (pass_sprintf_length::compute_format_length): Add debugging output.
38729 (try_substitute_return_value): Same.
38730 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38732 PR middle-end/78703
38733 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38734 (struct fmtresult, format_integer, format_floating): Adjust.
38735 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38736 (get_string_length, format_string,format_directive): Same.
38737 (pass_sprintf_length::compute_format_length): Same.
38738 (try_substitute_return_value): Simplify slightly.
38740 PR middle-end/78703
38741 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38742 (fmtresult::operator+=): Outlined.
38743 (struct fmtresult): Add ctors.
38744 (struct conversion_spec): Rename...
38745 (struct directive): ...to this. Add and remove data members.
38746 (directive::set_width, directive::set_precision): New functions.
38747 (format_percent): Use fmtresult ctor.
38748 (get_width_and_precision): Remove.
38749 (format_integer): Make naming changes. Avoid computing width and
38751 (format_floating): Same. Adjust indentation.
38752 (format_character, format_none): New functions.
38753 (format_string): Moved character handling to format_character.
38754 (format_directive): Remove arguments, change return type.
38755 (parse_directive): New function.
38756 (pass_sprintf_length::compute_format_length): Move directive
38757 parsing to parse_directive.
38759 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38761 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38762 (assign_assembler_name_if_needed): ... this.
38763 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38764 (assign_assembler_name_if_needed): ... this.
38765 (free_lang_data_in_cgraph): Adjust callers.
38766 * cgraphunit.c (cgraph_node::analyze): Likewise.
38767 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38770 2017-01-23 Richard Biener <rguenther@suse.de>
38772 PR tree-optimization/79088
38773 PR tree-optimization/79188
38774 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38775 resetting loop bounds after last path deletion. Reset loop
38776 bounds of the target loop, make code match the comments.
38777 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38778 Make sure loops need no fixups.
38780 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38782 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38783 exponent support with double type for first argument.
38784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38785 type returned by __builtin_vec_extract_sig,
38786 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38787 functions from "vector int" to "vector unsigned int" or from
38788 "vector long long int" to "vector unsigned long long int".
38789 Changed type returned by __builtin_vec_extract_exp,
38790 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38791 functions from "vector int" to "vector unsigned int" or from
38792 "vector long long int" to "vector unsigned long long int".
38793 Changed return type of __builtin_vec_test_data_class,
38794 __builtin_vec_test_data_class_sp, and
38795 __builtin_vec_test_data_class_dp from "vector int" to
38796 "vector bool int" or from "vector long long int" to "vector bool
38797 long long int" and changed second argument type from "unsigned
38798 int" to "int". Added new overloaded function forms "vector float
38799 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38800 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38801 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38802 double, vector unsigned long long int)" and "vector double
38803 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38804 long int)". Changed return type of
38805 __builtin_scalar_test_data_class and
38806 __builtin_scalar_test_data_class_sp and
38807 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38808 int" and changed second argument from "unsigned int" to "int".
38809 Changed type returned by __builtin_scalar_test_neg,
38810 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38811 from "int" to "bool int". Added new overloaded function form
38812 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38813 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38814 exponent double-precision with floating point first argument.
38815 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38816 documentation of scalar_test_data_class, scalar_test_neg,
38817 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38818 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38819 vec_test_data_class built-in functions to reflect refinements in
38820 their type signatures.
38822 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38824 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38826 (aarch64_elf_asm_destructor): Likewise.
38828 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38830 PR rtl-optimization/78634
38831 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38832 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38833 * ifcvt.c (noce_try_cmove): Add missing cost check.
38835 PR rtl-optimization/71724
38836 * combine.c (if_then_else_cond): Look for situations where it is
38837 beneficial to undo the work of one of the recursive calls.
38839 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38841 PR tree-optimization/70754
38842 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38843 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38844 combined stmt before it if not NULL.
38845 (combine_chains): Process refs reversely and compute dominance point
38848 2017-01-23 Martin Liska <mliska@suse.cz>
38850 PR tree-optimization/79196
38851 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38852 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38854 (strlen_optimize_stmt): Call the renamed function.
38856 2017-01-23 Michael Matz <matz@suse.de>
38858 PR tree-optimization/78384
38859 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38861 2017-01-23 Richard Biener <rguenther@suse.de>
38863 PR tree-optimization/79186
38864 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38865 both incoming edges before moving an assert.
38867 2017-01-23 Martin Jambor <mjambor@suse.cz>
38869 * ipa-prop.c (load_from_param_1): Removed.
38870 (load_from_unmodified_param): Bits from load_from_param_1 put back
38872 (load_from_param): Removed.
38873 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38874 with stmt. Reverted back to use of load_from_unmodified_param.
38876 2017-01-23 Martin Jambor <mjambor@suse.cz>
38879 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38880 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38881 field a pointer to garbage collected vector, mark lattices and
38882 ipcp_orig_node with GTY((skip)).
38883 (ipa_get_param_count): Adjust to descriptors being a pointer.
38884 (ipa_get_param): Likewise.
38885 (ipa_get_type): Likewise.
38886 (ipa_get_param_move_cost): Likewise.
38887 (ipa_set_param_used): Likewise.
38888 (ipa_get_controlled_uses): Likewise.
38889 (ipa_set_controlled_uses): Likewise.
38890 (ipa_is_param_used): Likewise.
38891 (ipa_node_params_t): Move into garbage collector. New methods insert
38893 (ipa_node_params_sum): Annotate wth GTY(()).
38894 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38896 (ipa_load_from_parm_agg): Adjust declaration.
38897 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38898 * ipa-profile.c (ipa_profile): Likewise.
38899 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38900 (ipa_populate_param_decls): Make descriptors parameter garbage
38902 (ipa_dump_param): Adjust to descriptors being a pointer.
38903 (ipa_alloc_node_params): Likewise.
38904 (ipa_initialize_node_params): Likewise.
38905 (load_from_param_1): Make descriptors parameter garbage collected.
38906 (load_from_unmodified_param): Likewise.
38907 (load_from_param): Likewise.
38908 (ipa_load_from_parm_agg): Likewise.
38909 (ipa_node_params::~ipa_node_params): Removed.
38910 (ipa_free_all_node_params): Remove call to delete operator.
38911 (ipa_node_params_t::insert): New.
38912 (ipa_node_params_t::remove): Likewise.
38913 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38914 copy known_csts and known_contexts vectors.
38915 (ipa_read_node_info): Adjust to descriptors being a pointer.
38916 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38918 (ipcp_transform_function): Make descriptors variable garbage
38921 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38923 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38924 * config/i386/avx512dqintrin.h: Ditto.
38925 * config/i386/avx512fintrin.h: Ditto.
38926 * config/i386/i386.c: Handle new builtins.
38927 * config/i386/i386-builtin.def: Add new builtins.
38928 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38929 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38931 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38932 Martin Liska <mliska@suse.cz>
38934 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38935 * asan.c (asan_expand_poison_ifn): Support stores and use
38936 appropriate ASAN report function.
38937 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38938 * internal-fn.def (ASAN_POISON_USE): Declare.
38939 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38940 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38941 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38942 ASAN_POISON calls w/o LHS.
38943 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38944 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38945 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38946 * gimplify.c (asan_poison_variables): Add attribute
38947 use_after_scope_memory to variables that really needs to live
38949 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38950 having the attribute.
38952 2017-01-23 Martin Liska <mliska@suse.cz>
38954 * asan.c (create_asan_shadow_var): New function.
38955 (asan_expand_poison_ifn): Likewise.
38956 * asan.h (asan_expand_poison_ifn): New declaration.
38957 * internal-fn.c (expand_ASAN_POISON): Likewise.
38958 * internal-fn.def (ASAN_POISON): New builtin.
38959 * sanopt.c (pass_sanopt::execute): Expand
38960 asan_expand_poison_ifn.
38961 * tree-inline.c (copy_decl_for_dup_finish): Make function
38963 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38964 * tree-ssa.c (is_asan_mark_p): New function.
38965 (execute_update_addresses_taken): Rewrite local variables
38966 (identified just by use-after-scope as addressable) into SSA.
38968 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38970 * doc/install.texi (Specific): opensource.apple.com uses https
38971 now. Remove trailing slash.
38973 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38975 * README.Portability: Remove note on an Irix compatibility issue.
38977 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38979 * gcov.c (INCLUDE_ALGORITHM): Define.
38980 (INCLUDE_VECTOR): Define.
38981 No longer include <vector> and <algorithm> directly.
38983 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38985 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38987 * doc/invoke.texi (Code Gen Options): Ditto.
38989 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38992 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38994 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38996 rtl-optimization/79125
38997 * cprop.c (local_cprop_pass): Handle cases where we make an
38998 unconditional trap.
