1 2017-12-21 Martin Liska <mliska@suse.cz>
4 * gcov-dump.c (dump_gcov_file): Do not read info about
5 support_unexecuted_blocks for gcda files.
7 2017-12-21 Jakub Jelinek <jakub@redhat.com>
9 PR rtl-optimization/82973
10 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
11 (valid_for_const_vector_p): ... this.
12 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
13 (valid_for_const_vector_p): ... this. Adjust function comment.
14 (gen_vec_duplicate): Adjust caller.
15 * optabs.c (expand_vector_broadcast): Likewise.
16 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
17 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
19 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
20 instead of GET_CODE == CONST_FIXED.
21 (simplify_subreg): Use CONST_FIXED_P macro instead of
22 GET_CODE == CONST_FIXED.
25 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
26 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
27 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
28 (ix86_option_override_internal): Adjust for
29 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
30 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
31 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
32 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
33 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
34 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
35 def_builtin2 for CET builtins.
36 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
37 up formatting in IX86_BUILTIN_RDPID code.
38 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
39 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
40 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
41 ix86_isa_flags2 to ix86_isa_flags.
42 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
44 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
45 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
46 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
48 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
49 Or in OPTION_MASK_ISA_AVX512F_SET.
50 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
51 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
52 (ix86_handle_option): Adjust for
53 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
54 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
56 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
57 Alan Hayward <alan.hayward@arm.com>
58 David Sherwood <david.sherwood@arm.com>
60 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
61 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
62 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
63 Track polynomial offsets.
65 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
66 Alan Hayward <alan.hayward@arm.com>
67 David Sherwood <david.sherwood@arm.com>
69 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
70 Treat drb->init as a poly_int. Fail if its misalignment wrt
71 vector_alignment isn't known.
73 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
74 Alan Hayward <alan.hayward@arm.com>
75 David Sherwood <david.sherwood@arm.com>
77 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
79 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
80 (vect_loop_versioning): Take the loop versioning threshold as a
82 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
83 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
85 (vect_analyze_loop_2): Compute the loop versioning threshold
86 whenever loop versioning is needed, and store it in the new
87 field rather than combining it with the cost model threshold.
88 (vect_transform_loop): Update call to vect_loop_versioning.
89 Try to combine the loop versioning and cost thresholds here.
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 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
96 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
97 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
98 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
99 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
100 (group_compare_offset): Update accordingly.
101 (split_small_address_groups_p): Likewise.
102 (record_use): Take addr_offset as a poly_uint64 rather than
103 an unsigned HOST_WIDE_INT.
104 (strip_offset): Return the offset as a poly_uint64 rather than
105 an unsigned HOST_WIDE_INT.
106 (record_group_use, split_address_groups): Track polynomial offsets.
107 (add_iv_candidate_for_use): Likewise.
108 (addr_offset_valid_p): Take the offset as a poly_int64 rather
109 than a HOST_WIDE_INT.
110 (strip_offset_1): Return the offset as a poly_int64 rather than
113 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
114 Alan Hayward <alan.hayward@arm.com>
115 David Sherwood <david.sherwood@arm.com>
117 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
118 rather than a HOST_WIDE_INT.
119 * tree.c (get_binfo_at_offset): Likewise.
121 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
122 Alan Hayward <alan.hayward@arm.com>
123 David Sherwood <david.sherwood@arm.com>
125 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
126 rather than a HOST_WIDE_INT.
127 * tree-sra.c (build_ref_for_offset): Likewise.
129 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
130 Alan Hayward <alan.hayward@arm.com>
131 David Sherwood <david.sherwood@arm.com>
133 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
135 * tree.c (mem_ref_offset): Likewise.
136 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
137 * builtins.c (get_object_alignment_2): Likewise.
138 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
139 * gimple-fold.c (get_base_constructor): Likewise.
140 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
141 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
143 * ipa-polymorphic-call.c
144 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
145 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
146 (get_ancestor_addr_info): Likewise.
147 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
148 * match.pd: Likewise.
149 * tree-data-ref.c (dr_analyze_innermost): Likewise.
150 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
151 * tree-eh.c (tree_could_trap_p): Likewise.
152 * tree-object-size.c (addr_object_size): Likewise.
153 * tree-ssa-address.c (copy_ref_info): Likewise.
154 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
155 (indirect_refs_may_alias_p): Likewise.
156 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
157 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
158 (non_rewritable_mem_ref_base): Likewise.
159 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
160 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
161 * varasm.c (decode_addr_const): Likewise.
163 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
164 Alan Hayward <alan.hayward@arm.com>
165 David Sherwood <david.sherwood@arm.com>
167 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
168 offsets for MEM_REFs.
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 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
175 as a poly_uint64 rather than an unsigned int.
176 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
178 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
179 Alan Hayward <alan.hayward@arm.com>
180 David Sherwood <david.sherwood@arm.com>
182 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
184 (decode_addr_const): Update accordingly.
186 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
187 Alan Hayward <alan.hayward@arm.com>
188 David Sherwood <david.sherwood@arm.com>
190 * tree.h (bit_field_size, bit_field_offset): New functions.
191 * hsa-gen.c (gen_hsa_addr): Use them.
192 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
193 (simplify_vector_constructor): Likewise.
194 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
195 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
196 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
197 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
200 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
201 Alan Hayward <alan.hayward@arm.com>
202 David Sherwood <david.sherwood@arm.com>
204 * expr.h (emit_group_load, emit_group_load_into_temps)
205 (emit_group_store): Take the size as a poly_int64 rather than an int.
206 * expr.c (emit_group_load_1, emit_group_load): Likewise.
207 (emit_group_load_into_temp, emit_group_store): Likewise.
209 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
210 Alan Hayward <alan.hayward@arm.com>
211 David Sherwood <david.sherwood@arm.com>
213 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
214 an unsigned int to a poly_uint64.
215 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
216 sizes as poly_uint64s rather than unsigned ints.
217 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
220 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
221 Alan Hayward <alan.hayward@arm.com>
222 David Sherwood <david.sherwood@arm.com>
224 * lra-constraints.c (emit_inc): Change inc_amount from an int
227 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
228 Alan Hayward <alan.hayward@arm.com>
229 David Sherwood <david.sherwood@arm.com>
231 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
233 (add_stack_var, stack_var_cmp, partition_stack_vars)
234 (dump_stack_var_partition): Update accordingly.
235 (alloc_stack_frame_space): Take the size as a poly_int64 rather
236 than a HOST_WIDE_INT.
237 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
238 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
239 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
240 rather than a HOST_WIDE_INT.
242 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
243 Alan Hayward <alan.hayward@arm.com>
244 David Sherwood <david.sherwood@arm.com>
246 * target.def (return_pops_args): Treat both the input and output
247 sizes as poly_int64s rather than HOST_WIDE_INTS.
248 * targhooks.h (default_return_pops_args): Update accordingly.
249 * targhooks.c (default_return_pops_args): Likewise.
250 * doc/tm.texi: Regenerate.
251 * emit-rtl.h (incoming_args): Change pops_args, size and
252 outgoing_args_size from int to poly_int64_pod.
253 * function.h (expr_status): Change x_pending_stack_adjust and
254 x_stack_pointer_delta from int to poly_int64.
255 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
256 (ARGS_SIZE_RTX): Update accordingly.
257 * calls.c (highest_outgoing_arg_in_use): Change from int to
259 (stack_usage_watermark, stored_args_watermark): New variables.
260 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
261 (emit_call_1): Change the stack_size and rounded_stack_size
262 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
264 (save_fixed_argument_area): Check stack_usage_watermark.
265 (initialize_argument_information): Change old_pending_adj from
266 a HOST_WIDE_INT * to a poly_int64_pod *.
267 (compute_argument_block_size): Return the size as a poly_int64
269 (finalize_must_preallocate): Track polynomial argument sizes.
270 (compute_argument_addresses): Likewise.
271 (internal_arg_pointer_based_exp): Track polynomial offsets.
272 (mem_overlaps_already_clobbered_arg_p): Rename to...
273 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
274 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
275 Check stored_args_used_watermark.
276 (load_register_parameters): Update accordingly.
277 (check_sibcall_argument_overlap_1): Likewise.
278 (combine_pending_stack_adjustment_and_call): Take the unadjusted
279 args size as a poly_int64 rather than an int. Return a bool
280 indicating whether the optimization was possible and return
281 the new adjustment by reference.
282 (check_sibcall_argument_overlap): Track polynomail argument sizes.
283 Update stored_args_watermark.
284 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
285 (expand_call): Likewise. Maintain stack_usage_watermark and
286 stored_args_watermark. Update calls to
287 combine_pending_stack_adjustment_and_call.
288 (emit_library_call_value_1): Handle polynomial argument sizes.
289 Call stack_region_maybe_used_p and mark_stack_region_used.
290 Maintain stack_usage_watermark.
291 (store_one_arg): Likewise. Update call to
292 mem_overlaps_already_clobbered_arg_p.
293 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
295 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
296 * config/microblaze/microblaze.c (microblaze_function_prologue):
298 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
299 TARGET_RETURN_POPS_ARGS interface.
300 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
302 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
303 * config/i386/i386.c (ix86_return_pops_args): Update for new
304 TARGET_RETURN_POPS_ARGS interface.
305 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
306 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
307 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
308 TARGET_RETURN_POPS_ARGS interface.
309 * config/vax/vax.c (vax_return_pops_args): Likewise.
310 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
311 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
312 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
313 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
314 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
315 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
316 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
317 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
318 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
319 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
320 and x_stack_pointer_delta from int to poly_int64.
321 * dojump.c (do_pending_stack_adjust): Update accordingly.
322 * explow.c (allocate_dynamic_stack_space): Handle polynomial
323 stack_pointer_deltas.
324 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
325 (pad_to_arg_alignment): Track polynomial offsets.
326 (assign_parm_find_stack_rtl): Likewise.
327 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
328 * toplev.c (output_stack_usage): Update reference to
329 current_function_pushed_stack_size.
331 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
332 Alan Hayward <alan.hayward@arm.com>
333 David Sherwood <david.sherwood@arm.com>
335 * function.c (in_arg_offset, var_offset, dynamic_offset)
336 (out_arg_offset, cfa_offset): Change from int to poly_int64.
337 (instantiate_new_reg): Return the new offset as a poly_int64_pod
338 rather than a HOST_WIDE_INT.
339 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
340 (instantiate_virtual_regs_in_insn): Likewise.
342 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
343 Alan Hayward <alan.hayward@arm.com>
344 David Sherwood <david.sherwood@arm.com>
346 * rtl.h (get_args_size, add_args_size_note): New functions.
347 (find_args_size_adjust): Return a poly_int64 rather than a
349 (fixup_args_size_notes): Likewise. Make the same change to the
350 end_args_size parameter.
351 * rtlanal.c (get_args_size, add_args_size_note): New functions.
352 * builtins.c (expand_builtin_trap): Use add_args_size_note.
353 * calls.c (emit_call_1): Likewise.
354 * explow.c (adjust_stack_1): Likewise.
355 * cfgcleanup.c (old_insns_match_p): Update use of
356 find_args_size_adjust.
357 * combine.c (distribute_notes): Track polynomial arg sizes.
358 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
359 end_true_args_size, beg_delay_args_size and end_delay_args_size
360 from HOST_WIDE_INT to poly_int64.
361 (add_cfi_args_size): Take the args_size as a poly_int64 rather
362 than a HOST_WIDE_INT.
363 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
364 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
365 polynomial arg sizes.
366 * emit-rtl.c (try_split): Use get_args_size.
367 * recog.c (peep2_attempt): Likewise.
368 * reload1.c (reload_as_needed): Likewise.
369 * expr.c (find_args_size_adjust): Return the adjustment as a
370 poly_int64 rather than a HOST_WIDE_INT.
371 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
372 to a poly_int64 and change the return type in the same way.
373 (emit_single_push_insn): Track polynomial arg sizes.
375 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
376 Alan Hayward <alan.hayward@arm.com>
377 David Sherwood <david.sherwood@arm.com>
379 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
380 from HOST_WIDE_INT to poly_int64.
381 * expr.c (push_block, emit_push_insn): Likewise.
383 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
384 Alan Hayward <alan.hayward@arm.com>
385 David Sherwood <david.sherwood@arm.com>
387 * function.h (frame_space): Change start and length from HOST_WIDE_INT
389 (get_frame_size): Return the size as a poly_int64 rather than a
391 (frame_offset_overflow): Take the offset as a poly_int64 rather
392 than a HOST_WIDE_INT.
393 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
394 (assign_stack_temp): Likewise for the size.
395 * function.c (get_frame_size): Return a poly_int64 rather than
397 (frame_offset_overflow): Take the offset as a poly_int64 rather
398 than a HOST_WIDE_INT.
399 (try_fit_stack_local): Take the start, length and size as poly_int64s
400 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
401 rather than a HOST_WIDE_INT.
402 (add_frame_space): Take the start and end as poly_int64s rather than
404 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
405 (assign_stack_temp): Likewise for the size.
406 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
408 (find_temp_slot_from_address): Handle polynomial offsets.
409 (combine_temp_slots): Likewise.
410 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
412 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
413 poly_int64 rather than a HOST_WIDE_INT.
414 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
415 than a HOST_WIDE_INT.
416 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
417 polynomial frame offsets.
418 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
419 as a poly_int64 rather than an int.
420 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
421 * config/v850/v850-protos.h (compute_frame_size): Likewise.
422 * config/v850/v850.c (compute_frame_size): Likewise.
423 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
424 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
425 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
426 * config/pa/pa.c (pa_compute_frame_size): Likewise.
427 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
428 rather than a HOST_WIDE_INT.
429 * explow.c (get_dynamic_stack_base): Likewise.
430 * final.c (final_start_function): Use the constant lower bound
431 of the frame size for -Wframe-larger-than.
432 * ira.c (do_reload): Adjust for new get_frame_size return type.
433 * lra.c (lra): Likewise.
434 * reload1.c (reload): Likewise.
435 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
436 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
437 * rtlanal.c (get_initial_register_offset): Return the offset as
438 a poly_int64 rather than a HOST_WIDE_INT.
440 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * reload1.c (elim_table): Change initial_offset, offset and
445 previous_offset from HOST_WIDE_INT to poly_int64_pod.
446 (offsets_at): Change the target array's element type from
447 HOST_WIDE_INT to poly_int64_pod.
448 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
449 (elimination_costs_in_insn, update_eliminable_offsets)
450 (verify_initial_elim_offsets, set_offsets_for_label)
451 (init_eliminable_invariants): Update after above changes.
453 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
454 Alan Hayward <alan.hayward@arm.com>
455 David Sherwood <david.sherwood@arm.com>
457 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
458 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
459 (decomposition): Change start and end from HOST_WIDE_INT
461 (decompose, immune_p): Update accordingly.
462 (find_inc_amount): Return a poly_int64 rather than an int.
463 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
466 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
467 Alan Hayward <alan.hayward@arm.com>
468 David Sherwood <david.sherwood@arm.com>
470 * tree.h (get_inner_reference): Return the bitsize and bitpos
471 as poly_int64_pods rather than HOST_WIDE_INT.
472 * fold-const.h (ptr_difference_const): Return the pointer difference
473 as a poly_int64_pod rather than a HOST_WIDE_INT.
474 * expr.c (get_inner_reference): Return the bitsize and bitpos
475 as poly_int64_pods rather than HOST_WIDE_INT.
476 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
478 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
479 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
480 (optimize_bit_field_compare): Update call to get_inner_reference.
481 (decode_field_reference): Likewise.
482 (fold_unary_loc): Track polynomial offsets and sizes.
483 (split_address_to_core_and_offset): Return the bitpos as a
484 poly_int64_pod rather than a HOST_WIDE_INT.
485 (ptr_difference_const): Likewise for the pointer difference.
486 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
487 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
488 * dbxout.c (dbxout_expand_expr): Likewise.
489 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
490 (loc_list_from_tree_1, fortran_common): Likewise.
491 * gimple-laddress.c (pass_laddress::execute): Likewise.
492 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
493 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
494 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
495 * tree-affine.c (tree_to_aff_combination): Likewise.
496 (get_inner_reference_aff): Likewise.
497 * tree-data-ref.c (split_constant_offset_1): Likewise.
498 (dr_analyze_innermost): Likewise.
499 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
500 * tree-sra.c (ipa_sra_check_caller): Likewise.
501 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
502 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
503 (instrument_bool_enum_load, instrument_object_size): Likewise.
504 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
505 to get_inner_reference.
506 * hsa-gen.c (gen_hsa_addr): Likewise.
507 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
508 * tsan.c (instrument_expr): Likewise.
509 * match.pd: Update call to ptr_difference_const.
511 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * fold-const.c (fold_comparison): Track sizes and offsets as
516 poly_int64s rather than HOST_WIDE_INTs when folding address
519 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
520 Alan Hayward <alan.hayward@arm.com>
521 David Sherwood <david.sherwood@arm.com>
523 * expr.h (get_bit_range): Return the bitstart and bitend as
524 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
525 as a poly_int64 rather than a HOST_WIDE_INT.
526 * expr.c (get_bit_range): Likewise.
527 (expand_assignment): Update call accordingly.
528 * fold-const.c (optimize_bit_field_compare): Likewise.
530 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
531 Alan Hayward <alan.hayward@arm.com>
532 David Sherwood <david.sherwood@arm.com>
534 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
535 (aarch64_legitimate_address_p): Use it instead of an rtx code,
536 as an optional final parameter.
537 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
538 (aarch64_legitimate_address_p): Likewise.
539 (aarch64_print_address_internal): Take an aarch64_addr_query_type
540 instead of an rtx code.
541 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
542 (aarch64_legitimate_address_hook_p): Likewise.
543 (aarch64_print_ldpstp_address): Likewise.
544 (aarch64_print_operand_address): Likewise.
545 (aarch64_address_cost): Likewise.
546 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
547 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
548 (aarch64_mem_pair_lanes_operand): Likewise.
550 2017-12-20 Richard Biener <rguenther@suse.de>
552 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
553 update_stmt_if_modified.
555 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
557 PR tree-optimization/83491
558 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
559 before walking uses. Improve coding style and comments.
561 2017-12-20 Tom de Vries <tom@codesourcery.com>
563 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
565 2017-12-20 Jakub Jelinek <jakub@redhat.com>
568 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
570 (pass_ipa_free_fn_summary): Add small_p private data member,
571 initialize to false in the ctor.
572 (pass_ipa_free_fn_summary::clone,
573 pass_ipa_free_fn_summary::set_pass_param,
574 pass_ipa_free_fn_summary::gate): New methods.
575 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
576 | TODO_dump_symtab if small_p.
577 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
578 entry and add another instance of the pass with false parm after
580 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
581 ipa_free_fn_summary here.
583 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
585 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
587 2017-12-20 Martin Sebor <msebor@redhat.com>
590 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
591 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
592 (expand_builtin_strncmp): Same.
594 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
597 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
598 allowed debug stmts before labels.
599 (expand_gimple_basic_block): Likewise.
600 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
601 * gimple-iterator.h (gsi_after_labels): Likewise.
602 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
603 rename reused variable, and simplify using gsi_move_before.