39000 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39004 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39005 read from, for big endian.
39007 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39009 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39010 register pauth builtins for LP64 only.
39012 2017-01-20 Marek Polacek <polacek@redhat.com>
39015 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39018 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39020 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39022 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39023 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39024 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39026 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39029 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39030 in big-endian mode when they are not a single duplicated value.
39032 2017-01-20 Richard Biener <rguenther@suse.de>
39034 * BASE-VER: Bump to 7.0.1.
39036 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39038 * omp-low.c (omplow_simd_context): New struct. Use it...
39039 (lower_rec_simd_input_clauses): ...here and...
39040 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39041 references to idx, lane, max_vf, is_simt.
39043 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39045 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39048 2017-01-20 Martin Jambor <mjambor@suse.cz>
39050 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39051 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39053 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39054 (GTFILES): Rename hsa.c to hsa-common.c.
39055 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39056 * hsa-dump.c: Likewise.
39057 * hsa-gen.c: Likewise.
39058 * hsa-regalloc.c: Likewise.
39059 * ipa-hsa.c: Likewise.
39060 * omp-expand.c: Likewise.
39061 * omp-low.c: Likewise.
39062 * toplev.c: Likewise.
39064 2017-01-20 Marek Polacek <polacek@redhat.com>
39067 * doc/invoke.texi: Document -Wduplicated-branches.
39068 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39069 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39070 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39071 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39072 return 0 only when not OEP_LEXICOGRAPHIC.
39073 (fold_build_cleanup_point_expr): Use the expression
39074 location when building CLEANUP_POINT_EXPR.
39075 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39076 * tree.c (add_expr): Handle error_mark_node.
39078 2017-01-20 Martin Liska <mliska@suse.cz>
39081 * tree-profile.c (init_ic_make_global_vars): Do not call
39083 (gimple_init_gcov_profiler): Likewise.
39085 2017-01-20 Martin Liska <mliska@suse.cz>
39088 * cgraph.h (maybe_create_reference): Remove argument and
39090 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39092 * ipa-cp.c (create_specialized_node): Likewise.
39093 * symtab.c (symtab_node::maybe_create_reference): Handle
39094 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39096 2017-01-20 Martin Liska <mliska@suse.cz>
39098 * read-rtl-function.c (function_reader::create_function): Use
39099 build_decl instread of build_decl_stat.
39101 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39103 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39104 * config/i386/avx512dqintrin.h: Ditto.
39105 * config/i386/avx512fintrin.h: Ditto.
39106 * config/i386/i386-builtin-types.def: Add new types.
39107 * config/i386/i386.c: Handle new types.
39108 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39109 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39110 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39111 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39112 (__builtin_ia32_kshiftridi): New.
39113 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39115 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39119 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39120 define to rs6000_init_stack_protect_guard.
39121 (rs6000_init_stack_protect_guard): New function.
39123 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39124 Yunqiang Su <yunqiang.su@imgtec.com>
39126 * config.gcc (supported_defaults): Add madd4.
39127 (with_madd4): Add validation.
39128 (all_defaults): Add madd4.
39129 * config/mips/mips.opt (mmadd4): New option.
39130 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39132 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39134 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39135 (ISA_HAS_FUSED_MADD4): Likewise.
39136 * doc/invoke.texi (-mmadd4): Document the new option.
39137 * doc/install.texi (--with-madd4): Document the new option.
39139 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39141 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39142 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39143 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39144 (aarch64_init_pauth_hint_builtins): New.
39145 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39146 (aarch64_expand_builtin): Expand new builtins.
39148 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39150 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39151 * combine-stack-adj.c (no_unhandled_cfa): Handle
39152 REG_CFA_TOGGLE_RA_MANGLE.
39153 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39154 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39155 info for return address signing.
39156 (aarch64_expand_epilogue): Likewise.
39158 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39160 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39161 * config/aarch64/aarch64-protos.h
39162 (aarch64_return_address_signing_enabled): New declaration.
39163 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39165 (aarch64_expand_prologue): Sign return address before it's pushed onto
39167 (aarch64_expand_epilogue): Authenticate return address fetched from
39169 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39170 (aarch64_attributes): New function attributes for "sign-return-address".
39171 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39172 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39173 ("*do_return"): Generate combined instructions according to key index.
39174 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39175 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39177 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39178 * config/aarch64/aarch64.opt (msign-return-address=): New.
39179 * doc/extend.texi (AArch64 Function Attributes): Documents
39180 "sign-return-address=".
39181 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39183 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39185 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39186 overall option summary.
39188 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39190 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39191 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39192 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39193 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39195 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39197 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39198 -mpower9-minmax by default for -mcpu=power9.
39199 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39200 128-bit floating point.
39202 2017-01-20 Alan Modra <amodra@gmail.com>
39204 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39205 optimizing for size.
39207 2017-01-20 Alan Modra <amodra@gmail.com>
39210 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39211 for strcmp and strncmp from corresponding builtin decl.
39213 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39215 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39216 instead of i386/rtems-64.h.
39217 * config/i386/rtems-64.h: Remove.
39219 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39223 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39225 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39227 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39229 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39230 Change int to HOST_WIDE_INT.
39231 * config/aarch64/aarch64-protos.h
39232 (aarch64_simd_gen_const_vector_dup): Likewise.
39233 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39235 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39237 * langhooks-def.h (lhd_type_for_size): New decl.
39238 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39239 * langhooks.c (lhd_type_for_size): New function, taken from
39242 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39244 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39245 define_bypass for CR latency.
39246 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39247 (power9-alu2): Add define_bypass for CR latency.
39249 (power9-mul): Update insn latency.
39250 (power9-mul-compare): Update insn latency, bypass latency and remove
39253 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39255 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39257 * config/aarch64/aarch64.md
39258 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39259 aarch64_nopcrelative_literal_loads.
39260 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39262 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39264 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39265 TARGET_LOONGSON_3A.
39266 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39268 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39271 * config.gcc (supported_defaults): Add lxc1-sxc1.
39272 (with_lxc1_sxc1): Add validation.
39273 (all_defaults): Add lxc1-sxc1.
39274 * config/mips/mips.opt (mlxc1-sxc1): New option.
39275 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39277 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39278 __mips_no_lxc1_sxc1.
39279 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39280 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39281 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39283 2017-01-19 Richard Biener <rguenther@suse.de>
39285 PR tree-optimization/72488
39286 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39287 sure to restore SSA info.
39288 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39290 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39292 PR rtl-optimization/79121
39293 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39294 of the inner type when shifting an extended value.
39296 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39299 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39300 interposable aliases.
39302 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39305 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39306 Use the evmergelohi instruction.
39307 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39308 (mov_sitf_e500_subreg8_2_be): Likewise.
39309 (mov_sitf_e500_subreg12_2_le): Likewise.
39310 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39311 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39312 (mov_sitf_e500_subreg8_2_le): Likewise.
39313 (mov_sitf_e500_subreg12_2_be): Likewise.
39315 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39317 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39318 attribute from vecsimple to vecperm.
39319 (altivec_vbpermq2): Likewise.
39321 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39324 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39326 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39327 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39328 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39329 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39330 case where N arg is SIZE_MAX.
39331 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39332 (cmpstrsi): Add pattern.
39334 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39336 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39337 __builtin_vec_revb builtins.
39338 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39339 built-in functions to support generation of the ISA 3.0 XXBR<x>
39340 vector byte reverse instructions.
39341 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39342 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39343 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39344 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39345 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39346 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39347 (P9V_BUILTIN_VEC_REVB): Likewise.
39348 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39349 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39350 (p9_xxbrq_v16qi): Likewise.
39351 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39352 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39353 (p9_xxbrh_v8hi): Likewise.
39354 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39355 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39356 vec_revb built-in functions.
39358 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39360 PR rtl-optimization/78952
39361 * config/i386/i386.md (any_extract): New code iterator.
39362 (*insvqi_2): Use any_extract for source operand.
39363 (*insvqi_3): Use any_shiftrt for source operand.
39365 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39367 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39369 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39371 2017-01-18 Matthias Klose <doko@ubuntu.com>
39373 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39375 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39377 * config/rs6000/altivec.h (vec_bperm): Change #define.
39378 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39379 (altivec_vbpermq2): New define_insn.
39380 (altivec_vbpermd): Likewise.
39381 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39382 function interface.
39383 (VBPERMD): Likewise.
39384 (VBPERM): New polymorphic function interface.
39385 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39386 Add entries for P9V_BUILTIN_VEC_VBPERM.
39387 * doc/extend.texi: Add interfaces for vec_bperm.
39389 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39391 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39392 first letter of error messages.
39393 (s390_resolve_overloaded_builtin): Likewise.
39394 * config/s390/s390.c (s390_expand_builtin): Likewise.
39395 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39396 (s390_valid_target_attribute_inner_p): Likewise.
39397 * config/s390/s390.md ("tabort"): Likewise.
39399 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39401 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39402 (ISA_AVOID_DIV_HILO): New macro.
39403 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39404 (ISA_HAS_DDIV): Likewise.
39406 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39408 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39410 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39412 * doc/invoke.texi (fabi-version): Spelling fix.
39414 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39417 * doc/invoke.texi (fabi-version): Mention mangling fix for
39420 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39423 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39425 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39428 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39429 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39431 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39433 (stack_protect_test): Ditto.
39434 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39435 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39437 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39438 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39439 -mstack-protector-guard-offset=.
39440 (RS/6000 and PowerPC Options): Ditto.
39442 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39444 * config/i386/i386.h (MASK_CLASS_P): New define.
39445 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39446 there are no registers from different register sets also when
39447 mask registers are used. Update function comment.
39448 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39449 to (*k/*r) and (*k/*km) alternatives.
39451 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39453 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39454 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39455 (EH_RETURN_HANDLER_RTX): New define.
39456 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39457 Force frame pointer in EH return functions.
39458 (aarch64_expand_epilogue): Add barrier for eh_return.
39459 (aarch64_final_eh_return_addr): Remove.
39460 (aarch64_eh_return_handler_rtx): New function.
39461 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39463 (aarch64_eh_return_handler_rtx): New prototype.
39465 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39467 * config/rs6000/altivec.h (vec_rlmi): New #define.
39468 (vec_vrlnm): Likewise.
39469 (vec_rlnm): Likewise.
39470 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39471 (UNSPEC_VRLNM): Likewise.
39472 (VIlong): New mode iterator.
39473 (altivec_vrl<VI_char>mi): New define_insn.
39474 (altivec_vrl<VI_char>nm): Likewise.
39475 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39477 (VRLDNM): Likewise.
39478 (RLNM): New polymorphic function entry.
39479 (VRLWMI): New monomorphic function entry.
39480 (VRLDMI): Likewise.
39481 (RLMI): New polymorphic function entry.
39482 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39483 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39484 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39487 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39490 * dwarf2out.c (field_byte_offset): Restore the
39491 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39492 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39493 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39496 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39499 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39501 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39504 * dwarf2out.c (add_data_member_location_attribute): For constant
39505 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39506 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39507 DW_AT_byte_size attributes.
39509 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39511 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39512 after forcing to constant memory when the code model is medium.
39514 2017-01-17 Julia Koval <julia.koval@intel.com>
39517 * config/i386/avx512fintrin.h
39518 (_mm512_i32gather_ps): Change __addr type to void const*.
39519 (_mm512_mask_i32gather_ps): Ditto.
39520 (_mm512_i32gather_pd): Ditto.
39521 (_mm512_mask_i32gather_pd): Ditto.
39522 (_mm512_i64gather_ps): Ditto.
39523 (_mm512_mask_i64gather_ps): Ditto.
39524 (_mm512_i64gather_pd): Ditto.
39525 (_mm512_mask_i64gather_pd): Ditto.
39526 (_mm512_i32gather_epi32): Ditto.
39527 (_mm512_mask_i32gather_epi32): Ditto.
39528 (_mm512_i32gather_epi64): Ditto.
39529 (_mm512_mask_i32gather_epi64): Ditto.
39530 (_mm512_i64gather_epi32): Ditto.
39531 (_mm512_mask_i64gather_epi32): Ditto.
39532 (_mm512_i64gather_epi64): Ditto.
39533 (_mm512_mask_i64gather_epi64): Ditto.
39534 (_mm512_i32scatter_ps): Change __addr type to void*.
39535 (_mm512_mask_i32scatter_ps): Ditto.
39536 (_mm512_i32scatter_pd): Ditto.
39537 (_mm512_mask_i32scatter_pd): Ditto.
39538 (_mm512_i64scatter_ps): Ditto.
39539 (_mm512_mask_i64scatter_ps): Ditto.
39540 (_mm512_i64scatter_pd): Ditto.
39541 (_mm512_mask_i64scatter_pd): Ditto.
39542 (_mm512_i32scatter_epi32): Ditto.
39543 (_mm512_mask_i32scatter_epi32): Ditto.
39544 (_mm512_i32scatter_epi64): Ditto.
39545 (_mm512_mask_i32scatter_epi64): Ditto.
39546 (_mm512_i64scatter_epi32): Ditto.
39547 (_mm512_mask_i64scatter_epi32): Ditto.
39548 (_mm512_i64scatter_epi64): Ditto.
39549 (_mm512_mask_i64scatter_epi64): Ditto.
39550 * config/i386/avx512pfintrin.h
39551 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39552 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39553 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39554 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39555 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39556 (_mm512_prefetch_i32scatter_ps): Ditto.
39557 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39558 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39559 (_mm512_prefetch_i64scatter_pd): Ditto.
39560 (_mm512_prefetch_i64scatter_ps): Ditto.
39561 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39562 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39563 * config/i386/avx512vlintrin.h
39564 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39565 (_mm_mmask_i32gather_ps): Ditto.
39566 (_mm256_mmask_i32gather_pd): Ditto.
39567 (_mm_mmask_i32gather_pd): Ditto.
39568 (_mm256_mmask_i64gather_ps): Ditto.
39569 (_mm_mmask_i64gather_ps): Ditto.
39570 (_mm256_mmask_i64gather_pd): Ditto.
39571 (_mm_mmask_i64gather_pd): Ditto.
39572 (_mm256_mmask_i32gather_epi32): Ditto.
39573 (_mm_mmask_i32gather_epi32): Ditto.
39574 (_mm256_mmask_i32gather_epi64): Ditto.
39575 (_mm_mmask_i32gather_epi64): Ditto.
39576 (_mm256_mmask_i64gather_epi32): Ditto.
39577 (_mm_mmask_i64gather_epi32): Ditto.
39578 (_mm256_mmask_i64gather_epi64): Ditto.
39579 (_mm_mmask_i64gather_epi64): Ditto.
39580 (_mm256_i32scatter_ps): Change __addr type to void*.
39581 (_mm256_mask_i32scatter_ps): Ditto.
39582 (_mm_i32scatter_ps): Ditto.
39583 (_mm_mask_i32scatter_ps): Ditto.
39584 (_mm256_i32scatter_pd): Ditto.
39585 (_mm256_mask_i32scatter_pd): Ditto.
39586 (_mm_i32scatter_pd): Ditto.
39587 (_mm_mask_i32scatter_pd): Ditto.
39588 (_mm256_i64scatter_ps): Ditto.
39589 (_mm256_mask_i64scatter_ps): Ditto.
39590 (_mm_i64scatter_ps): Ditto.
39591 (_mm_mask_i64scatter_ps): Ditto.
39592 (_mm256_i64scatter_pd): Ditto.
39593 (_mm256_mask_i64scatter_pd): Ditto.
39594 (_mm_i64scatter_pd): Ditto.
39595 (_mm_mask_i64scatter_pd): Ditto.
39596 (_mm256_i32scatter_epi32): Ditto.
39597 (_mm256_mask_i32scatter_epi32): Ditto.
39598 (_mm_i32scatter_epi32): Ditto.
39599 (_mm_mask_i32scatter_epi32): Ditto.
39600 (_mm256_i32scatter_epi64): Ditto.
39601 (_mm256_mask_i32scatter_epi64): Ditto.
39602 (_mm_i32scatter_epi64): Ditto.
39603 (_mm_mask_i32scatter_epi64): Ditto.
39604 (_mm256_i64scatter_epi32): Ditto.
39605 (_mm256_mask_i64scatter_epi32): Ditto.
39606 (_mm_i64scatter_epi32): Ditto.
39607 (_mm_mask_i64scatter_epi32): Ditto.
39608 (_mm256_i64scatter_epi64): Ditto.
39609 (_mm256_mask_i64scatter_epi64): Ditto.
39610 (_mm_i64scatter_epi64): Ditto.
39611 (_mm_mask_i64scatter_epi64): Ditto.
39612 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39613 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39614 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39615 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39616 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39617 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39618 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39619 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39620 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39621 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39622 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39623 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39624 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39625 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39626 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39627 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39628 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39629 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39630 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39631 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39632 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39633 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39634 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39635 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39636 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39637 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39638 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39639 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39640 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39641 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39642 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39643 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39644 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39645 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39646 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39647 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39648 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39649 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39650 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39651 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39652 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39653 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39654 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39655 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39656 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39657 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39658 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39659 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39660 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39661 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39662 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39663 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39664 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39665 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39666 definitions accordingly.