604 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
605 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
606 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
607 (gimple_verify_flow_info, gimple_block_label): Likewise.
608 (make_blocks): Move debug markers after adjacent labels.
609 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
610 allowed debug insns outside blocks.
611 * df-scan.c (df_insn_delete): Likewise.
612 * lra-constraints.c (update_ebb_live_info): Likewise.
613 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
614 (vt_initialize, delete_vta_debug_insns): Likewise.
615 (reemit_marker_as_note): Drop BB parm. Adjust callers.
617 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
618 Alan Hayward <alan.hayward@arm.com>
619 David Sherwood <david.sherwood@arm.com>
621 * poly-int-types.h (round_down_to_byte_boundary): New macro.
622 (round_up_to_byte_boundary): Likewise.
623 * expr.h (get_bit_range): Add temporary shim.
624 * gimple-ssa-store-merging.c (store_operand_info): Change the
625 bitsize, bitpos, bitregion_start and bitregion_end fields from
626 unsigned HOST_WIDE_INT to poly_uint64.
627 (merged_store_group): Likewise load_align_base.
628 (compatible_load_p, compatible_load_p): Update accordingly.
629 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
630 (split_group, imm_store_chain_info::output_merged_store): Likewise.
631 (mem_valid_for_store_merging): Return the bitsize, bitpos,
632 bitregion_start and bitregion_end as poly_uint64s rather than
633 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
634 (handled_load): Take the bitsize, bitpos,
635 bitregion_start and bitregion_end as poly_uint64s rather than
636 unsigned HOST_WIDE_INTs.
637 (pass_store_merging::process_store): Update call to
638 mem_valid_for_store_merging.
640 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
641 Alan Hayward <alan.hayward@arm.com>
642 David Sherwood <david.sherwood@arm.com>
644 * builtins.c (get_object_alignment_2): Track polynomial offsets
645 and sizes. Update the alignment handling.
647 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
648 Alan Hayward <alan.hayward@arm.com>
649 David Sherwood <david.sherwood@arm.com>
651 * tree.h (get_inner_reference): Add a version that returns the
652 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
653 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
654 the case in which bitpos is not associated with the first byte.
656 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * tree-affine.h (get_inner_reference_aff): Return the size as a
662 * tree-affine.c (get_inner_reference_aff): Likewise.
663 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
664 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
666 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * fold-const.c (pointer_may_wrap_p): Take the offset as a
671 HOST_WIDE_INT rather than a poly_int64.
673 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
678 HOST_WIDE_INT to poly_int64_pod.
679 (perform_symbolic_merge): Update accordingly.
680 (bswap_replace): Likewise.
682 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * tree-affine.h (aff_tree::offset): Change from widest_int
688 (wide_int_ext_for_comb): Delete.
689 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
690 constants as poly_widest_int rather than widest_int.
691 (aff_combination_constant_multiple_p): Return the multiplier
692 as a poly_widest_int.
693 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
695 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
696 version static and add an overload for poly_widest_int.
697 (aff_combination_const, aff_combination_add_cst)
698 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
699 the constants as poly_widest_int rather than widest_int.
700 (tree_to_aff_combination): Generalize INTEGER_CST case to
702 (aff_combination_to_tree): Track offsets as poly_widest_ints.
703 (aff_combination_add_product, aff_combination_mult): Handle
705 (aff_combination_constant_multiple_p): Return the multiplier
706 as a poly_widest_int.
707 * tree-predcom.c (determine_offset): Return the offset as a
709 (split_data_refs_to_components, suitable_component_p): Update
711 (valid_initializer_p): Update call to
712 aff_combination_constant_multiple_p.
713 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
714 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
715 as a poly_int64 rather than a HOST_WIDE_INT.
716 (get_address_cost): Handle polynomial offsets.
717 (iv_elimination_compare_lt): Likewise.
718 (rewrite_use_nonlinear_expr): Likewise.
720 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
721 Alan Hayward <alan.hayward@arm.com>
722 David Sherwood <david.sherwood@arm.com>
724 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
725 as a poly_int64_pod rather than a HOST_WIDE_INT.
726 (get_addr_base_and_unit_offset): Likewise.
727 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
728 (get_addr_base_and_unit_offset): Likewise.
729 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
730 to poly_int64 in example.
731 * fold-const.c (fold_binary_loc): Update call to
732 get_addr_base_and_unit_offset.
733 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
734 (maybe_canonicalize_mem_ref_addr): Likewise.
735 (gimple_fold_stmt_to_constant_1): Likewise.
736 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
738 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
739 * match.pd: Likewise.
740 * omp-low.c (lower_omp_target): Likewise.
741 * tree-sra.c (build_ref_for_offset): Likewise.
742 (build_debug_ref_for_model): Likewise.
743 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
744 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
745 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
746 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
747 (constant_pointer_difference): Likewise.
748 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
749 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
750 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
751 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
752 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
753 (set_ssa_val_to): Likewise.
754 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
755 (maybe_diag_stxncpy_trunc): Likewise.
756 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
757 * tree.c (build_simple_mem_ref_loc): Likewise.
758 (array_at_struct_end_p): Likewise.
760 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
761 Alan Hayward <alan.hayward@arm.com>
762 David Sherwood <david.sherwood@arm.com>
764 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
765 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
766 (get_ref_base_and_extent_hwi): Declare.
767 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
768 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
769 (get_ref_base_and_extent_hwi): New function.
770 * cfgexpand.c (expand_debug_expr): Update call to
771 get_ref_base_and_extent.
772 * dwarf2out.c (add_var_loc_to_decl): Likewise.
773 * gimple-fold.c (get_base_constructor): Return the offset as a
774 poly_int64_pod rather than a HOST_WIDE_INT.
775 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
776 * ipa-polymorphic-call.c
777 (ipa_polymorphic_call_context::set_by_invariant)
778 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
779 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
780 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
781 rather than get_ref_base_and_extent.
782 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
783 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
784 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
786 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
787 call to get_ref_base_and_extent.
788 * tree-sra.c (create_access, get_access_for_expr): Likewise.
789 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
790 (stmt_kills_ref_p): Likewise.
791 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
792 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
794 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
795 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
796 when calling native_encode_expr.
797 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
798 call to get_ref_base_and_extent.
799 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
800 get_ref_base_and_extent.
801 * var-tracking.c (track_expr_p): Likewise.
803 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
804 Alan Hayward <alan.hayward@arm.com>
805 David Sherwood <david.sherwood@arm.com>
807 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
808 HOST_WIDE_INT to poly_int64_pod.
809 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
810 polynomail parameter offsets.
812 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
813 Alan Hayward <alan.hayward@arm.com>
814 David Sherwood <david.sherwood@arm.com>
816 * gengtype.c (main): Handle poly_int64_pod.
817 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
818 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
819 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
820 from HOST_WIDE_INT to poly_int64_pod.
821 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
822 (copy_cfa): New function.
823 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
824 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
825 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
826 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
827 (initial_return_save): Treat offsets as poly_ints.
828 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
829 offset is nonconstant.
830 (reg_save): Take the offset as a poly_int64. Fall back to
831 DW_CFA_expression for nonconstant offsets.
832 (queue_reg_save): Take the offset as a poly_int64.
833 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
835 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
836 Alan Hayward <alan.hayward@arm.com>
837 David Sherwood <david.sherwood@arm.com>
839 * rtl.h (operand_subword, operand_subword_force): Take the offset
840 as a poly_uint64 an unsigned int.
841 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
843 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
848 'p' format code. Use INT_LIST rather than SUBREG as the example of
849 a code with an XINT and an XEXP. Remove the implication that
850 accessing an rtx field using XINT is expected to work.
851 * rtl.def (SUBREG): Change format from "ei" to "ep".
852 * rtl.h (rtunion::rt_subreg): New field.
853 (XCSUBREG): New macro.
854 (SUBREG_BYTE): Use it.
855 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
856 Update constructor accordingly.
857 (subreg_shape::operator ==): Update accordingly.
858 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
859 than an unsigned int.
860 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
861 a poly_uint64 rather than an unsigned int.
862 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
863 than an unsigned int.
864 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
865 (subreg_size_highpart_offset): Return a poly_uint64 rather than
866 an unsigned int. Take the sizes as poly_uint64s.
867 (subreg_offset_from_lsb): Return a poly_uint64 rather than
868 an unsigned int. Take the shift as a poly_uint64 rather than
870 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
871 as a poly_uint64 rather than an unsigned int.
872 (simplify_subreg_regno): Likewise.
873 (byte_lowpart_offset): Return the memory offset as a poly_int64
875 (subreg_memory_offset): Likewise. Take the subreg offset as a
876 poly_uint64 rather than an unsigned int.
877 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
878 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
879 poly_uint64 rather than an unsigned int.
880 * rtl.c (rtx_format): Describe 'p' in comment.
881 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
882 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
883 offset as a poly_uint64 rather than an unsigned int.
884 (byte_lowpart_offset): Return the memory offset as a poly_int64
886 (subreg_memory_offset): Likewise. Take the subreg offset as a
887 poly_uint64 rather than an unsigned int.
888 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
889 mode sizes as poly_uint64s rather than unsigned ints. Return a
890 poly_uint64 rather than an unsigned int.
891 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
892 (copy_insn_1): Handle 'p'.
893 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
894 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
895 an unsigned int. Return the shift in the same way.
896 (subreg_lsb): Return the shift as a poly_uint64 rather than an
898 (subreg_size_offset_from_lsb): Take the sizes and shift as
899 poly_uint64s rather than unsigned ints. Return the offset as
901 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
902 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
904 * rtlhash.c (add_rtx): Handle 'p'.
905 * genemit.c (gen_exp): Likewise.
906 * gengenrtl.c (type_from_format, gendef): Likewise.
907 * gensupport.c (subst_pattern_match, get_alternatives_number)
908 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
909 (subst_dup): Likewise.
910 * gengtype.c (adjust_field_rtx_def): Likewise.
911 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
912 (match_pattern_2): Likewise.
913 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
914 (rtx_test::subreg_field): New function.
915 (operator ==, safe_to_hoist_p, transition_parameter_type)
916 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
917 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
918 * genpeep.c (match_rtx): Likewise.
919 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
920 (rtx_writer::print_rtx_operand): Handle 'p'.
921 (print_value): Handle SUBREG.
922 * read-rtl.c (apply_int_iterator): Likewise.
923 (rtx_reader::read_rtx_operand): Handle 'p'.
924 * alias.c (rtx_equal_for_memref_p): Likewise.
925 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
926 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
928 * calls.c (expand_call): Likewise.
929 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
930 (make_extraction, gen_lowpart_for_combine): Likewise.
931 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
933 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
934 rather than an unsigned int. Treat subreg offsets as poly_ints.
935 (exp_equiv_p): Handle 'p'.
936 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
937 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
938 * dse.c (find_shift_sequence): Likewise.
939 * dwarf2out.c (rtl_for_decl_location): Likewise.
940 * expmed.c (extract_low_bits): Likewise.
941 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
942 (expand_expr_real_2): Likewise.
943 * final.c (alter_subreg): Likewise.
944 (leaf_renumber_regs_insn): Handle 'p'.
945 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
946 Treat subreg offsets as poly_ints.
947 * fwprop.c (forward_propagate_and_simplify): Likewise.
948 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
949 * ira.c (get_subreg_tracking_sizes): Likewise.
950 * ira-conflicts.c (go_through_subreg): Likewise.
951 * ira-lives.c (process_single_reg_class_operands): Likewise.
952 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
953 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
954 as a poly_uint64 rather than an unsigned int.
955 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
956 subreg offsets as poly_ints.
957 * lra-constraints.c (operands_match_p): Handle 'p'.
958 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
959 * lra-spills.c (assign_mem_slot): Likewise.
960 * postreload.c (move2add_valid_value_p): Likewise.
961 * recog.c (general_operand, indirect_operand): Likewise.
962 * regcprop.c (copy_value, maybe_mode_change): Likewise.
963 (copyprop_hardreg_forward_1): Likewise.
964 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
965 (record_subregs_of_mode): Likewise.
966 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
967 * reload.c (operands_match_p): Handle 'p'.
968 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
969 * reload1.c (alter_reg, choose_reload_regs): Likewise.
970 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
971 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
972 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
973 (simplify_const_poly_int_tests<N>::run): Likewise.
974 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
975 a poly_uint64 rather than an unsigned int.
976 * valtrack.c (debug_lowpart_subreg): Likewise.
977 * var-tracking.c (var_lowpart): Likewise.
978 (loc_cmp): Handle 'p'.
980 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
981 Alan Hayward <alan.hayward@arm.com>
982 David Sherwood <david.sherwood@arm.com>
984 * ira.c (get_subreg_tracking_sizes): New function.
985 (init_live_subregs): Take an integer size rather than a register.
986 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
987 to init_live_subregs.
989 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * expr.c (store_constructor_field): Change bitsize from a
994 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
995 HOST_WIDE_INT to a poly_int64.
996 (store_constructor): Change size from a HOST_WIDE_INT to
998 (store_field): Likewise bitsize and bitpos.
1000 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1001 Alan Hayward <alan.hayward@arm.com>
1002 David Sherwood <david.sherwood@arm.com>
1004 * expmed.h (store_bit_field): Change bitregion_start and
1005 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1006 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1007 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1008 (store_fixed_bit_field, store_split_bit_field): Likewise.
1009 * expr.c (store_constructor_field, store_field): Likewise.
1010 (optimize_bitfield_assignment_op): Likewise. Make the same change
1011 to bitsize and bitpos.
1012 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1013 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1014 same change in the constructor arguments.
1015 (get_best_mode): Change bitregion_start and bitregion_end from
1016 unsigned HOST_WIDE_INT to poly_uint64.
1017 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1018 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1020 (bit_field_mode_iterator::next_mode): Update for new types
1021 of m_bitregion_start and m_bitregion_end.
1022 (get_best_mode): Change bitregion_start and bitregion_end from
1023 unsigned HOST_WIDE_INT to poly_uint64.
1025 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1026 Alan Hayward <alan.hayward@arm.com>
1027 David Sherwood <david.sherwood@arm.com>
1029 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1030 accepts poly_uint64 offsets.
1031 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1032 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1033 * expmed.c (lowpart_bit_field_p): Likewise.
1034 (extract_bit_field_as_subreg): New function, split out from...
1035 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1036 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1037 extractions, check that BITSIZE matches the size of the extracted
1038 value and that BITNUM is an exact multiple of that size.
1039 If all else fails, try forcing the value into memory if
1040 BITNUM is variable, and adjusting the address so that the
1041 offset is constant. Split the part that can only handle constant
1042 bitsize and bitnum out into...
1043 (extract_integral_bit_field): ...this new function.
1044 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1045 rather than unsigned HOST_WIDE_INTs.
1047 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1048 Alan Hayward <alan.hayward@arm.com>
1049 David Sherwood <david.sherwood@arm.com>
1051 * expmed.h (store_bit_field): Take bitsize and bitnum as
1052 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1053 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1054 that returns the byte size.
1055 (store_bit_field_1): Take bitsize and bitnum as
1056 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1057 to simple_mem_bitfield_p. Split the part that can only handle
1058 constant bitsize and bitnum out into...
1059 (store_integral_bit_field): ...this new function.
1060 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1061 than unsigned HOST_WIDE_INTs.
1062 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
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 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1069 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1071 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1072 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1073 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1074 offset as a poly_int64 rather than an int.
1075 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1076 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1077 * lra-constraints.c (equiv_address_substitution): Track offsets
1079 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1080 (curr_insn_transform): Handle the new form of sp_offset.
1081 * lra-eliminations.c (lra_elim_table): Change previous_offset
1082 and offset from HOST_WIDE_INT to poly_int64.
1083 (print_elim_table, update_reg_eliminate): Update accordingly.
1084 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1085 (get_elimination): Update accordingly.
1086 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1087 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1088 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1089 poly_int64 offsets generally.
1090 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1091 (mark_not_eliminable, init_elimination): Update accordingly.
1092 (remove_reg_equal_offset_note): Return a bool and pass the new
1093 offset back by pointer as a poly_int64.
1094 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1095 rather than a HOST_WIDE_INT.
1096 (do_remat): Track offsets poly_int64s.
1097 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1099 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1100 Alan Hayward <alan.hayward@arm.com>
1101 David Sherwood <david.sherwood@arm.com>
1103 * rtl.h (mem_attrs): Add a default constructor. Change size and
1104 offset from HOST_WIDE_INT to poly_int64.
1105 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1106 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1107 (widen_memory_access): Take the sizes and offsets as poly_int64s
1108 rather than HOST_WIDE_INTs.
1109 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1110 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1112 (adjust_offset_for_component_ref): Change the offset from a
1113 HOST_WIDE_INT to a poly_int64.
1114 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1115 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1116 * dce.c (find_call_stack_args): Likewise.
1117 * dse.c (record_store): Likewise.
1118 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1119 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1120 * read-rtl-function.c (test_loading_mem): Likewise.
1121 * rtlanal.c (may_trap_p_1): Likewise.
1122 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1123 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1124 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1125 (mem_attrs::mem_attrs): New function.
1126 (set_mem_attributes_minus_bitpos): Change bitpos from a
1127 HOST_WIDE_INT to poly_int64.
1128 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1129 (clear_mem_offset, clear_mem_size, change_address)
1130 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1131 initialize mem_attrs.
1132 (set_mem_offset, set_mem_size, adjust_address_1)
1133 (adjust_automodify_address_1, offset_address, widen_memory_access):
1134 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1136 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1137 Alan Hayward <alan.hayward@arm.com>
1138 David Sherwood <david.sherwood@arm.com>
1140 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1141 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1142 rather than 0 to represent an unknown size. Assert that the size
1143 is known when the mode isn't BLKmode.
1144 (may_trap_p_1): Use -1 for unknown sizes.
1145 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1147 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1148 Alan Hayward <alan.hayward@arm.com>
1149 David Sherwood <david.sherwood@arm.com>
1151 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1152 to poly_int64. Update commentary for positions_needed.large.
1153 (read_info_type): Change offset and width from HOST_WIDE_INT
1155 (set_usage_bits): Likewise.
1156 (canon_address): Return the offset as a poly_int64 rather than
1157 a HOST_WIDE_INT. Use strip_offset_and_add.
1158 (set_all_positions_unneeded, any_positions_needed_p): Use
1159 positions_needed.large to track stores with non-constant widths.
1160 (all_positions_needed_p): Likewise. Take the offset and width
1161 as poly_int64s rather than ints. Assert that rhs is nonnull.
1162 (record_store): Cope with non-constant offsets and widths.
1163 Nullify the rhs of an earlier store if we can't tell which bytes
1165 (find_shift_sequence): Take the access_size and shift as poly_int64s
1167 (get_stored_val): Take the read_offset and read_width as poly_int64s
1168 rather than HOST_WIDE_INTs.
1169 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1170 non-constant offsets and widths.
1172 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1173 Alan Hayward <alan.hayward@arm.com>
1174 David Sherwood <david.sherwood@arm.com>
1176 * inchash.h (inchash::hash::add_poly_int): New function.
1177 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1178 Use poly_int64 rather than HOST_WIDE_INT.
1179 (ao_ref::max_size_known_p): New function.