39668 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39669 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39672 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39675 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39678 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39679 hard regs for inner regno.
39681 2017-01-17 Martin Liska <mliska@suse.cz>
39684 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39685 assumption and add comment.
39687 2017-01-17 Nathan Sidwell <nathan@acm.org>
39689 * ipa-visibility.c (localize_node): New function, broken out of ...
39690 (function_and_variable_visibility): ... here. Call it.
39692 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39694 PR middle-end/77445
39695 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39696 correctly set frequency of oudgoing edge.
39697 (duplicate_thread_path): Fix profile updating.
39699 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39702 * configure.ac: Add GCC_BASE_VER.
39703 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39704 version from BASE-VER file.
39705 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39706 (gcc.o): Depend on $(BASEVER).
39707 * common.opt (dumpfullversion): New option.
39708 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39709 * doc/invoke.texi: Document -dumpfullversion.
39710 * doc/install.texi: Document --with-gcc-major-version-only.
39711 * configure: Regenerated.
39713 2017-01-17 Richard Biener <rguenther@suse.de>
39715 PR tree-optimization/71433
39716 * tree-vrp.c (register_new_assert_for): Merge same asserts
39717 on all incoming edges.
39718 (process_assert_insertions_for): Handle insertions at the
39721 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39723 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39724 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39726 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39729 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39732 2017-01-17 Alan Modra <amodra@gmail.com>
39735 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39736 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39737 symbolic stack limit when pic.
39739 2017-01-16 Martin Sebor <msebor@redhat.com>
39741 PR tree-optimization/78608
39742 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39744 2017-01-16 Jeff Law <law@redhat.com>
39747 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39748 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39749 for several include directories that may be relative to sysroot.
39750 * config/i386/x-mingw32 (gplus_includedir): Define.
39751 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39752 (native_system_includedir): Likewise.
39753 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39754 override if TARGET_SYSTEM_ROOT is defined.
39755 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39757 PR tree-optimization/79090
39758 PR tree-optimization/33562
39759 PR tree-optimization/61912
39760 PR tree-optimization/77485
39761 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39762 and computed trims into the dump file.
39764 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39766 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39768 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39771 * gimplify.c (gimplify_init_constructor): If want_value and
39772 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39776 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39777 sequence. Formatting fixes.
39778 (doloop_optimize): Formatting fixes.
39781 * gcc.c (debug_level_greater_than_spec_func): New function.
39782 (static_spec_functions): Add debug-level-gt spec function.
39783 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39785 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39786 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39787 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39788 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39789 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39790 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39792 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39794 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39795 QImode fixups to general and mask registers only.
39797 2017-01-16 Carl Love <cel@us.ibm.com>
39799 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39800 for built-in functions
39801 vector signed char vec_nabs (vector signed char)
39802 vector signed short vec_nabs (vector signed short)
39803 vector signed int vec_nabs (vector signed int)
39804 vector signed long long vec_nabs (vector signed long long)
39805 vector float vec_nabs (vector float)
39806 vector double vec_nabs (vector double)
39807 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39809 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39810 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39811 * doc/extend.texi: Update the documentation file for the new built-in
39814 2017-01-16 Martin Sebor <msebor@redhat.com>
39816 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39819 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39821 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39822 UNSPEC_VSX__XXSPLTD to require special splat handling.
39824 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39827 * system.h: Poison strndup.
39829 2017-01-16 Alan Modra <amodra@gmail.com>
39832 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39835 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39837 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39839 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39841 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39842 call recog here. Assert that INSN_CODE (insn) is non-negative.
39844 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39847 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39849 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39850 in the currently scheduled RTL fragment.
39852 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39854 PR rtl-optimization/78751
39855 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39858 2017-01-14 Jeff Law <law@redhat.com>
39860 PR tree-optimization/79090
39861 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39862 variable length stores.
39863 (compute_trims): Delete dead assignment to *trim_tail.
39864 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39867 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39869 PR rtl-optimization/78626
39870 PR rtl-optimization/78727
39871 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39872 of a block, and split such blocks after everything else is finished.
39874 2017-01-14 Alan Modra <amodra@gmail.com>
39877 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39878 target legitimate_combined_insn.
39879 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39880 (rs6000_legitimate_combined_insn): New function.
39881 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39883 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39884 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39885 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39887 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39889 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39891 2017-01-13 Jeff Law <law@redhat.com>
39893 PR tree-optimization/33562
39894 PR tree-optimization/61912
39895 PR tree-optimization/77485
39896 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39898 (delete_dead_assignment): Likewise.
39899 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39900 statement to delete_dead_call and delete_dead_assignment.
39902 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39905 * substring-locations.c (format_warning_va): Strengthen case 1 so
39906 that both endpoints of the substring must be within the format
39907 range for just the substring to be printed.
39909 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39911 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39912 * config/i386/i386.c (ix86_target_string): Add missing options
39913 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39914 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39915 flags_other and ix86_target_other to flags2_other. Display unknown
39917 (ix86_valid_target_attribute_inner_p): Add missing options and
39918 reorder options by implied ISAs, as in ix86_target_string.
39920 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39922 * hash-table.h (hash_table::too_empty_p): New function.
39923 (hash_table::expand): Use it.
39924 (hash_table::traverse): Likewise.
39925 (hash_table::empty_slot): Use sizeof (value_type) instead of
39926 sizeof (PTR) to convert bytes to elements. Shrink the table
39927 if the current size is excessive for the current number of
39930 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39932 * ira-costs.c (record_reg_classes): Break from the inner loop
39933 early once alt_fail is known to be true. Update outer loop
39934 handling accordingly.
39936 2017-01-13 Jeff Law <law@redhat.com>
39938 * tree-ssa-dse.c (decrement_count): New function.
39939 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39940 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39941 when we know the partially dead statement is a mem* function.
39943 PR tree-optimization/61912
39944 PR tree-optimization/77485
39945 * tree-ssa-dse.c: Include expr.h.
39946 (maybe_trim_constructor_store): New function.
39947 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39949 PR tree-optimization/33562
39950 PR tree-optimization/61912
39951 PR tree-optimization/77485
39952 * doc/invoke.texi: Document new dse-max-object-size param.
39953 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39954 * tree-ssa-dse.c: Include params.h.
39955 (dse_store_status): New enum.
39956 (initialize_ao_ref_for_dse): New, partially extracted from
39958 (valid_ao_ref_for_dse, normalize_ref): New.
39959 (setup_live_bytes_from_ref, compute_trims): Likewise.
39960 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39961 (maybe_trim_partially_dead_store): Likewise.
39962 (maybe_trim_complex_store): Likewise.
39963 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39964 Track what bytes live from the original store. Return tri-state
39965 for dead, partially dead or live.
39966 (dse_dom_walker): Add constructor, destructor and new private members.
39967 (delete_dead_call, delete_dead_assignment): New extracted from
39969 (dse_optimize_stmt): Make a member of dse_dom_walker.
39970 Use initialize_ao_ref_for_dse.
39972 PR tree-optimization/33562
39973 PR tree-optimization/61912
39974 PR tree-optimization/77485
39975 * sbitmap.h (bitmap_count_bits): Prototype.
39976 (bitmap_clear_range, bitmap_set_range): Likewise.
39977 * sbitmap.c (bitmap_clear_range): New function.
39978 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39980 2017-01-13 Martin Liska <mliska@suse.cz>
39983 * function.c (set_cfun): Add new argument force.
39984 * function.h (set_cfun): Likewise.
39985 * ipa-inline-transform.c (inline_call): Use the function when
39986 strict alising from is dropped for function we inline to.
39988 2017-01-13 Richard Biener <rguenther@suse.de>
39990 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39991 for dumping GIMPLE INTEGER_CSTs.
39993 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39995 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39996 to 201112L since C++17.
39998 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40001 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40002 if -fsanitize=kernel-address is present.
40004 2017-01-13 Richard Biener <rguenther@suse.de>
40006 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40007 as _Literal ( type ) number in case usual suffixes do not
40008 preserve all information.
40010 2017-01-13 Richard Biener <rguenther@suse.de>
40012 PR tree-optimization/77283
40013 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40014 and ssa-iterators.h.
40015 (is_feasible_trace): Implement a cost model based on joiner
40018 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40021 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40022 char or short to __float128/_Float128 directly.
40024 2017-01-12 Martin Sebor <msebor@redhat.com>
40026 to -Wformat-overflow.
40027 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40028 (min_bytes_remaining): Same.
40029 (get_string_length): Same.
40030 (format_string): Same.