1180 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1181 rather than HOST_WIDE_INT.
1182 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1183 to temporaries until its interface is adjusted to match.
1184 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1185 (aliasing_component_refs_p, decl_refs_may_alias_p)
1186 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1187 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1188 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1190 * alias.c (ao_ref_from_mem): Likewise.
1191 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1192 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1193 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1194 (maybe_trim_complex_store, maybe_trim_constructor_store)
1195 (live_bytes_read, dse_classify_store): Likewise.
1196 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1197 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1198 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1199 (vn_reference_lookup_3): Likewise.
1200 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1202 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1203 Alan Hayward <alan.hayward@arm.com>
1204 David Sherwood <david.sherwood@arm.com>
1206 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1207 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1208 instead of ranges_overlap_p.
1210 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1211 Alan Hayward <alan.hayward@arm.com>
1212 David Sherwood <david.sherwood@arm.com>
1214 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1215 sizes as poly_int64s rather than HOST_WIDE_INTs.
1217 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1222 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1223 * gimple-fold.c (fold_ctor_reference): Likewise.
1225 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1226 Alan Hayward <alan.hayward@arm.com>
1227 David Sherwood <david.sherwood@arm.com>
1229 * target.def (dwarf_poly_indeterminate_value): New hook.
1230 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1231 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1232 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1233 * doc/tm.texi: Regenerate.
1234 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1235 offset as a poly_int64.
1236 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1237 offset as a poly_int64.
1238 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1239 Take the offset as a poly_int64.
1240 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1241 (frame_pointer_fb_offset): Change to a poly_int64.
1242 (int_loc_descriptor): Take the offset as a poly_int64. Use
1243 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1244 (based_loc_descr): Take the offset as a poly_int64.
1245 Use strip_offset_and_add to handle (plus X (const)).
1246 Use new_reg_loc_descr instead of an open-coded version of the
1247 previous implementation.
1248 (mem_loc_descriptor): Handle CONST_POLY_INT.
1249 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1250 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1252 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1253 Alan Hayward <alan.hayward@arm.com>
1254 David Sherwood <david.sherwood@arm.com>
1256 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1257 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1258 * inchash.h (inchash::hash::add_poly_hwi): New function.
1259 * gengtype.c (main): Register poly_int64.
1260 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1261 offset as a poly_int.
1262 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1263 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1264 offset as a poly_int64.
1265 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1266 * print-rtl.c (print_poly_int): New function.
1267 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1269 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1271 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1272 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1273 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1274 Enforce track_offset_p for parts of a PARALLEL.
1275 (vt_add_function_parameter): Use const_offset for the final
1276 offset to track. Use get_tracked_reg_offset for the parts
1279 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayward <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1284 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1285 * doc/tm.texi: Regenerate.
1286 * hooks.h (hook_bool_uint_uint_true): Delete.
1287 (hook_bool_puint64_puint64_true): Declare.
1288 * hooks.c (hook_bool_uint_uint_true): Delete.
1289 (hook_bool_puint64_puint64_true): New function.
1290 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1291 instead of unsigned ints.
1292 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1293 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1295 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1296 Alan Hayward <alan.hayward@arm.com>
1297 David Sherwood <david.sherwood@arm.com>
1299 * optabs.h (expand_operand): Add an int_value field.
1300 (create_expand_operand): Add an int_value parameter and use it
1301 to initialize the new expand_operand field.
1302 (create_integer_operand): Replace with a declaration of a function
1303 that accepts poly_int64s. Move the implementation to...
1304 * optabs.c (create_integer_operand): ...here.
1305 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1306 the mode preserves the value of int_value, instead of calling
1307 const_int_operand on the rtx. Use gen_int_mode to generate
1310 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1311 Alan Hayward <alan.hayward@arm.com>
1312 David Sherwood <david.sherwood@arm.com>
1314 * dumpfile.h (dump_dec): Declare.
1315 * dumpfile.c (dump_dec): New function.
1316 * pretty-print.h (pp_wide_integer): Turn into a function and
1317 declare a poly_int version.
1318 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1320 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1321 Alan Hayward <alan.hayward@arm.com>
1322 David Sherwood <david.sherwood@arm.com>
1324 * doc/generic.texi (POLY_INT_CST): Document.
1325 * tree.def (POLY_INT_CST): New tree code.
1326 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1327 * tree-core.h (tree_poly_int_cst): New struct.
1328 (tree_node): Add a poly_int_cst field.
1329 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1330 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1331 instead of a wide_int_ref.
1332 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1334 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1335 instead of an unsigned HOST_WIDE_INT.
1336 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1337 (ptrdiff_tree_p): Declare.
1338 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1339 extern inline implementations if the target doesn't use POLY_INT_CST.
1340 (poly_int_tree_p): New function.
1341 (wi::unextended_tree): New class.
1342 (wi::int_traits <unextended_tree>): New override.
1343 (wi::extended_tree): Add a default constructor.
1344 (wi::extended_tree::get_tree): New function.
1345 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1346 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1347 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1348 (wi::tree_to_poly_wide_ref): New typedefs.
1349 (wi::ints_for): Provide overloads for extended_tree and
1351 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1352 (wi::to_wide): New functions.
1353 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1354 * tree.c (poly_int_cst_hasher): New struct.
1355 (poly_int_cst_hash_table): New variable.
1356 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1357 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1359 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1360 (init_ttree): Initialize poly_int_cst_hash_table.
1361 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1362 a poly_int64 instead of a HOST_WIDE_INT.
1363 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1364 instead of an unsigned HOST_WIDE_INT.
1365 (wide_int_to_tree): Rename to...
1366 (wide_int_to_tree_1): ...this.
1367 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1368 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1369 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1370 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1371 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1372 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1374 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1375 (streamer_read_tree_body): Likewise.
1376 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1377 (streamer_write_tree_body): Likewise.
1378 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1379 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1380 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1381 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1382 * gimple-expr.h (is_gimple_constant): Likewise.
1383 * gimplify.c (maybe_with_size_expr): Likewise.
1384 * print-tree.c (print_node): Likewise.
1385 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1386 * tree-pretty-print.c (dump_generic_node): Likewise.
1387 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1388 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1389 * tree-vrp.c (compare_values_warnv): Likewise.
1390 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1391 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1392 (force_expr_to_var_cost): Likewise.
1393 * tree-ssa-loop.c (for_each_index): Likewise.
1394 * fold-const.h (build_invariant_address, size_int_kind): Take a
1395 poly_int64 instead of a HOST_WIDE_INT.
1396 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1397 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1399 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1400 (int_const_binop_2): New function, split out from...
1401 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1402 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1403 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1404 * gimple-ssa-strength-reduction.c (slsr_process_add)
1405 (slsr_process_mul): Check for INTEGER_CSTs before using them
1407 * stor-layout.c (bits_from_bytes): New function.
1408 (bit_from_pos): Use it.
1409 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1410 by BITS_PER_UNIT to get the TYPE_SIZE.
1411 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1412 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
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 * doc/rtl.texi (const_poly_int): Document. Also document the
1419 rtl sharing behavior.
1420 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1421 * rtl.h (const_poly_int_def): New struct.
1422 (rtx_def::u): Add a cpi field.
1423 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1424 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1425 (wi::rtx_to_poly_wide_ref): New typedef
1426 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1427 (poly_int_rtx_p): New functions.
1428 (trunc_int_for_mode): Declare a poly_int64 version.
1429 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1430 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1432 (strip_offset): Declare.
1433 (strip_offset_and_add): New function.
1434 * rtl.def (CONST_POLY_INT): New rtx code.
1435 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1436 (shared_const_p): Use poly_int_rtx_p.
1437 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1439 (gen_int_shift_amount): Likewise.
1440 * emit-rtl.c (const_poly_int_hasher): New class.
1441 (const_poly_int_htab): New variable.
1442 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1443 (const_poly_int_hasher::hash): New function.
1444 (const_poly_int_hasher::equal): Likewise.
1445 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1446 (immed_wide_int_const): Rename to...
1447 (immed_wide_int_const_1): ...this and make static.
1448 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1449 instead of a wide_int_ref.
1450 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1451 (gen_lowpart_common): Handle CONST_POLY_INT.
1452 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1453 * cselib.c (cselib_hash_rtx): Likewise.
1454 * dwarf2out.c (const_ok_for_output_1): Likewise.
1455 * expr.c (convert_modes): Likewise.
1456 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1457 * rtlhash.c (add_rtx): Likewise.
1458 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1459 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1460 Handle existing CONST_POLY_INT rtxes.
1461 * expmed.h (expand_shift): Take a poly_int64 instead of a
1463 * expmed.c (expand_shift): Likewise.
1464 * rtlanal.c (strip_offset): New function.
1465 (commutative_operand_precedence): Give CONST_POLY_INT the same
1466 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1468 * rtl-tests.c (const_poly_int_tests): New struct.
1469 (rtl_tests_c_tests): Use it.
1470 * simplify-rtx.c (simplify_const_unary_operation): Handle
1472 (simplify_const_binary_operation): Likewise.
1473 (simplify_binary_operation_1): Fold additions of symbolic constants
1474 and CONST_POLY_INTs.
1475 (simplify_subreg): Handle extensions and truncations of
1477 (simplify_const_poly_int_tests): New struct.
1478 (simplify_rtx_c_tests): Use it.
1479 * wide-int.h (storage_ref): Add default constructor.
1480 (wide_int_ref_storage): Likewise.
1481 (trailing_wide_ints): Use GTY((user)).
1482 (trailing_wide_ints::operator[]): Add a const version.
1483 (trailing_wide_ints::get_precision): New function.
1484 (trailing_wide_ints::extra_size): Likewise.
1486 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1487 Alan Hayward <alan.hayward@arm.com>
1488 David Sherwood <david.sherwood@arm.com>
1490 * emit-rtl.h (gen_int_shift_amount): Declare.
1491 * emit-rtl.c (gen_int_shift_amount): New function.
1492 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1494 * calls.c (shift_return_value): Likewise.
1495 * cse.c (fold_rtx): Likewise.
1496 * dse.c (find_shift_sequence): Likewise.
1497 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1498 (expand_shift, expand_smod_pow2): Likewise.
1499 * lower-subreg.c (shift_cost): Likewise.
1500 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1501 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1502 (expand_vec_perm_var): Likewise.
1503 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1504 (simplify_binary_operation_1): Likewise.
1505 * combine.c (try_combine, find_split_point, force_int_to_mode)
1506 (simplify_shift_const_1, simplify_shift_const): Likewise.
1507 (change_zero_ext): Likewise. Use simplify_gen_binary.
1509 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1511 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1513 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1515 * doc/invoke.texi (ARM Options): Document accepted extension options
1516 for -march=armv8.3-a.
1518 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1521 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1522 or --with-float=softfp, set the default CPU to arm10e.
1524 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1526 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1527 * config/visium/predicates.md (const_shift_operand): Likewise.
1528 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1529 (visium_legitimize_reload_address): Likewise.
1531 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1534 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1536 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1539 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1541 2017-12-20 Julia Koval <julia.koval@intel.com>
1543 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1544 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1545 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1546 * config.gcc: Include vpclmulqdqintrin.h.
1547 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1548 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1550 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1551 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1552 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1553 * config/i386/i386.c (isa2_opts): Add -mcx16.
1554 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1555 (ix86_option_override_internal): Move mcx16 to flags2.
1556 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1557 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1558 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1559 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1560 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1561 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1562 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1563 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1564 * doc/invoke.texi: Add -mvpclmulqdq.
1566 2017-12-20 Tom de Vries <tom@codesourcery.com>
1569 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1570 * calls.c (rtx_for_static_chain): ... here. New function.
1571 * calls.h (rtx_for_static_chain): Declare.
1572 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1573 instead of targetm.calls.static_chain.
1574 * df-scan.c (df_get_entry_block_def_set): Same.
1576 2017-12-19 Tom de Vries <tom@codesourcery.com>
1578 PR tree-optimization/83493
1579 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1582 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1584 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1585 inform with hardcoded english plural handling.
1587 2017-12-18 Jeff Law <law@redhat.com>
1589 PR tree-optimization/83477
1590 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1591 a non-virtual PHI, always push a new range.
1593 2017-12-19 Martin Sebor <msebor@redhat.com>
1596 * builtins.c (compute_objsize): Handle non-constant offsets.
1598 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1600 PR tree-optimization/83444
1601 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1602 character load case, if get_stridx on MEM_REF's operand doesn't
1603 look usable, retry with get_addr_stridx.
1605 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1608 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1609 (variable_tracking_main_1): Keep markers even when VTA fails.
1612 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1613 even if there are markers before it.
1614 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1616 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1618 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1619 typical order conditions.
1620 * sel-sched.c (extract_new_fences_from): Likewise.
1621 * config/visium/constraints.md (J, K, L): Likewise.
1622 * config/visium/predicates.md (const_shift_operand): Likewise.
1623 * config/visium/visium.c (visium_legitimize_address,
1624 visium_legitimize_reload_address): Likewise.
1625 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1626 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1627 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1628 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1629 SET_DUMP_DETAIL): Likewise.
1630 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1631 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1632 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1633 avr_set_core_architecture, avr_set_current_function,
1634 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1635 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1636 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1637 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1638 avr_map_decompose, avr_fold_builtin): Likewise.
1639 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1640 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1641 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1642 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1643 * config/m32c/m32c.c (m32c_conditional_register_usage,
1644 m32c_address_cost): Likewise.
1645 * config/m32c/predicates.md (shiftcount_operand,
1646 longshiftcount_operand): Likewise.
1647 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1648 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1649 can_use_cdx_ldstw): Likewise.
1650 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1651 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1653 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1654 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1655 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1656 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1657 * config/vax/vax.c (adjacent_operands_p): Likewise.
1658 * config/ft32/constraints.md (L, b, KA): Likewise.
1659 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1661 * cfgexpand.c (expand_stack_alignment): Likewise.
1662 * gcse.c (insert_expr_in_table): Likewise.
1663 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1664 * cgraphunit.c (cgraph_node::expand): Likewise.
1665 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1666 * emit-rtl.c (add_insn): Likewise.
1667 * input.c (dump_location_info): Likewise.
1668 * passes.c (NEXT_PASS): Likewise.
1669 * read-rtl-function.c (parse_note_insn_name,
1670 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1672 * sched-rgn.c (sched_rgn_init): Likewise.
1673 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1674 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1675 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1676 * explow.c (eliminate_constant_term): Likewise.
1677 * final.c (leaf_renumber_regs_insn): Likewise.
1678 * cfgrtl.c (print_rtl_with_bb): Likewise.
1679 * genhooks.c (emit_init_macros): Likewise.
1680 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1681 * tree-data-ref.c (conflict_fn): Likewise.
1682 * selftest.c (assert_streq): Likewise.
1683 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1684 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1685 fold_binary_loc, multiple_of_p): Likewise.
1686 * reload.c (push_reload, find_equiv_reg): Likewise.
1687 * et-forest.c (et_nca, et_below): Likewise.
1688 * dbxout.c (dbxout_symbol_location): Likewise.
1689 * reorg.c (relax_delay_slots): Likewise.
1690 * dojump.c (do_compare_rtx_and_jump): Likewise.
1691 * gengtype-parse.c (type): Likewise.
1692 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1693 simplify_const_relational_operation): Likewise.
1694 * reload1.c (do_output_reload): Likewise.
1695 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1696 * gengtype.c (type_for_name): Likewise.
1697 * gimple-ssa-sprintf.c (format_directive): Likewise.
1699 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1702 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1703 accessing it. Adjust comment.
1705 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1708 * predict.c (zero_one_minusone): New function.
1709 (apply_return_prediction): Avoid return prediction for functions
1710 returning only -1, 0 and 1 values, unless they only return -1 and 0
1713 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1715 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1716 fall-through warning.
1717 (arc_compute_frame_size): Remove unused variables.
1718 (arc_print_operand): Fix fprintif format.
1719 (arc_can_follow_jump): Clean fall-through warning.
1721 2017-12-19 Marek Polacek <polacek@redhat.com>
1724 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1727 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1729 * config/arc/arc.c (overriderregs): New variable.
1730 (arc_override_options): Track fixed/call saved/call options.
1731 (arc_conditional_register_usage): Check against overrideregs
1732 variable whenever we change register properties.
1734 2017-12-19 Nathan Sidwell <nathan@acm.org>
1736 * opts.c (finish_options): Don't prefix dump_base_name if it
1737 already contains directories.
1739 2017-12-19 Martin Liska <mliska@suse.cz>
1741 PR rtl-optimization/82675
1742 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1743 more element in sbitmap.
1745 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1747 * gimplify.c (gimplify_expr): Use error_operand_p.
1749 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1752 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1755 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1757 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1758 the jump is to a label.
1760 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1762 PR tree-optimization/83444
1763 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1768 * ipa-inline.c (flatten_remove_node_hook): New function.
1769 (ipa_inline): Keep only nodes with flatten attribute at the end of
1770 the array in the order from ipa_reverse_postorder, only walk that
1771 portion of array for flattening, if there is more than one such
1772 node, temporarily register a removal hook and ignore removed nodes.
1774 PR tree-optimization/80631
1775 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1776 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1777 IFN_REDUC_MAX or IFN_REDUC_MIN.
1779 2017-12-18 Jeff Law <law@redhat.com>
1781 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1782 record symbolic equivalences from backedges in the CFG.
1785 2017-11-19 Jeff Law <law@redhat.com>
1787 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1788 of degenerates resulting from ignoring an edge.
1790 2017-12-18 Martin Sebor <msebor@redhat.com>
1793 PR tree-optimization/78450
1794 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1795 (handle_builtin_strlen): Call it.
1797 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1799 PR rtl-optimization/83424
1800 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1802 2017-12-18 Marek Polacek <polacek@redhat.com>
1805 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1806 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1809 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1811 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1812 (umaddsidi4, umaddsidi_split): Likewise.
1814 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1816 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1819 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1822 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1824 2017-12-18 Richard Biener <rguenther@suse.de>
1826 PR tree-optimization/81877
1827 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1828 (outermost_indep_loop): Adjust.
1829 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1830 (can_sm_ref_p): Adjust.
1832 2017-12-18 Richard Biener <rguenther@suse.de>
1835 * tree.c (array_at_struct_end_p): Return true if the underlying
1836 object has space for at least one element in excess of what
1837 the array domain specifies.
1839 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1841 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1842 example to avoid overfull hbox.
1843 * doc/invoke.texi (Option Summary): Add missing @gol.
1844 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1847 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1848 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1850 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1851 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1852 (AArch64 Options): Document -mverbose-cost-dump.
1853 (ARM Options): Likewise, plus -mflip-thumb.
1855 2017-12-17 Martin Sebor <msebor@redhat.com>
1858 * gimple-ssa-warn-restrict.c
1859 (builtin_memref::offset_out_of_bounds): Correct the handling of
1862 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1864 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1866 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1868 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1870 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1872 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1875 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1877 * doc/invoke.texi: Fix some typos.
1879 2017-12-16 Martin Sebor <msebor@redhat.com>
1881 PR tree-optimization/78918
1882 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1883 * builtins.c (check_sizes): Rename...
1884 (check_access): ...to this. Rename function arguments for clarity.
1885 (check_memop_sizes): Adjust names.
1886 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1887 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1888 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1889 (check_strncat_sizes, expand_builtin_strncat): Same.
1890 (expand_builtin_strncpy, expand_builtin_memset): Same.
1891 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1892 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1893 (maybe_emit_sprintf_chk_warning): Same.
1894 (expand_builtin_strcpy): Adjust.
1895 (expand_builtin_stpcpy): Same.
1896 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1897 in pointer-checking forms of memcpy, memmove, and mempcpy.
1898 (gcall_to_tree_minimal, max_object_size): Define new functions.
1899 * builtins.h (max_object_size): Declare.
1900 * calls.c (alloc_max_size): Call max_object_size instead of
1901 hardcoding ssizetype limit.
1902 (get_size_range): Handle new argument.
1903 * calls.h (get_size_range): Add a new argument.
1904 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1905 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1906 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1908 (gimple_fold_builtin_memory_chk): Same.
1909 (gimple_fold_builtin_stxcpy_chk): New function.
1910 * gimple-ssa-warn-restrict.c: New source.
1911 * gimple-ssa-warn-restrict.h: New header.
1912 * gimple.c (gimple_build_call_from_tree): Propagate location.
1913 * passes.def (pass_warn_restrict): Add new pass.
1914 * tree-pass.h (make_pass_warn_restrict): Declare.
1915 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1917 (handle_builtin_strcat): Same.
1918 (strlen_optimize_stmt): Rename...
1919 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1920 stpncpy, strncpy, and their checking forms.
1922 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1924 PR rtl-optimization/82849
1925 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1926 and get_max_loop_iterations_int.
1928 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1929 Alan Hayward <alan.hayward@arm.com>
1930 David Sherwood <david.sherwood@arm.com>
1932 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1933 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1934 size as a poly_uint64.
1935 (mode_for_vector, mode_for_int_vector): Take the number of vector
1936 elements as a poly_uint64.
1937 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1938 size as a poly_uint64.
1939 (mode_for_vector, mode_for_int_vector): Take the number of vector
1940 elements as a poly_uint64.
1942 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1943 Alan Hayward <alan.hayward@arm.com>
1944 David Sherwood <david.sherwood@arm.com>
1946 * machmode.h (MACRO_MODE): New macro.
1947 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1948 * targhooks.c (default_libcall_value, default_secondary_reload)
1949 (default_memory_move_cost, default_register_move_cost)
1950 (default_class_max_nregs): Likewise.
1952 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1953 Alan Hayward <alan.hayward@arm.com>
1954 David Sherwood <david.sherwood@arm.com>
1956 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1957 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1959 * genautomata.c (main): Likewise.
1960 * genconditions.c (write_header): Likewise.
1961 * genemit.c (main): Likewise.
1962 * genextract.c (print_header): Likewise.
1963 * genopinit.c (main): Likewise.
1964 * genoutput.c (output_prologue): Likewise.
1965 * genpeep.c (main): Likewise.
1966 * genpreds.c (write_insn_preds_c): Likewise.
1967 * genrecog.c (writer_header): Likewise.
1968 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1969 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1970 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1971 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1972 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1973 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1974 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1975 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1976 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1977 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1978 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1979 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1980 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1981 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1982 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1983 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1984 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1985 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1986 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1987 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1988 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1989 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1990 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1991 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1992 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1993 * config/darwin.c (IN_TARGET_CODE): Likewise.
1994 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1995 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1996 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1997 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1998 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1999 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2000 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2001 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2002 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2003 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2004 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2005 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2006 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2007 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2008 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2009 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2010 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2011 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2012 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2013 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2014 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2015 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2016 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2017 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2018 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2019 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2020 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2021 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2022 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2023 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2024 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2025 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2026 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2027 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2028 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2029 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2030 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2031 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2032 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2033 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2034 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2035 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2036 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2037 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2038 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2039 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2040 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2041 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2042 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2043 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2044 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2045 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2046 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2047 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2048 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2049 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2050 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2051 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2052 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2053 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2054 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2055 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2056 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2057 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2058 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2059 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2060 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2061 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2062 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2063 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2064 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2065 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2066 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2067 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2068 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2069 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2070 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2071 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2072 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2073 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2074 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2075 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2076 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2077 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2078 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2079 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2080 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2081 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2082 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2083 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2084 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2085 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2086 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2087 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2088 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2089 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2090 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2091 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2092 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2093 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2094 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2095 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2096 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2097 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2098 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2099 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2101 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2102 Alan Hayward <alan.hayward@arm.com>
2103 David Sherwood <david.sherwood@arm.com>
2105 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2108 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2109 Alan Hayward <alan.hayward@arm.com>
2110 David Sherwood <david.sherwood@arm.com>
2112 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2113 * doc/md.texi (vec_series@var{m}): Document.
2114 * tree.def (VEC_SERIES_EXPR): New tree code.
2115 * tree.h (build_vec_series): Declare.
2116 * tree.c (build_vec_series): New function.
2117 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2118 * tree-pretty-print.c (dump_generic_node): Likewise.
2119 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2120 * tree-inline.c (estimate_operator_cost): Likewise.
2121 * expr.c (expand_expr_real_2): Likewise.
2122 * optabs-tree.c (optab_for_tree_code): Likewise.
2123 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2124 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2125 * expmed.c (make_tree): Handle VEC_SERIES.
2126 * optabs.def (vec_series_optab): New optab.
2127 * optabs.h (expand_vec_series_expr): Declare.
2128 * optabs.c (expand_vec_series_expr): New function.
2129 * tree-vect-generic.c (expand_vector_operations_1): Check that
2130 the operands also have vector type.
2132 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2133 Alan Hayward <alan.hawyard@arm.com>
2134 David Sherwood <david.sherwood@arm.com>
2136 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2137 (VEC_COND_EXPR): Add missing @tindex.
2138 * doc/md.texi (vec_duplicate@var{m}): Document.
2139 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2140 * tree.c (build_vector_from_val): Add stubbed-out handling of
2141 variable-length vectors, using VEC_DUPLICATE_EXPR.
2142 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2143 * cfgexpand.c (expand_debug_expr): Likewise.
2144 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2145 * tree-inline.c (estimate_operator_cost): Likewise.
2146 * tree-pretty-print.c (dump_generic_node): Likewise.
2147 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2148 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2149 (test_vec_duplicate_folding): New function.
2150 (fold_const_c_tests): Call it.
2151 * optabs.def (vec_duplicate_optab): New optab.
2152 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2153 * optabs.h (expand_vector_broadcast): Declare.
2154 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2155 vec_duplicate_optab.
2156 * expr.c (store_constructor): Try using vec_duplicate_optab for
2158 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2160 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2163 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2164 div/mod latencies a bit.
2166 2017-12-15 Jeff Law <law@redhat.com>
2168 PR tree-optimization/36550
2169 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2170 (mark_threaded_blocks): Rewrite code to avoid block copying when
2171 optimizing for size. Don't pessimize blocks which will be
2172 copied, but all the statements will be dead.
2174 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2176 PR tree-optimization/81165
2177 * tree-ssa-threadupdate.c (uses_in_bb): New.
2178 (estimate_threading_killed_stmts): New.
2179 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2180 * tree-ssa-threadedge.c
2181 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2184 2017-12-15 Jeff Law <law@redhat.com>
2186 PR tree-optimization/83410
2187 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2188 threads when parallelizing loops.
2190 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2192 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2194 * config/alpha/alpha.c (vms_attribute_table): Swap
2195 affects_type_identity and handler fields, adjust comments.
2196 * config/mips/mips.c (mips_attribute_table): Likewise.
2197 * config/visium/visium.c (visium_attribute_table): Likewise.
2198 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2199 * config/microblaze/microblaze.c (microblaze_attribute_table):
2201 * config/spu/spu.c (spu_attribute_table): Likewise.
2202 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2203 * config/arc/arc.c (arc_attribute_table): Likewise.
2204 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2205 * config/v850/v850.c (v850_handle_interrupt_attribute,
2206 v850_handle_data_area_attribute): Formatting fixes.
2207 (v850_attribute_table): Swap affects_type_identity and handler
2208 fields, adjust comments.
2209 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2210 * config/arm/arm.c (arm_attribute_table): Likewise.
2211 * config/avr/avr.c (avr_attribute_table): Likewise.
2212 * config/s390/s390.c (s390_attribute_table): Likewise.
2213 * config/sh/sh.c (sh_attribute_table): Likewise.
2214 * config/i386/i386.c (ix86_handle_cconv_attribute,
2215 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2216 (ix86_attribute_table): Swap affects_type_identity and handler
2217 fields, adjust comments.
2218 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2219 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2220 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2221 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2222 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2223 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2224 * config/rx/rx.c (rx_attribute_table): Likewise.
2225 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2226 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2227 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2228 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2229 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2230 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2231 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2232 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2233 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2234 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2235 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2237 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2239 (register_scoped_attributes, decl_attributes): Formatting fixes.
2241 PR tree-optimization/83269
2242 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2243 subtraction in arg0's type if type is signed and arg0 is unsigned.
2247 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2248 convert? on @0 instead of convert. Check type of @1, not @0.
2249 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2250 convert? on @0 instead of convert. Check type of @1, not @0.
2251 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2252 only optimize if either both @1 and @2 types are narrower
2253 precision, or both are wider or equal precision, and in the former
2254 case only if both have undefined overflow.
2256 2017-12-15 Richard Biener <rguenther@suse.de>
2259 * internal-fn.def (IFN_NOP): Add.
2260 * internal-fn.c (expand_NOP): Do nothing.
2261 * lto-streamer-in.c (input_function): Instead of removing
2262 sanitizer calls replace them with IFN_NOP calls.
2264 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2265 Alan Hayward <alan.hayward@arm.com>
2266 David Sherwood <david.sherwood@arm.com>
2268 * dse.c (store_info, read_info_type): Replace begin and end with
2270 (print_range): New function.
2271 (set_all_positions_unneeded, any_positions_needed_p)
2272 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2274 (record_store): Likewise. Optimize the case in which all positions
2276 (get_stored_val): Replace read_begin and read_end with read_offset
2278 (replace_read): Update call accordingly.
2280 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2282 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2283 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2284 (loop_cand::loop_cand): Initialize above members.
2285 (loop_cand::supported_operations): Delete.
2286 (loop_cand::can_interchange_p): Inline above function.
2287 (loop_cand::classify_simple_reduction): Record number of constant
2288 initialized simple reductions.
2289 (should_interchange_loops): New parameters. Check stmt cost of loops
2291 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2292 Update call to should_interchange_loops.
2293 (should_interchange_loop_nest): Update call to
2294 should_interchange_loops.
2296 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2299 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2300 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2301 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2303 2017-12-15 Julia Koval <julia.koval@intel.com>
2305 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2306 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2307 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2308 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2309 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2311 2017-12-15 Julia Koval <julia.koval@intel.com>
2313 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2314 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2315 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2316 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2317 _mm_aesenc_epi128): New intrinsics.
2319 2017-12-15 Julia Koval <julia.koval@intel.com>
2321 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2322 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2323 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2324 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2325 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2327 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2329 * gimple-ssa-strength-reduction.c (analyze_increments):
2330 Distinguish replacement costs for constant strides from those for
2333 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2335 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2337 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2339 * doc/invoke.texi: Document -Wcast-function-type.
2340 * recog.h (stored_funcptr): Change signature.
2341 * tree-dump.c (dump_node): Avoid warning.
2342 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2344 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2347 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2350 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2352 PR tree-optimization/83312
2353 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2354 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2355 "val" param, and to cope with arbitrary basic blocks.
2356 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2357 handle NULL_TREE for "val", dropping "bb" param.
2358 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2359 drop "bb" param. Handle NULL_TREE for "val".
2360 (find_case_label_for_value): Make "switch_stmt" param const.
2361 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2363 (vrp_prop::check_all_array_refs): Reimplement as...
2364 (check_array_bounds_dom_walker::before_dom_children): ...this new
2365 vfunc. Replace linear search through BB block list, excluding
2366 those with non-executable in-edges via dominator walk.
2368 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2370 * config/arm/arm.opt (mverbose-cost-dump): New option.
2371 * config/arm/arm.c (arm_rtx_costs): Use it.
2373 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2376 * reload1.c (emit_input_reload_insns): Skip debug markers.
2378 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2380 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2384 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2385 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2386 * config/sh/sh.c: Skip debug insns besides notes.
2387 * config/sh/sh.md: Likewise.
2388 * config/sh/sh_treg_combine.cc: Likewise.
2389 * config/sh/sync.md: Likewise.
2391 2017-12-14 Tom de Vries <tom@codesourcery.com>
2393 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2394 Add item for weak_undefined.
2396 2017-12-14 Richard Biener <rguenther@suse.de>
2398 PR tree-optimization/67842
2399 * sese.h (bb_in_region): Remove #if 0'ed code.
2401 2017-12-14 Richard Biener <rguenther@suse.de>
2403 PR tree-optimization/83326
2404 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2405 may_be_zero parameter and handle it by not marking the first
2406 peeled copy as not exiting the loop.
2407 (try_peel_loop): Likewise.
2408 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2409 to handle the case of constant or zero iterations and perform
2410 loop header copying on-the-fly.
2412 2017-12-14 Richard Biener <rguenther@suse.de>
2414 PR tree-optimization/83418
2415 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2416 Instead of asserting we don't get unfolded comparisons deal with
2419 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2422 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2423 basic blocks. Assert debug bind insns don't appear outside of bbs,
2424 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2427 PR tree-optimization/83198
2428 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2429 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2430 value if arg is a REAL_CST with incompatible type.
2432 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2433 Bin Cheng <bin.cheng@arm.com>
2436 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2438 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2441 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2443 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2444 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2446 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2447 Alan Hayward <alan.hayward@arm.com>
2448 David Sherwood <david.sherwood@arm.com>
2450 * poly-int.h: New file.
2451 * poly-int-types.h: Likewise.
2452 * coretypes.h: Include them.
2453 (POLY_INT_CONVERSION): Define.
2454 * target.def (estimated_poly_value): New hook.
2455 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2456 * doc/tm.texi: Regenerate.
2457 * doc/poly-int.texi: New file.
2458 * doc/gccint.texi: Include it.
2459 * doc/rtl.texi: Describe restrictions on subreg modes.
2460 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2461 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2462 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2463 * targhooks.h (default_estimated_poly_value): Declare.
2464 * targhooks.c (default_estimated_poly_value): New function.
2465 * target.h (estimated_poly_value): Likewise.
2466 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2467 (wi::unary_traits): Delete.
2468 (wi::binary_traits::signed_shift_result_type): Define for
2469 offset_int << HOST_WIDE_INT, etc.
2470 (generic_wide_int::operator <<=): Define for all types and use
2471 wi::lshift instead of <<.
2472 (wi::hwi_with_prec): Add a default constructor.
2473 (wi::ints_for): New class.
2474 (operator <<): Define for all wide-int types.
2475 (operator /): New function.
2476 (operator %): Likewise.
2477 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2478 (ASSERT_MAYBE_NE_AT): New macros.
2480 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2481 Dominik Vogt <vogt@linux.vnet.ibm.com>
2484 * emit-rtl.c (init_emit): Remove ??? comment.
2485 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2486 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2487 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2488 alignment of 3 virtual registers to BITS_PER_WORD.
2490 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2492 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2494 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2495 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2497 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2500 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2502 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2504 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2505 with the exception of debug begin stmt markers appear before
2509 * final.c (rest_of_handle_final): Call variable_tracking_main only
2510 if !flag_var_tracking.
2512 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2513 Jakub Jelinek <jakub@redhat.com>
2517 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2518 labels debug stmts that can only appear after labels.
2520 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2522 PR rtl-optimization/82398
2523 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2524 EXPR_USEFULNESS in priority comparison.
2526 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2528 PR rtl-optimization/83393
2529 * combine.c (move_deaths): If reg_stat points to a too new insn in
2530 last_death, do not use it: find the proper insn instead.
2532 2017-12-12 Jeff Law <law@redhat.com>
2534 PR tree-optimization/83298
2535 PR tree-optimization/83362
2536 PR tree-optimization/83383
2537 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2538 push_value_range a public interface. Add new argument to
2539 record_ranges_from_stmt.
2540 * gimple-ssa-evrp-analyze.c
2541 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2542 Update comments. Handle recording temporary equivalences.
2543 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2544 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2545 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2546 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2547 gimple-ssa-evrp-analyze.h.
2548 (record_temporary_equivalences_from_phis): Add new argument. When
2549 the PHI arg is an SSA_NAME, set the result's range to the range
2551 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2552 from statements too.
2553 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2554 Pass it down to children as needed.
2555 (thread_outgoing_edges): Likewise.
2556 (thread_across_edge): Likewise. Push/pop range state as needed.
2557 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2559 2017-12-12 Julia Koval <julia.koval@intel.com>
2561 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2562 (PTA_CANNONLAKE): Remove PTA_CLWB.
2564 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2567 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2568 (vcondv2div2df): Likewise.
2569 (vconduv2dfv2di): Likewise.
2571 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2573 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2574 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2575 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2576 ATTR_MATHFN_FPROUNDING_ERRNO.
2578 2017-12-12 Richard Biener <rguenther@suse.de>
2580 PR tree-optimization/83385
2581 * graphite-scop-detection.c (get_order, order): Remove.
2582 (bb_to_rpo): New global.
2584 (build_scops): Sort pbbs in RPO order.
2586 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2588 * combine.c (simplify_set): Do not transform subregs to zero_extends
2589 if the destination mode is a vector mode.
2591 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2593 PR tree-optimization/83359
2594 * tree-cfg.h (fold_loop_internal_call): Declare.
2595 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2596 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2597 (find_loop_dist_alias): New function.
2598 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2599 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2602 PR tree-optimization/80631
2603 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2604 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2605 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2606 hardcoding zero as the value if COND_EXPR is never true. For
2607 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2608 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2609 hardcoding MAX_EXPR as the reduction operation.
2610 (is_nonwrapping_integer_induction): Allow negative step.
2611 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2612 vect_create_epilog_for_reduction, if no value is suitable, don't
2613 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2615 2017-12-12 Richard Biener <rguenther@suse.de>
2617 PR tree-optimization/81889
2618 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2619 range info from the non-wrapping IV instead of just the range
2622 2017-12-12 Julia Koval <julia.koval@intel.com>
2624 * config.gcc: Add vaesintrin.h.
2625 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2627 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2628 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2630 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2631 * config/i386/immintrin.h: Include vaesintrin.h.
2632 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2633 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2634 _mm_aesdec_epi128): New intrinsics.
2636 2017-12-12 Julia Koval <julia.koval@intel.com>
2638 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2639 OPTION_MASK_ISA_VAES_UNSET): New.
2640 (ix86_handle_option): Handle -mvaes.
2641 * config/i386/cpuid.h: Define bit_VAES.
2642 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2643 * config/i386/i386-c.c (__VAES__): New.
2644 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2645 (ix86_valid_target_attribute_inner_p): Ditto.
2646 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2647 * config/i386/i386.opt: Add -mvaes.
2648 * doc/invoke.texi: Ditto.