40031 (format_directive): Same.
40033 (pass_sprintf_length::handle_gimple_call): Same.
40035 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40037 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40038 info.nowrite calls with no lhs that can't throw. Return bool
40039 whether gsi_remove has been called or not.
40040 (pass_sprintf_length::handle_gimple_call): Return bool whether
40041 try_substitute_return_value called gsi_remove. Formatting fix.
40042 (pass_sprintf_length::execute): Don't use gsi_remove if
40043 handle_gimple_call returned true.
40046 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40047 be removed due to side-effects, don't remove following barrier nor
40048 turn the successor edge into fallthru edge.
40050 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40053 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40054 element-reversing loads and stores as not swappable.
40056 2017-01-12 Nathan Sidwell <nathan@acm.org>
40057 Nicolai Stange <nicstange@gmail.com>
40059 * combine.c (try_combine): Don't ignore result of overlap checking
40060 loop. Combine overlap & asm check into single loop.
40062 2017-01-12 Richard Biener <rguenther@suse.de>
40064 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40065 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40067 2017-01-12 Richard Biener <rguenther@suse.de>
40069 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40070 and TS_TARGET_OPTION directly derive from TS_BASE.
40071 * tree-core.h (tree_optimization_option): Derive from tree_base.
40072 (tree_target_option): Likewise.
40074 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40076 * config/i386/i386.c (memory_address_length): Increase len
40077 only when rip_relative_addr_p returns false.
40079 2017-01-11 Julia Koval <julia.koval@intel.com>
40081 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40082 (OPTION_MASK_ISA_SGX_SET): New.
40083 (ix86_handle_option): Handle OPT_msgx.
40084 * config.gcc: Added sgxintrin.h.
40085 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40086 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40087 * config/i386/i386.c (ix86_target_string): Add -msgx.
40089 (ix86_option_override_internal): Handle new options.
40090 (ix86_valid_target_attribute_inner_p): Add sgx.
40091 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40092 * config/i386/i386.opt: Add msgx.
40093 * config/i386/sgxintrin.h: New file.
40094 * config/i386/x86intrin.h: Add sgxintrin.h.
40096 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40099 * fold-const.c (maybe_nonzero_address): Return 1 for function
40101 (tree_single_nonzero_warnv_p): Don't handle function local objects
40105 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40108 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40111 * opts.c: Include "spellcheck.h"
40112 (struct string_fragment): New struct.
40113 (struct edit_distance_traits<const string_fragment &>): New
40115 (get_closest_sanitizer_option): New function.
40116 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40118 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40120 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40122 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40123 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40124 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40125 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40126 for initial die_offset if dwarf_split_debug_info.
40127 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40128 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40130 (output_skeleton_debug_sections): Formatting fix. Use
40131 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40132 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40134 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40136 * config/arm/cortex-a53.md: Add bypasses for
40137 cortex_a53_r2f_cvt.
40138 (cortex_a53_r2f): Only use for transfers.
40139 (cortex_a53_f2r): Likewise.
40140 (cortex_a53_r2f_cvt): Add reservation for conversions.
40141 (cortex_a53_f2r_cvt): Likewise.
40143 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40145 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40146 to all inlined functions, change static to extern.
40148 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40151 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40153 (arm_assemble_integer): Likewise.
40155 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40157 * config.gcc: Use new awk script to check CPU, FPU and architecture
40158 parameters for --with-... options.
40159 * config/arm/parsecpu.awk: New file
40160 * config/arm/arm-cpus.in: New file.
40161 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40163 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40165 * config/arm/t-arm: Update dependency rules.
40166 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40167 of processing .def files.
40168 * config/arm/genopt.sh: Deleted.
40169 * config/arm/gentune.sh: Deleted.
40170 * config/arm/arm-cores.def: Deleted.
40171 * config/arm/arm-arches.def: Deleted.
40172 * config/arm/arm-fpus.def: Deleted.
40173 * config/arm/arm-tune.md: Regenerated.
40174 * config/arm/arm-tables.opt: Regenerated.
40175 * config/arm/arm-cpu.h: New generated file.
40176 * config/arm/arm-cpu-data.h: New generated file.
40177 * config/arm/arm-cpu-cdata.h: New generated file.
40179 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40182 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40184 (input_varpool_node): Unpack dynamically_initialized bit.
40186 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40188 PR rtl-optimization/79032
40189 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40190 the alignment of the adjusted memory reference against that of MODE,
40191 instead of the alignment of the original memory reference.
40193 2017-01-11 Martin Jambor <mjambor@suse.cz>
40195 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40197 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40198 decorated functions.
40200 2017-01-11 Richard Biener <rguenther@suse.de>
40202 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40203 set range/nonnull info for PHI results. Do not set it on
40204 stmts marked for removal.
40206 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40208 * expr.c (store_field): In the bitfield case, fetch the return value
40209 from the registers before applying a single big-endian adjustment.
40210 Always do a final load for a BLKmode value not larger than a word.
40212 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40215 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40216 that we correctly handle column numbers greater than
40217 LINE_MAP_MAX_COLUMN_NUMBER.
40219 2017-01-10 Martin Sebor <msebor@redhat.com>
40221 PR middle-end/78245
40222 * gimple-ssa-sprintf.c (get_destination_size): Call
40223 {init,fini}object_sizes.
40224 * tree-object-size.c (addr_object_size): Adjust.
40225 (pass_through_call): Adjust.
40226 (pass_object_sizes::execute): Adjust.
40227 * tree-object-size.h (fini_object_sizes): Declare.
40229 2017-01-10 Martin Sebor <msebor@redhat.com>
40231 PR tree-optimization/78775
40232 * builtins.c (get_size_range): Move...
40233 * calls.c: ...to here.
40234 (alloc_max_size): Accept zero argument.
40235 (operand_signed_p): Remove.
40236 (maybe_warn_alloc_args_overflow): Call get_size_range.
40237 * calls.h (get_size_range): Declare.
40239 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40241 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40242 from TI's devices.csv file as of September 2016.
40243 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40245 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40247 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40248 * doc/invoke.texi: Likewise.
40249 * doc/md.texi: Likewise.
40250 * doc/objc.texi: Likewise.
40252 2017-01-10 Joshua Conner <joshconner@google.com>
40254 * config/arm/fuchsia-elf.h: New file.
40255 * config/fuchsia.h: New file.
40256 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40257 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40259 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40261 2016-01-10 Richard Biener <rguenther@suse.de>
40263 PR tree-optimization/79034
40264 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40265 Propagate out degenerate PHIs in the joiner.
40267 2017-01-10 Martin Liska <mliska@suse.cz>
40269 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40270 (sort_congruence_classes_by_decl_uid): Likewise.
40271 (sort_congruence_class_groups_by_decl_uid): Likewise.
40272 (sem_item_optimizer::merge_classes): Sort class, groups in these
40273 classes and members in the groups by DECL_UID of declarations.
40274 This would make merge operations stable.
40276 2017-01-10 Martin Liska <mliska@suse.cz>
40278 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40279 usage of m_classes_vec.
40280 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40281 (sem_item_optimizer::get_group_by_hash): Likewise.
40282 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40283 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40284 (sem_item_optimizer::verify_classes): Likewise.
40285 (sem_item_optimizer::process_cong_reduction): Likewise.
40286 (sem_item_optimizer::dump_cong_classes): Likewise.
40287 (sem_item_optimizer::merge_classes): Likewise.
40288 * ipa-icf.h (congruence_class_hash): Rename from
40289 congruence_class_group_hash. Remove declaration of m_classes_vec.
40291 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40293 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40294 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40295 * config.gcc: Add avx512vpopcntdqintrin.h.
40296 * config/i386/avx512vpopcntdqintrin.h: New.
40297 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40298 * config/i386/i386-builtin-types.def: Add new types.
40299 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40300 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40301 __builtin_ia32_vpopcountq_v8di_mask): New.
40302 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40303 __AVX512VPOPCNTDQ__.
40304 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40305 (PTA_AVX512VPOPCNTDQ): Define.
40306 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40307 TARGET_AVX512VPOPCNTDQ_P): Define.
40308 * config/i386/i386.opt: Add mavx512vpopcntdq.
40309 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40310 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40312 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40314 PR middle-end/77484
40315 * predict.def (PRED_CALL): Set to 67.
40317 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40319 * expr.c (store_field): In the bitfield case, if the value comes from
40320 a function call and is of an aggregate type returned in registers, do
40321 not modify the field mode; extract the value in all cases if the mode
40322 is BLKmode and the size is not larger than a word.
40324 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40327 * config/i386/cpuid.h: Fix undefined behavior.