2650 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2652 * debug.h (gcc_debug_hooks): Add inline_entry.
2653 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2654 * debug.c (do_nothing_debug_hooks): Likewise.
2655 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2656 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2657 (dwarf2_lineno_debug_hooks): Likewise.
2659 * common.opt (gstatement-frontiers): New, setting
2660 debug_nonbind_markers_p.
2661 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2662 * toplev.c (process_options): Autodetect value for debug statement
2664 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2665 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2667 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2668 markers. Integrate source bind into debug stmt expand loop.
2669 (pass_expand::execute): Check debug marker limit. Avoid deep
2670 TER and expand debug locations for debug bind insns only.
2671 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2673 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2674 * final.c (reemit_insn_block_notes): Take current block from
2675 nonbind markers. Declare note where it's first set.
2676 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2677 begin stmt markers if enabled.
2678 (notice_source_line): Handle nonbind markers. Fail if their
2679 location is unknown or that of builtins.
2680 (rest_of_handle_final): Convert begin stmt markers to notes if
2681 var-tracking didn't run.
2682 (rest_of_clean_state): Skip begin stmt markers.
2683 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2685 * function.c (allocate_struct_function): Set begin_stmt_markers.
2686 * function.h (struct function): Add debug_marker_count counter
2687 and debug_nonbind_markers flag.
2688 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2689 * gimple-low.c (lower_function_body): Adjust
2690 debug_nonbind_markers.
2691 (lower_stmt): Drop or skip gimple debug stmts.
2692 (lower_try_catch): Skip debug stmts.
2693 * gimple.c (gimple_build_debug_begin_stmt): New.
2694 (gimple_copy): Increment debug_marker_count if copying one.
2695 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2696 * gimplify.c (rexpr_location): New.
2697 (rexpr_has_location): New.
2698 (warn_switch_unreachable_r): Handle gimple debug stmts.
2699 (shortcut_cond_r): Call expr_location.
2701 (find_goto_label): New.
2702 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2704 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2706 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2707 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2708 (LANG_HOOKS_INITIALIZER): ... this.
2709 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2710 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2712 (update_ebb_live_info): Skip debug insn markers.
2713 * lra.c (debug_insn_static_data): Rename to...
2714 (debug_bind_static_data): ... this.
2715 (debug_marker_static_data): New.
2716 (lra_set_insn_recog_data): Select one of the above depending
2718 (lra_update_isn_regno_info): Don't assume debug insns have
2720 (push_insns): Skip debug insns.
2721 * lto-streamer-in.c (input_function): Drop debug stmts
2722 depending on active options. Adjust debug_nonbind_markers.
2723 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2724 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2725 begin stmt marker notes.
2726 (print_insn): Likewise.
2727 * recog.c (extract_insn): Recognize rtl for debug markers.
2728 * rtl.def (DEBUG_MARKER): New.
2729 * tree-inline.c: Include params.h.
2730 (remap_gimple_stmt): Handle nonbind markers.
2731 (maybe_move_debug_stmts_to_successors): Likewise.
2732 (copy_debug_stmt): Likewise.
2733 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2734 markers regardless of no side effects.
2735 (tsi_link_before): Don't update container's side effects when adding
2736 a begin stmt marker.
2737 (tsi_link_after): Likewise.
2738 (expr_first): Skip begin stmt markers.
2739 (expr_last): Likewise.
2740 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2741 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2742 Disregard nonbind markers.
2743 * tree.c (make_node_stat): Don't set side effects for begin stmt
2745 (build1_stat): Likewise.
2746 * tree.def (DEBUG_BEGIN_STMT): New.
2747 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2748 * var-tracking.c (delete_debug_insns): Renamed to...
2749 (delete_vta_debug_insns): ... this.
2750 (reemit_marker_as_note): New.
2751 (vt_initialize): Reemit markers.
2752 (delete_vta_debug_insns): Likewise.
2753 (vt_debug_insns_local): Reemit or delete markers.
2754 (variable_tracking_main_1): Likewise.
2755 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2756 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2757 (gimple_debug_nonbind_marker_p): New.
2758 (gimple_build_debug_bind): Adjust.
2759 (gimple_build_debug_begin_stmt): New.
2760 * doc/invoke.texi (max-debug-marker-count): New param.
2761 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2762 (debug_parameter_ref, debug_marker): New.
2763 (NOTE_INSN_BEGIN_STMT): New.
2764 (DEBUG_INSN): Describe begin stmt markers.
2766 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2767 without debug insns, we wouldn't, but clean up debug insns
2768 after a control flow insn nevertheless.
2770 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2771 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2772 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2773 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2774 (rtl_tidy_fallthru_edge): Likewise.
2775 (rtl_verify_fallthru): Likewise.
2776 (rtl_verify_bb_layout): Likewise.
2777 (skip_insns_after_block): Likewise.
2778 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2779 * dwarf2out.c: Include print-rtl.h.
2780 (dwarf2out_next_real_insn): New.
2781 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2782 Dump debug binds in asm comments.
2783 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2784 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2785 callers to use gsi_start_nondebug_bb instead.
2786 (gsi_after_labels): Skip gimple debug stmts.
2787 (gsi_start_nondebug): New.
2788 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2789 (proper_loop_form_for_interchange): Adjust.
2790 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2791 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2792 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2793 (collect_fallthrough_labels): Likewise.
2794 (should_warn_for_implicit_fallthrough): Likewise.
2795 (warn_implicit_fallthrough_r): Likewise.
2796 (expand_FALLTHROUGH_r): Likewise.
2797 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2798 (graphite_copy_stmts_from_block): Skip nonbind markers.
2799 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2800 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2801 * jump.c (clean_barriers): Skip debug insns.
2802 * omp-expand.c (expand_parallel_call): Skip debug insns.
2803 (expand_task_call): Likewise.
2804 (remove_exit_barrier): Likewise.
2805 (expand_omp_taskreg): Likewise.
2806 (expand_omp_for_init_counts): Likewise.
2807 (expand_omp_for_generic): Likewise.
2808 (expand_omp_for_static_nochunk): Likewise.
2809 (expand_omp_for_static_chunk): Likewise.
2810 (expand_omp_simd): Likewise.
2811 (expand_omp_taskloop_for_outer): Likewise.
2812 (expand_omp_taskloop_for_inner): Likewise.
2813 (expand_oacc_for): Likewise.
2814 (expand_omp_sections): Likewise.
2815 (expand_omp_single): Likewise.
2816 (expand_omp_synch): Likewise.
2817 (expand_omp_atomic_load): Likewise.
2818 (expand_omp_atomic_store): Likewise.
2819 (expand_omp_atomic_fetch_op): Likewise.
2820 (expand_omp_atomic_pipeline): Likewise.
2821 (expand_omp_atomic_mutex): Likewise.
2822 (expand_omp_target): Likewise.
2823 (grid_expand_omp_for_loop): Likewise.
2824 (grid_expand_target_grid_body): Likewise.
2825 (build_omp_regions_1): Likewise.
2826 * omp-low.c (check_combined_parallel): Skip debug stmts.
2827 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2828 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2829 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2831 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2832 (make_edges): Likewise.
2833 (cleanup_dead_labels): Likewise.
2834 (gimple_can_merge_blocks_p): Likewise.
2835 (stmt_starts_bb_p): Likewise.
2836 (gimple_block_label): Likewise.
2837 (gimple_redirect_edge_and_branch): Likewise.
2838 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2840 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2842 * tree-pretty-print (print_declaration): Omit initializer in slim
2844 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2846 (eliminate_unnecessary_stmts): Stabilize block removal order.
2847 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2848 * var-tracking.c (get_first_insn): New.
2849 (vt_emit_notes): Call it.
2850 (vt_initialize): Walk any insns before the first BB.
2851 (delete_debug_insns): Likewise.
2853 * gimple.h (enum gimple_debug_subcode): Add
2854 GIMPLE_DEBUG_BEGIN_STMT.
2855 (gimple_debug_begin_stmt_p): New.
2856 (gimple_debug_nonbind_marker_p): New.
2857 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2858 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2859 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2860 * insn-notes.def (BEGIN_STMT): New.
2861 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2862 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2863 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2864 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2865 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2866 (INSN_DEBUG_MARKER_KIND): New.
2867 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2868 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2869 (INSN_VAR_LOCATION_PTR): New.
2870 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2872 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2873 deps only if debug bind insns are enabled.
2874 (pass_expand::execute): Avoid deep TER and expand
2875 debug locations for debug bind insns only.
2876 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2877 debug stmts special handling down to debug bind stmts.
2878 * combine.c (try_combine): Narrow debug insns special handling
2879 down to debug bind insns.
2880 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2881 Narrow debug insns preexisting special handling down to debug
2883 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2884 handling down to debug bind insns.
2885 * function.c (instantiate_virtual_regs): Skip debug markers,
2886 adjust handling of debug binds.
2887 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2888 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2889 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2890 insns to debug bind insns.
2891 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2892 special handling of debug stmts to debug bind stmts.
2893 * ipa-split.c (split_function): Likewise.
2894 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2895 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2897 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2898 * regrename.c (build_def_use): Likewise.
2899 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2900 (pass_cprop_hardreg): Narrow special casing of debug insns to
2902 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2903 * reload1.c (reload): Likewise.
2904 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2905 casing of debug stmts to debug bind stmts.
2906 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2907 * ssa-iterators.h (num_imm_uses): Likewise.
2908 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2909 debug stmts to debug bind stmts.
2910 * tree-inline.c (tree_function_versioning): Narrow special casing
2911 of debug stmts to debug bind stmts.
2912 * tree-loop-distribution.c (generate_loops_for_partition):
2913 Narrow special casing of debug stmts to debug bind stmts.
2914 * tree-sra.c (analyze_access_subtree): Narrow special casing
2915 of debug stmts to debug bind stmts.
2916 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2917 stmts to debug bind stmts.
2918 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2919 casing of debug stmts to debug bind stmts.
2920 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2921 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2922 casing of debug stmts to debug bind stmts.
2923 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2925 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2926 debug stmts to debug bind stmts.
2927 (gimple_replace_ssa_lhs): Likewise.
2928 (insert_debug_temp_for_var_def): Likewise.
2929 (insert_debug_temps_for_defs): Likewise.
2930 (reset_debug_uses): Likewise.
2931 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2932 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2933 (adjust_debug_stmts): Likewise.
2934 (adjust_phi_and_debug_stmts): Likewise.
2935 (vect_do_peeling): Likewise.
2936 * tree-vect-loop.c (vect_transform_loop): Likewise.
2937 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2938 * var-tracking.c (adjust_mems): Narrow special casing of debug
2939 insns to debug bind insns.
2940 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2941 (compute_bb_dataflow, vt_find_locations): Likewise.
2942 (vt_expand_loc, emit_notes_for_changes): Likewise.
2943 (vt_init_cfa_base): Likewise.
2944 (vt_emit_notes): Likewise.
2945 (vt_initialize): Likewise.
2946 (vt_finalize): Likewise.
2948 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2949 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2950 (next_nonnote_nondebug_insn_bb): New.
2951 (prev_nonnote_nondebug_insn_bb): New.
2952 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2953 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2954 (prev_nonnote_nondebug_insn_bb): Declare.
2955 (next_nonnote_nondebug_insn_bb): Declare.
2956 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2957 * cfgrtl.c (get_last_bb_insn): Likewise.
2958 * lra.c (push_insns): Likewise.
2960 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2963 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2965 * selftest.c (selftest::test_runner::test_runner): New ctor.
2966 (selftest::test_runner::~test_runner): New dtor.
2967 * selftest.h (class selftest::test_runner): New class.
2969 2017-12-11 Carl Love <cel@us.ibm.com>
2971 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2972 vec_extract_fp32_from_shortl]): Add #defines.
2973 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2974 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2975 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2976 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2977 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2978 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2979 * doc/extend.texi: Add documentation for the added builtins.
2981 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2983 PR rtl-optimization/80693
2984 PR rtl-optimization/81019
2985 PR rtl-optimization/81020
2986 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2987 are not mentioned in i3. Place the REG_UNUSED note on i2,
2988 possibly modified to REG_DEAD, if it did not originate in i3.
2990 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2992 * recog.c (store_data_bypass_p_1): New function.
2993 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2994 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2996 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2998 PR rtl-optimization/83361
2999 * ifcvt.c (if_convert): Call fixup_partitions.
3001 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3003 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3004 early folding of splat_u{8,16,32}.
3006 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3008 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3009 output_operand_lossage first argument with capital letter.
3010 (aarch64_override_options): Don't start error and sorry first argument
3011 with capital letter.
3013 2017-12-11 Andi Kleen <ak@linux.intel.com>
3015 PR gcov-profile/83355
3016 * auto-profile.c (string_table::get_index_by_decl): Don't
3017 recurse when abstract origin points to itself.
3019 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3021 PR tree-optimization/83320
3022 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3023 (prune_datarefs_not_in_loop): Ditto.
3025 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3027 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3029 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3031 PR tree-optimization/83337
3032 * gimple-loop-interchange.cc (compute_access_stride): Handle
3033 bitfield DRs properly.
3035 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3037 PR tree-optimization/83338
3038 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3039 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3040 vector of unsigned integers to vector of signed integers.
3042 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3044 PR rtl-optimization/83317
3045 * lra-constraints.c (process_address_1): Add insn code check.
3047 2017-12-08 Michael Matz <matz@suse.de>
3049 Fix PR tree-optimization/83323
3050 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3051 head-controlled loops and loop BBs.
3052 * common.opt (funroll-and-jam): Remove, instead ...
3053 (floop-unroll-and-jam): ... reuse this option.
3054 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3055 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3056 (-floop-unroll-and-jam): ... this option.
3058 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3060 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3061 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3062 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3064 2017-12-08 Richard Biener <rguenther@suse.de>
3067 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3068 handle accesses outside of zero-sized vars.
3070 2017-12-08 Martin Jambor <mjambor@suse.cz>
3072 PR tree-optimization/83141
3073 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3074 test for MEM_REFs implicitely changing types with padding. Remove
3076 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3078 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3080 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3082 * config/avr/avr.c (avr_attribute_table): Likewise.
3083 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3084 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3085 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3086 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3087 of struct attribute_spec.
3088 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3090 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3093 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3095 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3098 * profile-count.c (profile_count::from_gcov_type): Move from
3099 profile-count.h; handle overflow.
3100 * profile-count.h (profile_count::from_gcov_type): Move offline.
3102 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3104 PR rtl-optimization/83304
3105 * combine.c (move_deaths): If we do not know where a register died,
3108 2017-12-08 Richard Biener <rguenther@suse.de>
3110 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3111 Provide -fopt-info-loop feedback when we interchange in a nest.
3113 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3115 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3116 for armv6 ARM CPU IDs.
3118 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3120 * common/config/arm/arm-common.c: Include <algorithm>.
3121 (INCLUDE_VECTOR): Define.
3122 (compare_opt_names): New function.
3123 (arm_rewrite_selected_arch): Only strip out extensions that can be
3124 expressed through -mfpu. Sort the remaining extensions
3127 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3129 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3130 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3131 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3132 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3134 2017-06-08 Tristan Gingold <gindold@adacore.com>
3137 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3138 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3140 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3142 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3143 of struct attribute_spec.
3145 2017-12-08 Julia Koval <julia.koval@intel.com>
3147 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3148 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3149 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3150 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3151 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3152 _mm_maskz_dpwssds_epi32): Ditto.
3154 2017-12-08 Richard Biener <rguenther@suse.de>
3156 PR tree-optimization/81303
3157 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3158 conditions try to create a comparison vector type matching
3159 the data vector type.
3160 (vectorizable_condition): Adjust.
3161 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3162 Leave invariant conditions alone in case we can vectorize those.
3164 2017-12-08 Julia Koval <julia.koval@intel.com>
3166 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3167 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3168 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3169 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3170 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3172 2017-12-08 Julia Koval <julia.koval@intel.com>
3174 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3175 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3176 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3177 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3178 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3179 _mm_maskz_dpbusds_epi32): New intrinsics.
3181 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3183 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3185 (ld<bh>io_signed): Likewise.
3186 (st<bhw>io): Likewise.
3187 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3188 SMALL_INT12 constant integer operand.
3190 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3192 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3194 (nios2_large_constant_allowed): ...this. Adjust uses.
3195 (nios2_plus_symbolic_constant_p): Rename to...
3196 (nios2_plus_large_constant_p): ...this. Adjust uses.
3197 (nios2_legitimate_address_p): Correct CONST_INT handling.
3198 (nios2_symbolic_memory_operand_p): Rename to...
3199 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3200 (nios2_large_constant_p): Check for large constant integers too.
3201 (nios2_split_large_constant): Handle constant integers.
3202 (nios2_split_symbolic_memory_operand): Rename to...
3203 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3204 (nios2_legitimize_constant_address): Handle constant integers.
3205 (r0rel_constant_p): Handle small constant integers.
3206 (nios2_print_operand_address): Handle r0-relative integer addresses.
3207 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3208 * config/nios2/nios2.md: Adjust for renamed functions.
3210 2017-12-07 Andrew Waterman <andrew@sifive.com>
3212 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3213 (TARGET_HAVE_SRODATA_SECTION): New define.
3214 (riscv_select_section): New function.
3216 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3217 Alexander Monakov <amonakov@ispras.ru>
3218 Jakub Jelinek <jakub@redhat.com>
3221 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3223 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3225 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3226 a bool success value. Don't call output_operand_lossage here.
3227 (aarch64_print_ldpstp_address): Return a bool success value.
3228 (aarch64_print_operand_address): Call output_addr_const if
3229 aarch64_print_address_internal fails.
3230 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3231 'y'; call output_operand_lossage instead. Call output_operand_lossage
3232 if aarch64_print_ldpstp_address fails.
3234 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3236 * tree-vector-builder.h
3237 (tree_vector_builder::binary_encoded_nelts): Declare.
3238 * tree-vector-builder.c
3239 (tree_vector_builder::binary_encoded_nelts): New function.
3240 * fold-const.c (negate_expr_p): Likewise.
3241 (operand_equal_p, fold_checksum_tree): Likewise.
3242 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3243 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3244 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3245 (uniform_vector_p): Likewise.
3246 * varasm.c (const_hash_1, compare_constant): Likewise.
3247 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3248 (valid_lattice_transition): Operate directly on the VECTOR_CST
3250 * ipa-icf.c: Include tree-vector-builder.h.
3251 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3252 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3254 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3256 * tree.c (build_vector): Delete.
3257 * tree.h (build_vector): Make static and move into the self-testing
3260 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3262 * vector-builder.h (vector_builder::derived): New const overload.
3263 (vector_builder::elt): New function.
3264 * tree-vector-builder.h (tree_vector_builder::type): New function.
3265 (tree_vector_builder::apply_step): Declare.
3266 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3268 * gimple-fold.h (tree_vector_builder): Declare.
3269 (gimple_build_vector): Take a tree_vector_builder instead of a
3270 type and vector of elements.
3271 * gimple-fold.c (gimple_build_vector): Likewise.
3272 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3274 (get_initial_defs_for_reduction): Likewise.
3275 (vectorizable_induction): Likewise.
3277 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3279 * tree-vector-builder.h
3280 (tree_vector_builder::new_binary_operation): Declare.
3281 * tree-vector-builder.c
3282 (tree_vector_builder::new_binary_operation): New function.
3283 * fold-const.c (fold_relational_const): Use it.
3284 (const_binop): Likewise. Check that both input vectors have
3285 the same number of elements, thus excluding things like WIDEN_SUM.
3286 Check whether it is possible to operate directly on the encodings
3289 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3291 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3292 new_unary_operation, operating only on the encoded elements.
3293 (const_unop): Likewise.
3294 (exact_inverse): Likewise.
3295 (distributes_over_addition_p): New function.
3296 (const_binop): Use tree_vector_builder and new_unary_operation
3297 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3298 on the encoded elements unless the encoding is strided and the
3299 operation does not distribute over addition.
3300 (fold_convert_const): Use tree_vector_builder and
3301 new_unary_operation. Operate only on the encoded elements
3302 for truncating integer conversions, or for non-stepped encodings.
3304 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3306 * config/sparc/sparc.c: Include tree-vector-builder.h.
3307 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3308 * expmed.c: Include tree-vector-builder.h.
3309 (make_tree): Use tree_vector_builder instead of build_vector.
3310 * fold-const.c: Include tree-vector-builder.h.
3311 (const_binop): Use tree_vector_builder instead of build_vector.
3312 (const_unop): Likewise.
3313 (native_interpret_vector): Likewise.
3314 (fold_vec_perm): Likewise.
3315 (fold_ternary_loc): Likewise.
3316 * gimple-fold.c: Include tree-vector-builder.h.
3317 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3319 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3320 (simplify_vector_constructor): Use tree_vector_builder instead
3322 * tree-vect-generic.c: Include tree-vector-builder.h.
3323 (add_rshift): Use tree_vector_builder instead of build_vector.
3324 (expand_vector_divmod): Likewise.
3325 (optimize_vector_constructor): Likewise.
3326 * tree-vect-loop.c: Include tree-vector-builder.h.
3327 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3328 of build_vector. Explicitly use a stepped encoding for
3330 * tree-vect-slp.c: Include tree-vector-builder.h.
3331 (vect_get_constant_vectors): Use tree_vector_builder instead
3333 (vect_transform_slp_perm_load): Likewise.
3334 (vect_schedule_slp_instance): Likewise.
3335 * tree-vect-stmts.c: Include tree-vector-builder.h.
3336 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3337 (vect_gen_perm_mask_any): Likewise.
3338 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3339 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3341 (build_vector_from_val): Likewise. Explicitly use a duplicate
3344 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3346 * doc/generic.texi (VECTOR_CST): Describe new representation of
3348 * vector-builder.h: New file.
3349 * tree-vector-builder.h: Likewise.
3350 * tree-vector-builder.c: Likewise.
3351 * Makefile.in (OBJS): Add tree-vector-builder.o.
3352 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3353 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3354 (tree_vector): Change the number of elements to
3355 vector_cst_encoded_nelts.
3356 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3357 (VECTOR_CST_ELTS): Delete.
3358 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3359 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3360 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3361 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3362 (VECTOR_CST_ENCODED_ELT): Likewise.
3363 (vector_cst_encoded_nelts): New function.
3364 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3365 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3366 (vector_cst_int_elt, vector_cst_elt): Declare.
3367 * tree.c: Include tree-vector-builder.h.
3368 (tree_code_size): Abort if passed VECTOR_CST.
3369 (tree_size): Update for new VECTOR_CST layout.
3370 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3371 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3372 (build_vector): Use tree_vector_builder.
3373 (vector_cst_int_elt, vector_cst_elt): New functions.
3374 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3375 encoded elements and then create the vector in the canonical form.
3376 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3377 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3378 (tree_c_tests): Call test_vector_cst_patterns.
3379 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3381 (hash_tree): Likewise.
3382 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3383 (streamer_write_tree_header): Likewise.
3384 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3385 (streamer_alloc_tree): Likewise. Update call to make_vector.
3386 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3388 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3390 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3391 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3392 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3394 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3395 Richard Biener <rguenther@suse.de>
3397 PR tree-optimization/81303
3398 * Makefile.in (gimple-loop-interchange.o): New object file.
3399 * common.opt (floop-interchange): Reuse the option from graphite.
3400 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3401 -floop-interchange and mention it for -O3.
3402 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3403 * gimple-loop-interchange.cc: New file.
3404 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3405 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3406 * passes.def (pass_linterchange): New pass.
3407 * timevar.def (TV_LINTERCHANGE): New time var.
3408 * tree-pass.h (make_pass_linterchange): New declaration.
3409 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3410 interchange. Record IV before/after increment in new parameters.
3411 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3412 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3414 (check_reduction_path): ...New function here.
3415 * tree-vectorizer.h (check_reduction_path): New declaration.
3417 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3420 PR rtl-optimization/80818
3421 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3422 always early clobbered.
3423 * lra-lives.c (process_bb_lives): Check input hard regs for early
3424 clobbered non-operand hard reg.
3426 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3429 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3430 types_compatible_p, just that TYPE_MODE is the same.
3432 2017-12-07 Martin Sebor <msebor@redhat.com>
3435 * attribs.c (empty_attribute_table): Initialize new member of
3436 struct attribute_spec.
3437 (decl_attributes): Add argument. Handle mutually exclusive
3438 combinations of attributes.
3439 (selftests::test_attribute_exclusions): New function.
3440 (selftests::attribute_c_tests): Ditto.
3441 * attribs.h (decl_attributes): Add default argument.
3442 * selftest.h (attribute_c_tests): Declare.
3443 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3444 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3446 * doc/extend.texi (Common Function Attributes): Update const and pure.
3447 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3448 of struct attribute_spec.
3449 * config/arc/arc.c (arc_attribute_table): Same.
3450 * config/arm/arm.c (arm_attribute_table): Same.
3451 * config/avr/avr.c ( avr_attribute_table): Same.
3452 * config/bfin/bfin.c (bfin_attribute_table): Same.
3453 * config/cr16/cr16.c (cr16_attribute_table): Same.
3454 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3455 * config/h8300/h8300.c (h8300_attribute_table): Same.
3456 * config/i386/i386.c (ix86_attribute_table): Same.
3457 * config/ia64/ia64.c (ia64_attribute_table): Same.
3458 * config/m32c/m32c.c (m32c_attribute_table): Same.
3459 * config/m32r/m32r.c (m32r_attribute_table): Same.
3460 * config/m68k/m68k.c (m68k_attribute_table): Same.
3461 * config/mcore/mcore.c (mcore_attribute_table): Same.
3462 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3463 * config/mips/mips.c (mips_attribute_table): Same.
3464 * config/msp430/msp430.c (msp430_attribute_table): Same.
3465 * config/nds32/nds32.c (nds32_attribute_table): Same.
3466 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3467 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3468 * config/rl78/rl78.c (rl78__attribute_table): Same.
3469 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3470 * onfig/rx/rx.c (rx_attribute_table): Same.
3471 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3472 * config/sh/sh.c (sh_attribute_table): Same.
3473 * config/sparc/sparc.c (sparc_attribute_table): Same.
3474 * config/spu/spu.c (spu_attribute_table): Same.
3475 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3476 * config/v850/v850.c (v850_attribute_table): Same.
3477 * config/visium/visium.c (visium_attribute_table): Same.
3479 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3482 * config/arm/arm.c (INCLUDE_STRING): Define.
3483 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3484 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3487 2017-12-07 Michael Matz <matz@suse.de>
3489 Add unroll and jam pass
3491 * gimple-loop-jam.c: New file.
3492 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3493 * common.opt (funroll-and-jam): New option.
3494 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3495 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3496 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3497 * passes.def: Add pass_loop_jam.
3498 * timevar.def (TV_LOOP_JAM): Add.
3499 * tree-pass.h (make_pass_loop_jam): Declare.
3500 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3501 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3502 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3503 to flow_loop_tree_node_add.
3504 (duplicate_subloops, copy_loops_to): Append to sibling list.
3505 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3506 * doc/invoke.texi (-funroll-and-jam): Document new option.
3507 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3509 2017-12-07 Richard Biener <rguenther@suse.de>
3511 PR tree-optimization/83296
3512 PR tree-optimization/67769
3513 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3514 flow sensitive info in an unrelated BB.
3515 (value_replacement): Use reset_flow_sensitive_info.
3516 (minmax_replacement): Reset flow sensitive info on the def
3517 we move. Do not reset flow sensitive info in the whole BB
3518 we move the stmt to.
3519 (abs_replacement): Likewise.
3521 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3525 rs6000_cpu to the given -mcpu=, or to the default processor.
3527 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3529 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3530 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3531 instead of rs6000_cpu_attr.
3532 (rs6000_adjust_cost): Ditto.
3533 (is_microcoded_insn): Ditto.
3534 (rs6000_adjust_priority): Ditto.
3535 (rs6000_issue_rate): Ditto.
3536 (rs6000_use_sched_lookahead): Ditto.
3537 (rs6000_use_sched_lookahead_guard): Ditto.
3538 (rs6000_sched_reorder): Ditto.
3539 (force_new_group): Ditto.
3540 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3541 (group_ending_nop): Ditto.
3543 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3545 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3546 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3547 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3548 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3550 (rs6000_reassociation_width): Ditto.
3551 (rs6000_emit_epilogue): Ditto.
3552 (rs6000_adjust_cost): Ditto.
3553 (is_microcoded_insn): Ditto.
3554 (is_cracked_insn): Ditto.
3555 (rs6000_adjust_priority): Ditto.
3556 (rs6000_sched_reorder): Ditto.
3557 (rs6000_sched_reorder2): Ditto.
3558 (insn_must_be_first_in_group): Ditto.
3559 (insn_must_be_last_in_group): Ditto.
3560 (rs6000_register_move_cost): Ditto.
3561 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3564 2017-12-07 Julia Koval <julia.koval@intel.com>
3566 * config.gcc: Add vaesintrin.h.
3567 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3569 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3570 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3572 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3573 * config/i386/immintrin.h: Include vaesintrin.h.
3574 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3575 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3576 _mm_aesdec_epi128): New intrinsics.
3578 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3580 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3581 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3582 c-family/c-spellcheck.cc.
3583 (best_macro_match::best_macro_match): Likewise.
3585 (struct edit_distance_traits<cpp_hashnode *>): Move to
3586 c-family/c-spellcheck.h.
3587 (class best_macro_match): Likewise.
3589 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3591 PR tree-optimization/83293
3592 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3593 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3594 might insert into empty bb.
3597 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3598 simplify for plus with :c added, and pointer_plus without that.
3599 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3600 with undefined overflow and the conversion is not widening,
3601 perform negation in utype and only convert to type afterwards.
3602 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3603 simplify for plus with :c added, and pointer_plus without that.
3604 If type is integral with undefined overflow and the conversion is
3605 not widening, perform minus in utype and only convert to type
3606 afterwards. Move the last pointer_diff_expr simplify into the
3609 2017-12-06 Martin Sebor <msebor@redhat.com>
3611 PR tree-optimization/82646
3612 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3613 strncpy, not maxlen.
3615 2017-12-06 Martin Sebor <msebor@redhat.com>
3617 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3620 PR tree-optimization/83075
3621 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3622 strncat/strncpy don't change length of source string.
3624 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3627 2017-11-29 Martin Aberg <maberg@gaisler.com>
3629 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3630 to prevent b2bst errata sequence.
3631 (sqrtdf2_fix): Likewise.
3633 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3635 PR tree-optimization/81945
3636 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3637 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3638 to dest_cfun has orig_loop_num set, either remap it to the new
3639 loop number if the loop got moved too, or clear it.
3641 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3643 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3645 (thunderx2t99_nothing): New insn reservation.
3646 (thunderx2t99_mrs): New insn reservation.
3647 (thunderx2t99_multiple): New insn reservation.
3648 (thunderx2t99_alu_basi): Add bfx to reservation.
3649 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3651 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3654 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3656 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3658 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3659 tree-ssa-pre.c::remove_dead_inserted_code.
3660 * tree-ssa-dce.h: New file.
3661 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3662 (remove_dead_inserted_code): Move and rename to function
3663 tree-ssa-dce.c::simple_dce_from_worklist.
3664 (pass_pre::execute): Update use.
3666 2017-12-05 Richard Biener <rguenther@suse.de>
3668 PR tree-optimization/83277
3669 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3670 to code-gen liveout vars.
3672 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3674 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3675 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3676 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3679 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3681 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3683 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3684 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3687 2017-12-05 Richard Biener <rguenther@suse.de>
3689 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3690 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3691 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3692 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3694 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3696 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3698 2017-12-05 Martin Liska <mliska@suse.cz>
3699 Jakub Jelinek <jakub@redhat.com>
3701 * doc/invoke.texi: Document the options.
3702 * flag-types.h (enum sanitize_code): Add
3703 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3704 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3705 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3706 * opts.c: Define new sanitizer options.
3707 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3708 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3710 2017-12-05 Julia Koval <julia.koval@intel.com>
3712 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3713 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3714 (ix86_handle_option): Handle -mavx512vnni.
3715 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3716 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3717 * config/i386/i386-c (__AVX512VNNI__): New.
3718 * config/i386/i386.c (ix86_target_string): Handle new option.
3719 (ix86_valid_target_attribute_inner_p): Handle new option.
3720 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3721 * config/i386/i386.opt (mavx512vnni): New option.
3723 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3726 * athlon.md: Disable for generic.
3727 * haswell.md: Enable for generic.
3728 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3729 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3731 (ix86_adjust_cost): Move generic to haswell path.
3733 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3735 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3736 instead of MEM_P in a couple more places. Fix formatting issues.
3738 2017-12-04 Jim Wilson <jimw@sifive.com>
3740 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3741 instead of GP_REG_LAST-1.
3742 (riscv_adjust_libcall_cfi_prologue): Likewise.
3743 (riscv_adjust_libcall_cri_epilogue): Likewise.
3744 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3747 2017-12-04 Luis Machado <luis.machado@linaro.org>
3749 * ipa-pure-const.c (check_decl): Add missing newline.
3750 (state_from_flags): Likewise.
3752 2017-12-04 Jeff Law <law@redhat.com>
3754 PR tree-optimizatin/78496
3755 * gimple-ssa-evrp-analyze.h
3756 (evrp_range_analyzer::get_vr_values): Simplify.
3757 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3758 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3759 and gimple-ssa-evrp-analyze.h.
3760 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3761 (simplify_stmt_for_jump_threading): Copy a blob of code from
3762 tree-vrp.c to use ranges to simplify statements.
3763 (dom_opt_dom_walker::before_dom_children): Call
3764 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3765 (dom_opt_dom_walker::after_dom_children): Similarly for
3766 evrp_range_analyzer::leave.
3767 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3770 * gimple-ssa-evrp-analyze.c
3771 (evrp_range_analyzer::extract_range_from_stmt): Always use
3772 vr_values::update_value_range so preexisting range info is
3773 medged with new range info, even if the new range is VR_VARYING.
3775 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3777 * combine.c: Adjust comment.
3778 (use_crosses_set_p): Delete.
3779 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3780 (try_combine): Ditto.
3782 2017-12-04 Richard Biener <rguenther@suse.de>
3784 PR tree-optimization/83255
3785 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3786 Re-add zero-iteration check.
3788 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3790 PR rtl-optimization/83245
3791 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3792 hard registers as earlyclobber, also if not in an asm.
3794 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3798 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3802 rs6000_cpu based on cpu_index, not tune_index.
3804 2017-12-04 Richard Biener <rguenther@suse.de>
3806 PR tree-optimization/83238
3807 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3808 code match comment, rejecting invalid SESE regions.
3810 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3812 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3813 require base operand is a REG_POINTER prior to reload on targets
3814 with non-equivalent space registers.
3816 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3818 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3819 (update_specialized_profile): Fix updating of counts.
3820 (perhaps_add_new_callers): Likewise.
3822 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3825 * x86-tune.def: Remove obsolette FIXMEs.
3826 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3827 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3828 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3830 (X86_TUNE_PAD_RETURNS): Disable for generic.
3832 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3834 PR tree-optimization/83170
3835 PR tree-optimization/83241
3836 * gimple-ssa-store-merging.c
3837 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3838 gimple_vuse (ins_stmt) in case it has changed.
3839 (imm_store_chain_info::output_merged_store): Likewise.
3841 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3845 * tree-cfg.c (pass_warn_function_return::execute): Handle
3846 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3847 with BUILTINS_LOCATION.
3851 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3852 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3853 instead of DECL_MODE.
3855 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3856 last argument to rtx pointer.
3857 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3858 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3859 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3860 depending on the chosen ISAs.
3861 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3862 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3863 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3865 * config/i386/sse.md (mov<mode>_internal): Likewise.
3866 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3868 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3870 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3872 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3874 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3875 parameters from prototype.
3876 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3877 parameters. Don't print a comment.
3878 (emit_fusion_gpr_load): Adjust.
3879 (emit_fusion_load_store): Adjust.
3880 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3881 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3882 comment on the second line.
3884 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3888 rs6000_cpu based on cpu_index, not tune_index.
3890 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3892 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3893 which_alternative instead of which_alternative + 1.
3894 (output_asm_insn): Print an extra tab if the template is short.
3896 2017-12-01 Jim Wilson <jimw@sifive.com>
3898 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3900 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3901 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3903 * doc/tm.texi: Regenerate.
3905 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3908 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3909 whether we can allocate pseudos before trying to fix an address.
3910 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3911 memory address is indexed or indirect.
3912 (floatuns_<mode>si2_hw2): Likewise.
3914 2017-12-01 Jason Merrill <jason@redhat.com>
3916 * Makefile.in (TAGS): Add c-family/*.cc.
3918 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3920 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3921 (call_value_insn): Likewise.
3922 (sibcall_insn): Likewise.
3923 (sibcall_value_insn): Likewise.
3924 (movsi_aarch64): Likewise.
3925 (movdi_aarch64): Likewise.
3926 (add_losym_): Likewise.
3927 (ldr_got_small_): Likewise.
3928 (ldr_got_small_sidi): Likewise.
3929 (ldr_got_small_28k_): Likewise.
3930 (ldr_got_small_28k_sidi): Likewise.
3931 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3932 Move output_addr_const to symbolic case. Add error check.
3934 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3936 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3937 * config/s390/s390.md: Disable MVC merging peephole if it would
3938 disable operand forwarding.
3939 (new peephole2): Split MVCs if it would turn them into up to 2
3942 2017-12-01 Richard Biener <rguenther@suse.de>
3944 PR tree-optimization/83232
3945 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3946 detection of same access. Instead of breaking the group here
3947 do not consider the duplicate. Add comment explaining real fix.
3949 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3951 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3953 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3955 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3957 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3959 * function.h (struct function): Remove cilk_frame_decl,
3960 is_cilk_function and calls_cilk_spawn fields.
3961 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3963 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3964 cilk_elemental field.
3965 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3966 * target.def: Adjust comment.
3967 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3968 Don't test cilk_elemental.
3970 PR tree-optimization/83233
3971 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3972 bswap_stat name for the struct.
3975 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3976 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3977 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3978 SWITCH_BREAK_LABEL_P set on the label.