40329 2017-01-04 Jeff Law <law@redhat.com>
40331 PR tree-optimization/79007
40332 PR tree-optimization/67955
40333 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40334 conservative for pt.null when flag_non_call_exceptions is on.
40336 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40338 PR translation/79019
40339 PR translation/79020
40340 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40341 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40342 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40344 * config/avr/avr.opt (maccumulate-args): Likewise.
40345 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40346 * common.opt (freport-bug): Likewise.
40347 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40348 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40349 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40350 translatable string.
40351 * config/i386/i386.c (function_value_32): Likewise.
40352 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40353 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40355 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40356 * common/config/msp430/msp430-common.c (msp430_handle_option):
40358 * symtab.c (symtab_node::verify_base): Likewise.
40359 * opts.c (set_debug_level): Likewise.
40360 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40362 missing whitespace to translatable strings.
40363 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40364 * config/sh/superh.h: Likewise.
40365 * config/i386/xopintrin.h: Likewise.
40366 * config/i386/znver1.md: Likewise.
40367 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40368 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40369 * double-int.h (struct double_int): Likewise.
40370 * double-int.c (div_and_round_double): Likewise.
40371 * wide-int.cc: Likewise.
40372 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40373 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40374 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40375 (crossjumps_occurred): ... this.
40376 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40379 PR tree-optimization/78899
40380 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40381 returning bool return struct loop *, NULL for failure and the new
40383 (versionable_outer_loop_p): Don't version outer loop if it has
40384 dont_vectorized bit set.
40385 (tree_if_conversion): When versioning outer loop, ensure
40386 tree_if_conversion is performed also on the inner loop of the
40387 non-vectorizable outer loop copy.
40388 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40389 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40390 prevent vectorization of it.
40391 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40392 the outer loop vectorization of the non-scalar version is attempted
40393 before vectorization of the inner loop in scalar version. If
40394 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40395 vectorization of its inner loop.
40396 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40397 has 2 inner loops, rename also on edges from bb whose single pred
40398 is outer_loop->header. Fix typo in function comment.
40400 2017-01-09 Martin Sebor <msebor@redhat.com>
40403 * asan.c (asan_emit_stack_protection): Increase local buffer size
40404 to avoid snprintf truncation warning.
40406 2017-01-09 Andrew Pinski <apinski@cavium.com>
40408 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40409 to reference thunderx2t99 for the tuning structure
40410 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40412 (thunderx2t99_extra_costs): This.
40413 * config/aarch64/aarch64-tune.md: Regenerate.
40414 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40415 (vulcan_addrcost_table): This.
40416 (vulcan_regmove_cost): Rename to ...
40417 (thunderx2t99_regmove_cost): This.
40418 (vulcan_vector_cost): Rename to ...
40419 (thunderx2t99_vector_cost): this.
40420 (vulcan_branch_cost): Rename to ...
40421 (thunderx2t99_branch_cost): This.
40422 (vulcan_tunings): Rename to ...
40423 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40424 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40426 2017-01-09 Martin Jambor <mjambor@suse.cz>
40430 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40431 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40432 (propagate_vr_accross_jump_function): Use the above function for all
40433 value range computations for pass-through jump functions and type
40434 converasion from explicit value range values.
40435 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40436 parameters from TYPE_ARG_TYPES.
40437 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40438 (ipa_write_node_info): Stream type of the actual argument.
40439 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40441 2017-01-09 Martin Liska <mliska@suse.cz>
40444 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40445 (lookup_compiler): Do not show error message with have_E.
40447 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40449 PR tree-optimization/78938
40450 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40451 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40452 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40453 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40456 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40458 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40461 2017-01-09 Richard Biener <rguenther@suse.de>
40463 PR tree-optimization/78997
40464 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40465 name condition properly.
40467 2017-01-09 Richard Biener <rguenther@suse.de>
40470 * dwarf2out.c (is_cxx): New overload with context.
40471 (is_naming_typedef_decl): Use it.
40473 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40475 * invoke.texi (Option Summary): Correct spacing in option lists
40476 and add line breaks to fix over-long lines.
40478 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40480 PR middle-end/17660
40482 * extend.texi (Common Variable Attributes): Add xref to GCC
40483 Internals manual to explain mode attribute keywords.
40485 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40488 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40489 and Preprocessor Options.
40490 (Options for Linking): Document -pthread here....
40491 (RS/6000 and PowerPC Options): ...not here.
40492 (Solaris 2 Options): ...or here.
40493 * doc/cppopts.texi: Document -pthread.
40495 2017-01-08 Martin Sebor <msebor@redhat.com>
40497 PR middle-end/77708
40498 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40499 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40500 New member functions.
40501 (format_directive): Used them.
40503 (pass_sprintf_length::handle_gimple_call): Same.
40504 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40505 to avoid truncation for any argument.
40506 (extract_affine_mul): Same.
40507 * tree.c (get_file_function_name): Same.
40509 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40511 PR middle-end/77484
40512 * predict.def (PRED_INDIR_CALL): Set to 86.
40514 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40516 PR preprocessor/54124
40517 * doc/cppopts.texi: Reformat -d subtable to list the full name
40518 of the options. Add cross-reference to the docs for the general
40519 compiler -d options.
40520 * doc/invoke.texi (Developer Options): Add cross-reference to the
40521 preprocessor-specific -d option documentation.
40523 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40525 PR preprocessor/13498
40526 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40527 redudant material, and reflect new command-line options.
40528 (System Headers): Likewise.
40530 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40532 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40533 -isystem, and -idirafter. Copy-edit.
40534 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40535 default for -ftrack-macro-expansion. Delete obsolete and
40536 badly-formatted implementation details about -fdebug-cpp output.
40537 * doc/cppwarnopts.texi: Copy-edit.
40539 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40542 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40543 that the transition from a max line width >= 1<<10 to narrower
40544 lines works correctly.
40546 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40548 * doc/options.texi (PerFunction): New.
40549 * opt-functions.awk (switch_flags): Map both Optimization and
40550 PerFunction to CL_OPTIMIZATION.
40551 * opth-gen.awk: Test for PerFunction flag along with
40553 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40554 it only when the latter is present. Skip those that don't in
40555 the hash function generator.
40556 * common.opt (fvar-tracking): Mark as PerFunction instead of
40558 (fvar-tracking-assignments): Likewise.
40559 (fvar-tracking-assignments-toggle): Likewise.
40560 (fvar-tracking-uninit): Likewise.
40562 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40564 PR translation/79018
40565 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40568 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40571 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40572 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40573 TARGET_LONG_JUMP_TABLE_OFFSETS.
40574 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40575 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40576 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40577 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40578 * config/m68k/m68k.md (tablejump expander): Likewise.
40579 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40580 TARGET_LONG_JUMP_TABLE_OFFSETS.
40581 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40582 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40584 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40585 David Holsgrove <david.holsgrove@xilinx.com>
40587 * common/config/microblaze/microblaze-common.c
40588 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40589 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40591 * config/microblaze/microblaze.c (microblaze_must_save_register)
40592 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40594 (microblaze_eh_return): New function.
40595 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40596 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40597 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40598 * config/microblaze/microblaze.md (eh_return): New pattern.
40600 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40602 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40603 GCC_DIAGNOSTIC_STRINGIFY): Define.
40605 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40607 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40609 * config/arm/arm.md (<mcrr>): New.
40611 * config/arm/arm.c (arm_arch5te): New.
40612 (arm_option_override): Set arm_arch5te.
40613 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40615 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40616 (arm_mcrr_qualifiers): ... this. New.
40617 (MRRC_QUALIFIERS): Define to...
40618 (arm_mrrc_qualifiers): ... this. New.
40619 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40621 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40622 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40623 (MRRCI, mrrc, MRRC): New.
40624 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40625 VUNSPEC_MRRC2): New.
40627 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40629 * config/arm/arm.md (<mcr>): New.
40631 * config/arm/arm.c (arm_coproc_builtin_available): Add
40632 support for mcr, mrc, mcr2 and mrc2.
40633 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40634 (arm_mcr_qualifiers): ... this. New.
40635 (MRC_QUALIFIERS): Define to ...
40636 (arm_mrc_qualifiers): ... this. New.
40637 (MCR_QUALIFIERS): Define to ...
40638 (arm_mcr_qualifiers): ... this. New.
40639 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40641 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40642 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40643 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40644 VUNSPEC_MRC2): New.
40646 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40648 * config/arm/arm.md (*ldc): New.
40652 * config/arm/arm.c (arm_coproc_builtin_available): Add
40653 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40654 (arm_coproc_ldc_stc_legitimate_address): New.
40655 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40656 'qualifier_const_pointer'.
40657 (LDC_QUALIFIERS): Define to...
40658 (arm_ldc_qualifiers): ... this. New.
40659 (STC_QUALIFIERS): Define to...
40660 (arm_stc_qualifiers): ... this. New.
40661 * config/arm/arm-protos.h
40662 (arm_coproc_ldc_stc_legitimate_address): New.
40663 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40664 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40665 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40666 stc2, stcl, stc2l): New.
40667 * config/arm/constraints.md (Uz): New.
40668 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40669 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40670 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40671 VUNSPEC_STC2L): New.
40673 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40675 * config/arm/arm.md (<cdp>): New.
40676 * config/arm/arm.c (neon_const_bounds): Rename this ...
40677 (arm_const_bounds): ... this.
40678 (arm_coproc_builtin_available): New.
40679 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40680 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40681 (CDP_QUALIFIERS): Define to...
40682 (arm_cdp_qualifiers): ... this. New.
40684 (arm_expand_builtin_args): Add case for 6 arguments.
40685 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40686 (arm_const_bounds): ... this.
40687 (arm_coproc_builtin_available): New.
40688 * config/arm/arm_acle.h (__arm_cdp): New.
40690 * config/arm/arm_acle_builtins.def (cdp): New.
40692 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40693 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40694 'arm_const_bounds'.
40695 * config/arm/types.md (coproc): New.
40696 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40697 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40698 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40699 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40701 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40703 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40704 (UBINOP_QUALIFIERS): New.
40706 (acle_builtin_data): New. Change comment.
40707 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40708 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40709 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40710 arm_acle_builtins.def.
40711 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40712 (arm_init_acle_builtins): New.
40713 (CRC32_BUILTIN): Remove.
40714 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40715 crc32cb, crc32ch and crc32cw.
40716 (arm_init_crc32_builtins): Remove.
40717 (arm_init_builtins): Use arm_init_acle_builtins rather
40718 than arm_init_crc32_builtins.
40719 (arm_expand_acle_builtin): New.
40720 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40721 * config/arm/arm_acle_builtins.def: New.
40723 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40725 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40726 (arm_builtin_datum): ... this.
40727 (arm_init_neon_builtin): Rename to ...
40728 (arm_init_builtin): ... this. Add a new parameters PREFIX
40729 and USE_SIG_IN_NAME.
40730 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40731 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40732 'arm_builtin_datum'.
40733 (arm_init_vfp_builtins): Likewise.
40734 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40735 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40736 (arm_expand_neon_args): Rename to ...
40737 (arm_expand_builtin_args): ... this. Rename builtin_arg
40738 enum values and differentiate between ARG_BUILTIN_MEMORY
40739 and ARG_BUILTIN_NEON_MEMORY.
40740 (arm_expand_neon_builtin_1): Rename to ...
40741 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40742 values, arm_expand_builtin_args and add bool parameter NEON.
40743 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40744 (arm_expand_vfp_builtin): Likewise.
40745 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40747 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40749 PR middle-end/77484
40750 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40751 * predict.c (tree_estimate_probability_bb): Reverse direction of
40752 polymorphic call predictor.
40754 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40756 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40757 (determine_pass_name_match): ...this new function and...
40758 (should_skip_pass_p): ...this new function.
40760 2017-01-06 Nathan Sidwell <nathan@acm.org>
40762 * ipa-visibility.c (function_and_variable_visibility): Reformat
40763 comments and long lines. Remove extrneous if.
40764 * symtab.c (symtab_node::make_decl_local): Fix code format.
40765 (symtab_node::set_section_for_node): Fix comment typo.
40767 2017-01-06 Martin Liska <mliska@suse.cz>
40770 * lra-constraints.c: Rename invariant to lra_invariant.
40771 * predict.c (set_even_probabilities): Initialize e to NULL.
40773 2017-01-05 Martin Sebor <msebor@redhat.com>
40775 PR tree-optimization/78910
40776 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40777 (format_integer): Correct off-by-one error in the handling
40778 of precision with negative numbers in signed conversions..
40780 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40782 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40784 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40786 PR tree-optimization/71016
40787 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40788 factor_out_conditional_conversion. Formatting fix.
40789 (factor_out_conditional_conversion): Add cond_stmt argument.
40790 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40791 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40794 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40796 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40797 read-rtl-function.o, and selftest-rtl.o.
40798 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40799 (selftest::aarch64_test_loading_full_dump): New function.
40800 (selftest::aarch64_run_selftests): New function.
40801 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40802 selftest::aarch64_run_selftests.
40803 * config/i386/i386.c
40804 (selftest::ix86_test_loading_dump_fragment_1): New function.
40805 (selftest::ix86_test_loading_call_insn): New function.
40806 (selftest::ix86_test_loading_full_dump): New function.
40807 (selftest::ix86_test_loading_unspec): New function.
40808 (selftest::ix86_run_selftests): Call the new functions.
40809 * emit-rtl.c (maybe_set_max_label_num): New function.
40810 * emit-rtl.h (maybe_set_max_label_num): New decl.
40811 * function.c (instantiate_decls): Guard call to
40812 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40813 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40815 * gensupport.c (gen_reader::gen_reader): Pass "false"
40816 for new "compact" param of rtx_reader.
40817 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40818 rather than an empty string for NULL strings.
40819 * read-md.c: Potentially include config.h rather than bconfig.h.
40820 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40821 (have_error): New global, copied from errors.c.
40822 (md_reader::read_name): Rename to...
40823 (md_reader::read_name_1): ...this, adding "out_loc" param,
40824 and converting "missing name or number" to returning false, rather
40826 (md_reader::read_name): Reimplement in terms of read_name_1.
40827 (md_reader::read_name_or_nil): New function.
40828 (md_reader::read_string): Handle "(nil)" by returning NULL.
40829 (md_reader::md_reader): Add new param "compact".
40830 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40831 (md_reader::read_file): New method.
40832 * read-md.h (md_reader::md_reader): Add new param "compact".
40833 (md_reader::read_file): New method.
40834 (md_reader::is_compact): New accessor.
40835 (md_reader::read_name): Convert return type from void to file_location.
40836 (md_reader::read_name_or_nil): New decl.
40837 (md_reader::read_name_1): New decl.
40838 (md_reader::m_compact): New field.
40839 (noop_reader::noop_reader): Pass "false" for new "compact" param
40841 (rtx_reader::rtx_reader): Add new "compact" param.
40842 (rtx_reader::read_rtx_operand): Make virtual and convert return
40843 type from void to rtx.
40844 (rtx_reader::read_until): New decl.
40845 (rtx_reader::handle_any_trailing_information): New virtual function.
40846 (rtx_reader::postprocess): New virtual function.
40847 (rtx_reader::finalize_string): New virtual function.
40848 (rtx_reader::m_in_call_function_usage): New field.
40849 (rtx_reader::m_reuse_rtx_by_id): New field.
40850 * read-rtl-function.c: New file.
40851 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40852 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40853 (selftest::verify_three_block_rtl_cfg): New decl.
40854 * read-rtl-function.h: New file.
40855 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40856 For host, include function.h, memmodel.h, and emit-rtl.h.
40857 (one_time_initialization): New function.
40858 (struct compact_insn_name): New struct.
40859 (compact_insn_names): New array.
40860 (find_code): Handle insn codes in compact dumps.
40861 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40862 (bind_subst_iter_and_attr): Likewise.
40863 (add_condition_to_string): Likewise.
40864 (add_condition_to_rtx): Likewise.
40865 (apply_attribute_uses): Likewise.
40866 (add_current_iterators): Likewise.
40867 (apply_iterators): Likewise.
40868 (initialize_iterators): Guard usage of apply_subst_iterator with
40869 #ifdef GENERATOR_FILE.
40870 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40871 (md_reader::read_mapping): Likewise.
40872 (add_define_attr_for_define_subst): Likewise.
40873 (add_define_subst_attr): Likewise.
40874 (read_subst_mapping): Likewise.
40875 (check_code_iterator): Likewise.
40876 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40878 (one_time_initialization): New function.
40879 (rtx_reader::read_until): New method.
40880 (read_flags): New function.
40881 (parse_reg_note_name): New function.
40882 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40883 Handle reuse_rtx ids.
40884 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40885 Add parsing support for RTL dumps, mirroring the special-cases in
40886 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40887 values, and calling handle_any_trailing_information.
40888 (rtx_reader::read_rtx_operand): Convert return type from void
40889 to rtx, returning return_rtx. Handle case 'e'. Call
40890 finalize_string on XSTR and XTMPL fields.
40891 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40892 "(nil)" values were omitted. Call the postprocess vfunc on the
40894 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40895 class ctor. Initialize m_in_call_function_usage. Call
40896 one_time_initialization.