3979 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3980 added for default case if it was missing and not all cases covered.
3981 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3982 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3984 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3985 to avoid -Wimplicit-fallthrough warning.
3986 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3987 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3989 PR tree-optimization/83221
3990 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3992 (init_reassoc): Formatting fix.
3995 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3996 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4000 * tree-cfg.c: Include asan.h.
4001 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4003 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4005 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4007 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4009 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4010 vpcmpeqd instruction.
4012 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4014 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4015 registers type for 128bit mode.
4017 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4019 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4021 * gengtype-state.c (read_state_pair): Likewise.
4022 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4023 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4024 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4025 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4026 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4027 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4028 * builtins.c (fold_builtin_3): Likewise.
4029 * graphite-scop-detection.c
4030 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4031 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4034 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4037 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4038 parameter. Return true if ignore_decl_rtl_set_p is true and other
4039 conditions are satisfied.
4040 * asan.h (asan_protect_global): Add new parameter.
4041 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4042 to asan_protect_global calls.
4044 2017-11-30 Jim Wilson <jimw@sifive.com>
4046 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4047 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4048 -mstrict-align, add info on default value. Delete redundant lines for
4049 -mabi. Add missing -mexplicit-relocs docs.
4051 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4053 * config/arc/arc.md (trap): New pattern.
4055 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4057 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4058 instruction to end into a delay slot.
4059 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4060 can be placed into a delay slot against reg_note.
4062 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4064 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4065 labels number of usages.
4067 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4069 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4071 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4073 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4076 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4077 multiplication by power of 2 constant into two shifts + comparison.
4079 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4082 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4084 2017-11-30 Richard Biener <rguenther@suse.de>
4086 PR tree-optimization/83202
4087 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4088 allow_peel argument and guard peeling.
4089 (canonicalize_loop_induction_variables): Likewise.
4090 (canonicalize_induction_variables): Pass false.
4091 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4092 peeling from cunrolli.
4094 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4096 * combine.c (try_combine): Print a message to dump file whenever
4097 I0, I1, or I2 cannot be combined into I3.
4099 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4101 PR rtl-optimization/83156
4102 PR rtl-optimization/82621
4103 * combine.c (try_combine): Don't split an I2 if one of the dests is
4104 set again before I3. Allow unused dests.
4106 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4108 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4110 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4112 PR rtl-optimization/80818
4113 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4114 recursively. Use insn code for clobber.
4115 (lra_set_insn_recog_data): Pass the new arg to
4116 collect_non_operand_hard_regs.
4117 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4119 (lra_update_insn_regno_info): Pass insn to
4120 add_regs_to_insn_regno_info.
4122 2017-11-29 Jim Wilson <jimw@sifive.com>
4123 Andrew Waterman <andrew@sifive.com>
4125 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4126 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4127 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4128 use SINGLE_SHIFT_COST.
4129 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4130 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4132 2017-11-29 Julia Koval <julia.koval@intel.com>
4134 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4135 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4136 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4137 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4138 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4139 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4140 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4141 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4142 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4143 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4144 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4145 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4146 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4147 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4148 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4149 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4150 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4151 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4152 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4153 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4154 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4155 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4156 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4157 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4158 __builtin_ia32_vpshldv_v2di_mask,
4159 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4160 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4161 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4163 2017-11-29 Julia Koval <julia.koval@intel.com>
4165 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4166 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4167 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4168 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4169 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4170 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4171 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4172 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4173 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4174 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4175 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4176 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4177 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4178 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4179 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4180 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4181 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4182 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4183 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4184 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4185 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4187 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4189 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4190 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4193 2017-11-29 Martin Aberg <maberg@gaisler.com>
4195 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4196 to prevent b2bst errata sequence.
4197 (sqrtdf2_fix): Likewise.
4199 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4201 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4202 (div_sqrt_insn_p): New function.
4203 (sparc_do_work_around_errata): Insert NOP instructions to
4204 prevent sequences that could trigger the TN-0013 errata for
4205 certain LEON3 processors.
4206 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4207 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4208 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4209 (in_branch_delay): Prevent div and sqrt in delay slot if
4211 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4213 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4215 * config/sparc/sparc.c (atomic_insn_p): New function.
4216 (sparc_do_work_around_errata): Insert NOP instructions to
4217 prevent sequences that could trigger the TN-0010 errata for
4219 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4220 instruction referable in atomic_insns_p.
4222 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4224 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4225 (atomic_compare_and_swap_leon3_1): Likewise.
4228 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4230 * config/sparc/sparc.c (fpop_insn_p): New function.
4231 (sparc_do_work_around_errata): Insert NOP instructions to
4232 prevent sequences that could trigger the TN-0012 errata for
4234 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4235 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4236 (in_branch_annul_delay): Prevent floating-point instructions
4237 in delay slot of annulled integer branch.
4239 2017-11-29 Richard Biener <rguenther@suse.de>
4241 PR tree-optimization/83202
4242 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4244 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4245 nodes vectorized to the same stmts multiple times.
4246 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4247 (vect_analyze_slp_instance): Adjust.
4248 (scalar_stmts_to_slp_tree_map_t): New typedef.
4249 (vect_schedule_slp_instance): Add a map recording the SLP node
4250 representing the vectorized stmts for a set of scalar stmts.
4251 Avoid code-generating redundancies.
4252 (vect_schedule_slp): Allocate map and pass it down.
4254 2017-11-29 Nathan Sidwell <nathan@acm.org>
4257 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4260 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4262 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4263 cases for printing LDP/STP memory addresses.
4264 (aarch64_print_address_internal): Renamed from
4265 aarch64_print_operand_address, added parameter, add Pmode check.
4266 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4267 (aarch64_print_operand_address): Indirect to
4268 aarch64_print_address_internal.
4269 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4270 'y' operand output specifier.
4272 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4275 * tree.c (build_simple_mem_ref_loc): Handle
4276 get_addr_base_and_unit_offset returning a MEM_REF.
4279 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4282 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4285 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4287 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4289 (layout::move_to_column): Likewise.
4291 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4294 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4295 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4297 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4298 Martin Jambor <mjambor@suse.cz>
4301 * tree.h (expr_type_first_operand_type_p): Declare
4302 * tree.c (expr_type_first_operand_type_p): New function.
4303 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4304 (ipa_value_from_jfunc): Adjust declaration.
4305 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4306 Use it as result type for arithmetics, unless it is NULL in which case
4307 be more conservative.
4308 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4309 ipa_get_jf_pass_through_result.
4310 (propagate_vals_across_pass_through): Likewise.
4311 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4312 is to propagate_vals_across_pass_through.
4313 (propagate_constants_across_call): Pass PARM_TYPE to
4314 propagate_scalar_across_jump_function.
4315 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4316 ipa_value_from_jfunc.
4317 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4318 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4319 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4320 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4321 target_type, pass it to ipa_value_from_jfunc.
4322 (update_indirect_edges_after_inlining): Pass parameter type to
4323 try_make_edge_direct_simple_call.
4325 2017-11-28 Jeff Law <law@redhat.com>
4327 * gimple-ssa-evrp-analyze.c
4328 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4329 refine ranges if scev_initialized_p returns true.
4330 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4332 2017-11-28 Julia Koval <julia.koval@intel.com>
4334 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4335 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4336 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4337 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4338 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4339 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4340 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4341 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4342 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4343 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4344 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4345 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4346 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4347 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4348 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4349 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4350 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4351 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4352 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4353 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4354 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4355 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4357 2017-11-28 Julia Koval <julia.koval@intel.com>
4359 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4360 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4361 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4362 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4363 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4364 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4365 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4366 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4367 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4368 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4369 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4370 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4371 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4372 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4373 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4374 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4375 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4376 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4377 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4378 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4379 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4380 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4381 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4382 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4383 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4384 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4385 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4386 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4387 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4388 __builtin_ia32_vpshld_v2di_mask): New builtins.
4389 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4390 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4392 2017-11-28 Richard Biener <rguenther@suse.de>
4394 PR tree-optimization/80776
4395 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4397 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4399 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4400 If the incoming edge is an effective fallthru because the other
4401 edge only reaches a __builtin_unreachable () then record ranges
4402 derived from the controlling condition in SSA info.
4403 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4404 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4406 2017-11-28 Olivier Hainque <hainque@adacore.com>
4408 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4409 on mingw build hosts.
4411 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4414 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4416 2017-11-28 Richard Biener <rguenther@suse.de>
4419 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4420 copies generated from memcpy use a character array as reference
4423 2017-11-28 Julia Koval <julia.koval@intel.com>
4424 Sebastian Peryt <sebastian.peryt@intel.com>
4426 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4427 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4428 cilk-common.o, cilk.h, cilk-common.c): Remove.
4430 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4431 * builtins.c (is_builtin_name): Remove cilkplus condition.
4432 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4433 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4434 cilk-builtins.def, cilkplus.def): Remove.
4435 * cif-code.def (CILK_SPAWN): Remove.
4436 * cilk-builtins.def: Delete.
4437 * cilk-common.c: Ditto.
4439 * cilkplus.def: Ditto.
4440 * config/darwin.h (fcilkplus): Delete.
4441 * cppbuiltin.c: Ditto.
4442 * doc/extend.texi: Remove cilkplus doc.
4443 * doc/generic.texi: Ditto.
4444 * doc/invoke.texi: Ditto.
4445 * doc/passes.texi: Ditto.
4446 * gcc.c (fcilkplus): Remove.
4447 * gengtype.c (cilk.h): Remove.
4448 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4450 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4452 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4454 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4455 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4456 cilkplus conditions.
4457 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4458 inline_read_section): Ditto.
4459 * ipa-inline-analysis.c (cilk.h): Remove.
4460 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4461 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4462 append_linker_options): Remove condition for fcilkplus.
4463 * lto/lto-lang.c (cilk.h): Remove.
4464 (lto_init): Remove condition for fcilkplus.
4465 * omp-expand.c (expand_cilk_for_call): Delete.
4466 (expand_omp_taskreg, expand_omp_for_static_chunk,
4467 expand_omp_for): Remove cilkplus
4469 (expand_cilk_for): Delete.
4470 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4471 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4472 execute_lower_omp, diagnose_sb_0): Ditto.
4473 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4474 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4475 * tree-nested.c: Ditto.
4476 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4477 (dump_generic_node): Ditto.
4478 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4479 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4481 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4483 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4485 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4486 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4487 for vector mode and !TARGET_SIMD.
4489 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4491 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4493 * tree.h (SWITCH_LABELS): Remove.
4494 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4495 assert SWITCH_BODY is non-NULL.
4496 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4498 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4500 PR tree-optimization/80788
4501 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4502 has TREE_OVERFLOW set, call drop_tree_overflow.
4504 2017-11-28 Richard Biener <rguenther@suse.de>
4506 PR tree-optimization/83158
4507 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4509 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4512 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4513 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4515 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4517 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4518 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4520 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4523 * varasm.c (compare_constant): Take the mode of the constants into
4524 account when comparing floating point constants.
4526 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4528 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4529 from explicit instantiation of debug_helper.
4530 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4532 2017-11-27 Richard Biener <rguenther@suse.de>
4534 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4537 2017-11-27 Richard Biener <rguenther@suse.de>
4539 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4540 (type_hash_canon): Also clear int_cst_hash_table entry for
4542 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4544 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4546 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4547 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4548 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4550 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4552 * hash-map.h (gt_cleare_cache): Avoid UB.
4554 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4556 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4557 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4558 (decide_unroll_runtime_iterations): Likewise.
4559 (decide_unroll_stupid): Likewise.
4561 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4564 * config/i386/i386.md: Add a loop with incssp.
4566 2017-11-27 Martin Jambor <mjambor@suse.cz>
4568 PR tree-optimization/81248
4569 * tree-sra.c (splice_param_accesses): Remove size check.
4570 (decide_one_param_reduction): Fix size check.
4571 * gimple-pretty-print.c (dump_profile): Silence warning.
4572 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4574 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4577 * dbxout.c (lastlineno): New variable.
4578 (dbx_debug_hooks): Use dbxout_switch_text_section as
4579 switch_text_section debug hook.
4580 (dbxout_function_end): Switch to current_function_section
4581 rather than function_section. If crtl->has_bb_partition,
4582 output just one N_FUN, depending on in_cold_section_p.
4583 (dbxout_source_line): Remember last lineno in lastlineno.
4584 (dbxout_switch_text_section): New function.
4585 (dbxout_function_decl): Adjust dbxout_block caller.
4586 (dbx_block_with_cold_children): New function.
4587 (dbxout_block): Return true if any LBRAC/RBRAC have been
4588 emitted. Use dbx_block_with_cold_children at depth == 0
4589 in second partition. Add PARENT_BLOCKNUM argument, pass
4590 it optionally adjusted to children. Output LBRAC/RBRAC
4591 around recursive call only if the block is in the current
4592 partition, if not and anything was output, emit empty
4594 * final.c (final_scan_insn): Compute cold_function_name
4595 before calling switch_text_section debug hook. Call
4596 that hook even if dwarf2out_do_frame if not emitting
4600 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4601 TREE_READONLY decls.
4603 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4605 PR rtl-optimization/82488
4606 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4608 2017-11-26 Julia Koval <julia.koval@intel.com>
4610 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4612 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4615 2017-11-26 Julia Koval <julia.koval@intel.com>
4617 * config/i386/driver-i386.c (host_detect_local_cpu):
4618 Detect skylake-avx512.
4620 2017-11-26 Julia Koval <julia.koval@intel.com>
4622 * config.gcc: Add -march=cannonlake.
4623 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4624 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4625 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4626 (PTA_CANNONLAKE): New.
4627 (processor_target_table): Add cannonlake.
4628 (ix86_option_override_internal): Ditto.
4629 (fold_builtin_cpu): Ditto.
4630 (get_builtin_code_for_version): Handle cannonlake.
4631 (M_INTEL_COREI7_CANNONLAKE): New.
4632 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4633 * doc/invoke.texi: Add -march=cannonlake.
4635 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4637 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4638 (try_init_one_plugin): Alternative implementation for MinGW.
4639 * Makefile.in (plugin_implib): New.
4640 (gengtype-lex.c): Fix broken AIX workaround.
4641 * configure: Regenerate.
4642 * doc/plugins.texi: Document support for MinGW.
4644 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4646 PR rtl-optimization/81553
4647 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4648 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4649 is a shift where C1 has different mode than the whole shift, use C1's
4650 mode for MULT rather than the shift's mode.
4653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4654 builtins not enabled in the currently selected ISA.
4656 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4658 PR tree-optimization/71026
4659 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4660 (insert_reciprocals): Change to insert reciprocals before a division
4661 by a square and to insert the square of a reciprocal.
4662 (execute_cse_reciprocals_1): Change to consider division by a square.
4663 (register_division_in): Add importance parameter.
4665 2017-11-24 Richard Biener <rguenther@suse.de>
4667 PR tree-optimization/82402
4668 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4669 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4671 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4673 * match.pd (0-ptr): New transformation.
4675 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4678 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4679 when skipping checking.
4681 2017-11-24 Richard Biener <rguenther@suse.de>
4683 PR tree-optimization/83128
4684 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4685 (vn_reference_lookup_3): Likewise.
4687 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4690 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4691 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4694 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4695 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4696 check for the latter with gimple_call_builtin_p. Do not handle
4697 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4699 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4701 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4704 2017-11-23 Julia Koval <julia.koval@intel.com>
4706 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4707 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4708 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4709 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4710 _mm512_maskz_expandloadu_epi16): New intrinsics.
4711 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4712 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4713 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4714 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4715 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4716 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4717 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4718 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4719 _mm256_maskz_expandloadu_epi8): New intrinsics.
4720 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4721 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4722 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4723 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4724 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4725 * config/i386/sse.md (VI248_VLBW): New iterator.
4726 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4728 2017-11-23 Julia Koval <julia.koval@intel.com>
4730 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4731 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4732 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4733 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4734 _mm512_mask_compressstoreu_epi16): New.
4735 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4736 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4737 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4738 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4739 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4740 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4741 _mm256_mask_compressstoreu_epi8): New.
4742 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4743 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4744 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4745 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4746 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4747 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4748 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4749 __builtin_ia32_compresshi128_mask,
4750 __builtin_ia32_compressstoreuqi512_mask,
4751 __builtin_ia32_compressstoreuhi512_mask,
4752 __builtin_ia32_compressstoreuqi256_mask,
4753 __builtin_ia32_compressstoreuqi128_mask,
4754 __builtin_ia32_compressstoreuhi256_mask,
4755 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4756 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4758 (ix86_expand_special_args_builtin): Handle new types.
4759 (s4fma_expand): Handle new builtin array.
4760 * config/i386/immintrin.h: Include new headers.
4761 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4762 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4764 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4767 fixes. Declare temp and g variables at the top in order to avoid
4768 {} in most of the cases.
4770 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4772 * match.pd (ptr-0): New transformation.
4774 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4776 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4777 (struct addr_mode_cost_table): New.
4778 (struct tune_params): Add field addr_mode_costs.
4779 * config/arm/arm.c (generic_addr_mode_costs): New.
4780 (arm_slowmul_tune): Initialise addr_mode_costs field.
4781 (arm_fastmul_tune): Likewise.
4782 (arm_strongarm_tune): Likewise.
4783 (arm_xscale_tune): Likewise.
4784 (arm_9e_tune): Likewise.
4785 (arm_marvell_pj4_tune): Likewise.
4786 (arm_v6t2_tune): Likewise.
4787 (arm_cortex_tune): Likewise.
4788 (arm_cortex_a8_tune): Likewise.
4789 (arm_cortex_a7_tune): Likewise.
4790 (arm_cortex_a15_tune): Likewise.
4791 (arm_cortex_a35_tune): Likewise.
4792 (arm_cortex_a53_tune): Likewise.
4793 (arm_cortex_a57_tune): Likewise.
4794 (arm_exynosm1_tune): Likewise.
4795 (arm_xgene1_tune): Likewise.
4796 (arm_cortex_a5_tune): Likewise.
4797 (arm_cortex_a9_tune): Likewise.
4798 (arm_cortex_a12_tune): Likewise.
4799 (arm_cortex_a73_tune): Likewise.
4800 (arm_v7m_tune): Likewise.
4801 (arm_cortex_m7_tune): Likewise.
4802 (arm_v6m_tune): Likewise.
4803 (arm_fa726te_tune): Likewise.
4804 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4806 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4808 * config/arm/arm.c (arm_mem_costs): New function.
4809 (arm_rtx_costs_internal): Use arm_mem_costs.
4811 2017-11-23 Mark Wielaard <mark@klomp.org>
4813 * dwarf2out.c (init_sections_and_labels): Use generation to create
4814 unique ranges_section_label and ranges_base_label. Return generation.
4815 (output_rnglists): Add generation argument. Use generation to create
4816 unique ranges labels.
4817 (dwarf2out_finish): Get generation from init_sections_and_labels
4818 and pass generation to output_rnglists.
4820 2017-11-23 Mike Stump <mikestump@comcast.net>
4821 Eric Botcazou <ebotcazou@adacore.com>
4823 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4824 * cfgloop.h (struct loop): Add unroll field.
4825 * function.h (struct function): Add has_unroll bitfield.