40897 * rtl-tests.c (selftest::test_uncond_jump): Call
40898 set_new_first_and_last_insn.
40899 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40900 * selftest-rtl.c: New file.
40901 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40902 (selftest::get_insn_by_uid): New decl.
40903 * selftest-run-tests.c (selftest::run_tests): Call
40904 read_rtl_function_c_tests.
40905 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40906 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40909 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40911 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40912 operands in a special way. Assert that pos+len <= mode precision.
40914 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40916 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40917 3 argument Alias with unlimited for the negative form.
40918 (fno-vect-cost-model): Removed.
40920 2017-01-05 Martin Liska <mliska@suse.cz>
40922 * hsa-gen.c (gen_hsa_divmod): New function.
40923 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40925 2017-01-05 Martin Liska <mliska@suse.cz>
40928 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40931 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40933 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40934 small constant length operands.
40936 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40938 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40939 between loop iterations.
40941 2017-01-05 Martin Liska <mliska@suse.cz>
40944 * gimplify.c (gimplify_decl_expr): Compare to
40945 asan_poisoned_variables instread of checking flags.
40946 (gimplify_target_expr): Likewise.
40947 (gimplify_expr): Likewise.
40948 (gimplify_function_tree): Conditionally initialize
40949 asan_poisoned_variables.
40951 2017-01-04 Jeff Law <law@redhat.com>
40953 PR tree-optimizatin/78812
40954 * rtl.h (contains_mem_rtx_p): Prototype.
40955 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40956 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40957 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40958 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40959 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40961 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40963 * input.c (assert_char_at_range): Default-initialize actual_range.
40965 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40967 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40968 to match the caller.
40970 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40972 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40973 insns after final jump in test to emit dummy move.
40975 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40977 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40978 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40980 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40982 * multiple_target.c (create_dispatcher_calls): Init e_next.
40983 * tree-ssa-loop-split.c (split_loop): Init border.
40984 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40987 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40992 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40993 (altivec_register_operand): Do not return true if the operand
40994 contains a SUBREG mixing SImode and SFmode.
40995 (vsx_register_operand): Likewise.
40996 (vsx_reg_sfsubreg_ok): New predicate.
40997 (vfloat_operand): Do not return true if the operand contains a
40998 SUBREG mixing SImode and SFmode.
40999 (vint_operand): Likewise.
41000 (vlogical_operand): Likewise.
41001 (gpc_reg_operand): Likewise.
41002 (int_reg_operand): Likewise.
41003 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41004 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41005 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41007 (rs6000_emit_move_si_sf_subreg): New helper function.
41008 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41009 fixup SUBREGs involving SImode and SFmode.
41010 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41011 numbers for the new peephole2 optimization.
41012 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41013 the GLIBC math library that do AND/IOR/XOR operations on single
41014 precision floating point.
41015 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41016 target macros to say whether we need to avoid SUBREGs mixing
41018 (TARGET_ALLOW_SF_SUBREG): Likewise.
41019 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41020 (UNSPEC_SI_FROM_SF): Likewise.
41021 (iorxor): Change spacing.
41022 (and_ior_xor): New iterator for AND, IOR, and XOR.
41023 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41024 (movdi_from_sf_zero_ext): Likewise.
41025 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41026 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41027 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41028 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41029 (fms<mode>4): Likewise.
41030 (fnma<mode>4): Likewise.
41031 (fnms<mode>4): Likewise.
41032 (nfma<mode>4): Likewise.
41033 (nfms<mode>4): Likewise.
41035 2017-01-04 Marek Polacek <polacek@redhat.com>
41038 * doc/invoke.texi: Document -Wpointer-compare.
41040 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41042 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41045 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41046 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41049 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41052 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41053 documentation of the powerpc_popcntb_ok attribute.
41054 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41055 code to issue warning messages if a requested CPU configuration is
41056 not supported by the binary (assembler and loader) toolchain.
41057 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41058 made to define a built-in function that has been disabled.
41059 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41060 made to define a built-in function that has been disabled.
41061 (altivec_init_builtins): Add comment explaining why definition
41062 of the DST built-in functions is not preceded by an assertion
41063 check. Add assertions to prevent ICE if attempts are made to
41064 define an altivec predicate or an abs* built-in function that has
41066 (htm_init_builtins): Add comment explaining why definition of the
41067 htm built-in functions is not preceded by an assertion check.
41069 2017-01-04 Jeff Law <law@redhat.com>
41071 PR tree-optimizatin/67955
41072 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41073 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41074 the points-to solution does not include pt_null. Use DECL_PT_UID
41077 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41079 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41080 Use gen_int_mode instead of gen_lopwart for const_int operands.
41082 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41084 PR tree-optimization/71563
41085 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41086 out of range value - has low bits known to be zero.
41088 2017-01-04 Alan Modra <amodra@gmail.com>
41090 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41091 * configure: Regenerate.
41092 * config.in: Regenerate.
41094 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41097 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41098 a substring of the message, but strcmp with the whole message. Ifdef
41099 ENABLE_NLS, translate the message first using dgettext.
41101 2017-01-03 Jeff Law <law@redhat.com>
41103 PR tree-optimizatin/78856
41104 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41105 (mark_threaded_blocks): Remove code to truncate thread paths that
41106 cross multiple loop headers. Instead invalidate the cached loop
41107 iteration information and handle case of a thread path walking
41108 into an irreducible region.
41110 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41113 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41114 assertions. Add support for doing the signbit if the IEEE 128-bit
41115 floating point value is in a GPR.
41116 * config/rs6000/rs6000.md (Fsignbit): Delete.
41117 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41118 Update the length attribute if the value is in a GPR.
41119 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41120 the sign or zero extension instruction, since the value is always 0/1.
41121 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41124 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41125 extracting SImode to a GPR register so that we can generate a
41126 store, limit the vector to be in a traditional Altivec register
41127 for the vextuwrx instruction.
41129 2017-01-03 Ian Lance Taylor <iant@google.com>
41131 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41133 2017-01-03 Martin Sebor <msebor@redhat.com>
41135 PR tree-optimization/78696
41136 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41137 precision. Use MPFR for %f for greater fidelity. Correct handling
41139 (pass_sprintf_length::compute_format_length): Set width and precision
41140 specified by asrerisk to void_node for vararg functions.
41141 (try_substitute_return_value): Adjust dump output.
41143 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41145 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41147 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41149 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41150 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41151 -mlra/-mno-lra was passed to the compiler.
41153 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41155 PR rtl-optimization/65618
41156 * emit-rtl.c (try_split): Move initialization of "before" and
41157 "after" to just before the call to emit_insn_after_setloc.
41159 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41161 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41163 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41165 * dwarf2out.c (gen_enumeration_type_die): When
41166 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41168 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41170 PR tree-optimization/78965
41171 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41172 Change first argument from const call_info & to call_info &. For %n
41173 set info.nowrite to false.
41175 PR middle-end/78901
41176 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41177 possibly throwing calls.
41179 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41180 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41181 and fns handling, rather than in a separate case SSA_NAME.
41183 2017-01-02 Jeff Law <law@redhat.com>
41185 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41186 fixes for first_period and second_period variables.
41188 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41191 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41192 (*insvqi_1): New insn pattern.
41193 (*insvqi_1_mem_rex64): Ditto.
41194 (*insvqi_2): Ditto.
41195 (*insvqi_3): Rename from *insvqi.
41197 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41199 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41201 * doc/cfg.texi (Edges): Remove reference to Java.
41202 (Maintaining the CFG): Ditto.
41204 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41206 PR middle-end/77674
41207 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41208 transparent aliases.
41210 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41212 PR middle-end/77484
41213 * predict.def (PRED_CALL): Update hitrate.
41214 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41215 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41216 into direct/indirect/polymorphic variants.
41218 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41220 Update copyright years.
41222 * gcc.c (process_command): Update copyright notice dates.
41223 * gcov-dump.c (print_version): Ditto.
41224 * gcov.c (print_version): Ditto.
41225 * gcov-tool.c (print_version): Ditto.
41226 * gengtype.c (create_file): Ditto.
41227 * doc/cpp.texi: Bump @copying's copyright year.
41228 * doc/cppinternals.texi: Ditto.
41229 * doc/gcc.texi: Ditto.
41230 * doc/gccint.texi: Ditto.
41231 * doc/gcov.texi: Ditto.
41232 * doc/install.texi: Ditto.
41233 * doc/invoke.texi: Ditto.
41235 Copyright (C) 2017 Free Software Foundation, Inc.
41237 Copying and distribution of this file, with or without modification,
41238 are permitted in any medium without royalty provided the copyright
41239 notice and this notice are preserved.