4826 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4827 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4828 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4829 (pass_rtl_unroll_loops::gate): Likewise.
4830 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4831 for which loop->unroll==1.
4832 (decide_unroll_constant_iterations): Use note for consistency and
4833 take loop->unroll into account. Return early if loop->unroll is set.
4834 Fix thinko in existing test.
4835 (decide_unroll_runtime_iterations): Use note for consistency and
4836 take loop->unroll into account.
4837 (decide_unroll_stupid): Likewise.
4838 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4839 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4840 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4842 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4843 (print_loop): Print loop->unroll if set.
4844 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4845 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4846 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4848 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4849 loop->unroll is set and smaller than the trip count. Otherwise bypass
4850 entirely the heuristics if loop->unroll is set. Remove dead note.
4851 Fix off-by-one bug in other note.
4852 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4853 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4855 (tree_unroll_loops_completely): Make static.
4856 (pass_complete_unroll::execute): Use correct type for variable.
4857 (pass_complete_unrolli::execute): Fix formatting.
4858 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4860 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4862 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4863 enable when TARGET_PREFER_AVX128 is set.
4865 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4867 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4869 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4870 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4872 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4874 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4875 frequencies. Fix estimation of aggregate parameters.
4877 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4879 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4882 2017-11-23 Tom de Vries <tom@codesourcery.com>
4884 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4886 2017-11-23 Tom de Vries <tom@codesourcery.com>
4888 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4891 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4894 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4895 sibcall_value_pcrel_fdpic): Use local variable instead of
4897 (calli_tbr_rel): Add missing operand 2.
4898 (call_valuei_tbr_rel): Add missing operand 3.
4900 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4903 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4904 bitpos/bitsize covering the whole destination, use store_expr only if
4905 the complex mode is the same. Otherwise, use expand_normal and if
4906 it returns CONCAT, subreg each part separately instead of trying to
4907 subreg the whole result.
4909 2017-11-23 Richard Biener <rguenther@suse.de>
4911 PR tree-optimization/23094
4912 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4913 come here from walking over backedges in the first iteration.
4914 (vn_reference_lookup_3): Skip clobbers that store the same value.
4916 2017-11-23 Richard Biener <rguenther@suse.de>
4918 PR tree-optimization/81403
4919 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4920 a block we need a leader relative to.
4921 (phi_translate_1): For nary processing require a leader from
4922 get_representative_for given we run expression simplification
4923 using match-and-simplify. Remove previous fix.
4925 2017-11-22 Jeff Law <law@redhat.com>
4927 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4928 Use new method allocate_value_range rather than accessing the
4929 vrp_value_range_pool data member directly.
4930 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4931 to use extract_range_from_stmt method to avoid need for
4932 extract_range_from_assignment method.
4933 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4934 method rather than setting values_propgated data member directly.
4935 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4936 and values propagated data members and extract_range_from_assignment
4937 method. Reorder private data members to conform to standards.
4938 Add new methods set_lattice_propagation_complete and
4939 allocate_value_range.
4941 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4943 PR rtl-optimization/83030
4944 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4945 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4946 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4947 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4948 CROSSING_JUMP_P flag is already set before setting it.
4949 * cfgrtl.c (fixup_partition_crossing): Likewise.
4950 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4953 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4955 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4956 Handle the case where both arguments are using gen_const_vec_series.
4958 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4961 * pretty-print.c (pp_format): Move quoting implementation to
4962 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4963 to pass address of "quote" local.
4964 (pp_begin_quote): New function.
4965 (pp_end_quote): New function.
4966 * pretty-print.h (printer_fn): Convert penultimate param from bool
4968 (pp_begin_quote): New decl.
4969 (pp_end_quote): New decl.
4970 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4971 param from bool to bool *.
4972 * tree-diagnostic.h (default_tree_printer): Likewise.
4974 2017-11-22 Jeff Law <law@redhat.com>
4976 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4977 Initialize vr_values.
4978 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4979 vr_values via vr_values class instance rather than delegators.
4980 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4981 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4982 (evrp_range_analyzer::push_value_range): Likewise.
4983 (evrp_range_analyzer::pop_value_range): Likewise.
4984 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4985 most delegators. Those remaining are exposed as public interfaces.
4986 Make vr_values a pointer and private.
4987 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4989 (evrp_range_analyzer::get_vr_value): New method.
4990 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4991 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4992 (class evrp_dom_walker): Attach evrp_folder class, initialize
4993 it in the ctor. Remove temporary delegators.
4994 (evrp_dom_walker::before_dom_children): Call methods in attached
4995 evrp_range_analyzer class via class instance pointer. Use
4996 free value_range_constant_singleton to remove need for
4997 op_with_constant_singleton_value delegator method. Do not
4998 create a vrp_prop class instance for every call! Narrow
4999 scope of a couple variables.
5000 (evrp_dom_walker::cleanup): Call methods in attached
5001 evrp_range_analyzer class via class instance pointer.
5002 * vr-values.h (class vr_values): Privatize many methods and
5005 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5007 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5008 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5009 * expr.c (expand_expr_real_2): Likewise.
5010 * fold-const.c (const_unop): Likewise.
5011 * optabs-tree.c (optab_for_tree_code): Likewise.
5012 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5013 * tree-inline.c (estimate_operator_cost): Likewise.
5014 * tree-pretty-print.c (dump_generic_node): Likewise.
5015 (op_code_prio): Likewise.
5016 (op_symbol_code): Likewise.
5017 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5018 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5019 * internal-fn.c (direct_internal_fn_optab): New function.
5020 (direct_internal_fn_array, direct_internal_fn_supported_p
5021 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5022 * fold-const-call.c (fold_const_reduction): New function.
5023 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5025 * tree-vect-loop.c: Include internal-fn.h.
5026 (reduction_code_for_scalar_code): Rename to...
5027 (reduction_fn_for_scalar_code): ...this and return an internal
5029 (vect_model_reduction_cost): Take an internal_fn rather than
5031 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5033 (vectorizable_reduction): Use internal functions rather than tree
5034 codes for the reduction operation. Update calls to the functions
5036 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5037 Use calls to internal functions rather than REDUC tree codes.
5038 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5040 2017-11-22 Olivier Hainque <hainque@adacore.com>
5042 * config/vxworks.c (vxworks_override_options): Pick default
5043 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5044 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5045 DWARF_GNAT_ENCODINGS_DEFAULT.
5046 * config/vxworksae.h: Likewise.
5048 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5050 PR tree-optimization/83104
5051 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5054 2017-11-22 Marek Polacek <polacek@redhat.com>
5055 H.J. Lu <hongjiu.lu@intel.com>
5056 Jason Merrill <jason@redhat.com>
5061 * calls.c (initialize_argument_information): Call
5062 warn_parameter_passing_abi target hook.
5063 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5064 argument onto stack.
5065 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5066 * common.opt: Update -fabi-version description.
5067 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5068 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5070 (ix86_is_empty_record): New function.
5071 (ix86_warn_parameter_passing_abi): New function.
5072 (TARGET_EMPTY_RECORD_P): Redefine.
5073 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5074 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5075 * doc/tm.texi: Regenerated.
5076 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5077 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5078 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5079 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5080 instead of int_size_in_bytes.
5081 * expr.c (copy_blkmode_to_reg): Likewise.
5082 * function.c (aggregate_value_p): Return 0 for empty types.
5083 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5085 (locate_and_pad_parm): Call arg size_in_bytes instead
5087 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5088 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5089 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5091 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5092 (std_gimplify_va_arg_expr): Skip empty records. Call
5093 arg_size_in_bytes instead size_in_bytes.
5094 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5095 * tree-core.h (tree_type_common): Add empty_flag.
5096 (tree_decl_common): Update comments.
5097 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5099 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5100 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5102 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5103 * tree.c (default_is_empty_type): New function.
5104 (default_is_empty_record): New function.
5105 (arg_int_size_in_bytes): New function.
5106 (arg_size_in_bytes): New function.
5107 (get_ultimate_context): New function.
5108 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5109 TRANSLATION_UNIT_WARN_EMPTY_P.
5110 (default_is_empty_record, arg_int_size_in_bytes,
5111 arg_size_in_bytes, get_ultimate_context): Declare.
5113 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5115 * config/arm/arm.c (cmse_clear_registers): New function.
5116 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5117 code by call to cmse_clear_registers.
5118 (cmse_nonsecure_entry_clear_before_return): Likewise.
5120 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5122 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5123 (vdot_s32, vdotq_s32): New.
5124 (vdot_lane_u32, vdotq_lane_u32): New.
5125 (vdot_lane_s32, vdotq_lane_s32): New.
5128 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5131 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5132 for unsigned values with fewer HWIs than the precision.
5133 (test_overflow): New function.
5134 (wide_int_cc_tests): Call it.
5136 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5138 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5139 has the same mode class as Pmode.
5141 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5143 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5144 padding_bits_to_clear_ptr.
5145 (cmse_nonsecure_entry_clear_before_return): Likewise.
5147 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5149 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5150 auto_sbitap instead of integer bitfield to control register needing
5153 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5155 PR tree-optimization/83044
5156 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5157 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5158 that need the upper bound. Subtract offset from
5159 get_addr_base_and_unit_offset only if positive and subtract it
5160 before division by eltsize rather than after it.
5163 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5164 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5165 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5169 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5170 expand_binop with *mul_widen_optab, make sure at least one of the
5171 operands doesn't have VOIDmode.
5174 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5176 PR rtl-optimization/82044
5177 PR tree-optimization/82042
5178 * dse.c (record_store): Check for overflow.
5179 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5180 add_wild_read instead of clear_rhs_from_active_local_stores on
5181 overflow and log it into dump_file.
5183 2017-11-22 Richard Biener <rguenther@suse.de>
5185 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5186 fake edges to exit when looking for a place to insert.
5187 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5189 (insert_into_preds_of_block): Commit edge insertion immediately,
5190 assert that doesn't require new BBs.
5191 (fini_pre): Release expressions.
5192 (pass_pre::execute): Shuffle things around a bit, if the fn
5193 is too large do not compute AVAIL either as this is really the
5196 2017-11-22 Richard Biener <rguenther@suse.de>
5198 PR tree-optimization/83089
5199 * tree-if-conv.c (pass_if_conversion::execute): If anything
5200 changed reset SCEV and free the number of iteration estimates.
5202 2017-11-21 Martin Sebor <msebor@redhat.com>
5204 PR tree-optimization/82945
5205 * calls.h (warn_nonstring_bound): Remove unused function.
5207 2017-11-21 Martin Sebor <msebor@redhat.com>
5209 PR tree-optimization/82945
5210 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5211 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5212 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5214 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5215 * calls.h (get_attr_nonstring_decl): Declare new function.
5216 * doc/extend.texi (attribute nonstring): Update.
5217 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5218 get_attr_nonstring_decl and handle it.
5219 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5220 detection of nul-termination.
5221 (strlen_to_stridx): Change to a pointer.
5222 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5223 (pass_strlen::execute): Same.
5225 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5227 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5228 for the new option -mprefer-vector-width=[none|128|256|512].
5229 * config/i386/i386.c (ix86_target_string): remove old style options
5230 -mprefer-avx256 and make -mprefer-avx128 as alias.
5231 (ix86_option_override_internal): Apply defaults for the
5232 -mprefer-vector-width=[128|256] option.
5233 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5234 Implement macros to work with -mprefer-vector-width=.
5235 * config/i386/i386.opt: Implemented option
5236 -mprefer-vector-width=[none|128|256|512].
5237 * doc/invoke.texi: Documentation for
5238 -mprefer-vector-width=[none|128|256|512].
5240 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5242 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5244 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5246 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5248 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5250 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5252 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5253 * expr.c (expand_expr_real_2): Likewise.
5254 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5255 fold_binary_loc): Likewise.
5256 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5257 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5258 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5259 MINUS_EXPR transformations.
5260 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5261 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5262 * tree-inline.c (estimate_operator_cost): Likewise.
5263 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5264 op_symbol_code): Likewise.
5265 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5266 * vr-values.c (extract_range_from_binary_expr): Likewise.
5267 * varasm.c (initializer_constant_valid_p_1): Likewise.
5268 * tree.def: New tree code POINTER_DIFF_EXPR.
5270 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5272 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5273 integer suffix to movbe mnemonic.
5274 (*bswaphi2_movbe): Ditto.
5275 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5277 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5280 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5281 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5282 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5284 2017-11-21 Martin Liska <mliska@suse.cz>
5286 * tree-inline.c (expand_call_inline): Remove not needed
5289 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5290 Jakub Jelinek <jakub@redhat.com>
5293 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5294 Remove static keyword from f variable.
5296 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5298 PR tree-optimization/83086
5299 * gimple-ssa-store-merging.c
5300 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5301 rather than n.base_addr.
5303 2017-11-21 Martin Liska <mliska@suse.cz>
5305 PR rtl-optimization/82044
5306 PR tree-optimization/82042
5307 * dse.c (check_mem_read_rtx): Check for overflow.
5309 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5311 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5314 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5316 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5317 and its corresponding call arg location note.
5319 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5320 Andrew Burgess <andrew.burgess@embecosm.com>
5322 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5324 (arc_return_slot_offset): Likewise.
5325 (arc_eh_return_address_location): New declaration.
5326 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5327 (MUST_SAVE_REGISTER): Add exception handler case.
5328 (MUST_SAVE_RETURN_ADDR): Likewise.
5329 (arc_frame_pointer_required): Likewise.
5330 (arc_frame_pointer_needed): New function.
5331 (arc_compute_frame_size): Changed.
5332 (arc_expand_prologue): Likewise.
5333 (arc_expand_epilogue): Likewise.
5334 (arc_initial_elimination_offset): Likewise.
5335 (arc_return_slot_offset): Delete.
5336 (arc_eh_return_address_location): New function.
5337 (arc_builtin_setjmp_frame_value): Likewise.
5338 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5339 (EH_RETURN_STACKADJ_RTX): Define.
5340 (EH_RETURN_HANDLER_RTX): Likewise.
5341 * config/arc/arc.md (eh_return): Delete.
5343 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5345 * print-tree.h (debug_vec_tree): Remove prototype.
5346 * gdbinit.in (pvt): Remove macro.
5348 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5350 PR tree-optimization/83047
5351 * gimple-ssa-store-merging.c
5352 (imm_store_chain_info::output_merged_store): If the loads with the
5353 same vuse are in different basic blocks, for load_gsi pick a load
5354 location that is dominated by the other loads.
5357 * config/i386/i386.c (ix86_memmodel_check): Start
5358 -Winvalid-memory-model diagnostics with lowercase letter.
5361 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5362 set in_cold_section_p to the partition containing loc_list->first.
5363 When seeing loc_list->last_before_switch node, update secname and
5364 perform range_across_switch second partition handling only after that.
5367 * run-rtl-passes.c: Include debug.h.
5368 (run_rtl_passes): Call debug_hooks->assembly_start.
5369 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5373 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5374 OPTAB_DIRECT in calls to expand_simple_binop.
5376 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5379 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5381 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5384 * spellcheck.h (best_match::blithely_get_best_candidate): New
5387 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5389 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5390 with lower case letter.
5392 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5394 * config/i386/i386.md (bswaphi2): New expander.
5395 (*bswaphi2_movbe): New insn pattern.
5396 (bswaphi -> rorhi pepehole2): New peephole pattern.
5398 2017-11-20 Jeff Law <law@redhat.com>
5400 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5401 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5402 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5403 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5404 gimple-ssa-evrp-analyze.h.
5406 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5407 set BB_VISITED here.
5408 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5410 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5412 * tree-predcom.c: Add general comment on Store-Store chains.
5413 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5415 (get_chain_last_ref_at): Rename into...
5416 (get_chain_last_write_at): ...this.
5417 (get_chain_last_write_before_load): New function.
5418 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5419 CT_STORE_STORE when write reference is added.
5420 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5421 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5422 (initialize_root_vars_store_elim_1): Ditto.
5423 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5424 definition is created.
5425 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5426 chain by replacing it with dominant stored value.
5428 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5430 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5432 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5434 * vr-values.c (extract_range_from_binary_expr): Use a full range
5437 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5439 * config/arm/arm.md (R4_REGNUM): Define constant.
5440 (nonsecure_call_internal): Remove r4 clobber.
5441 (nonsecure_call_value_internal): Likewise.
5442 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5443 clobber and resequence match_operands.
5444 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5445 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5446 (nonsecure_call_value_reg_thumb2): Likewise.
5448 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5450 PR tree-optimization/78821
5451 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5452 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5453 to the address of the base rather than the base itself.
5454 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5455 (find_bswap_or_nop_finalize): New function.
5456 (find_bswap_or_nop): Use it.
5457 (bswap_replace): Return a tree rather than bool, change first
5458 argument from gimple * to gimple_stmt_iterator, allow inserting
5459 into an empty sequence, allow ins_stmt to be NULL - then emit
5460 all stmts into gsi. Fix up MEM_REF address gimplification.
5461 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5462 (struct store_immediate_info): Add N and INS_STMT non-static
5464 (store_immediate_info::store_immediate_info): Initialize them
5465 from newly added ctor args.
5466 (merged_store_group::apply_stores): Formatting fixes. Sort by
5468 (stmts_may_clobber_ref_p): For stores call also
5469 refs_anti_dependent_p.
5470 (gather_bswap_load_refs): New function.
5471 (imm_store_chain_info::try_coalesce_bswap): New method.
5472 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5473 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5474 (imm_store_chain_info::output_merged_store): Fail if number of
5475 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5476 and NOP_EXPR rhs_code.
5477 (pass_store_merging::process_store): Compute n and ins_stmt, if
5478 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5479 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5481 (pass_store_merging::execute): Calculate dominators.
5483 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5484 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5485 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5486 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5487 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5488 class pass_optimize_bswap, bswap_replace,
5489 pass_optimize_bswap::execute): Moved to ...
5490 * gimple-ssa-store-merging.c: ... this file.
5491 Include optabs-tree.h.
5492 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5493 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5494 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5495 anonymous namespace, remove static keywords.
5496 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5497 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5499 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5502 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5504 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5506 * vec.h (debug_helper): New function.
5507 (DEFINE_DEBUG_VEC): New macro.
5508 * hash-set.h (debug_helper): New function.
5509 (DEFINE_DEBUG_HASH_SET): New macro.
5510 * cfg.c (debug_slim (edge)): New function.
5511 Call DEFINE_DEBUG_VEC for edges.
5512 Call DEFINE_DEBUG_HASH_SET for edges.
5513 * cfghooks.c (debug_slim (basic_block)): New function.
5514 Call DEFINE_DEBUG_VEC for basic blocks.
5515 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5516 * print-tree.c (debug_slim): New function to handle trees.
5517 Call DEFINE_DEBUG_VEC for trees.
5518 Call DEFINE_DEBUG_HASH_SET for trees.
5519 (debug (vec<tree, va_gc>) &): Remove.
5520 (debug (<vec<tree, va_gc>) *): Remove.
5521 * print-rtl.c (debug_slim): New function to handle const_rtx.
5522 Call DEFINE_DEBUG_VEC for rtx_def.
5523 Call DEFINE_DEBUG_VEC for rtx_insn.
5524 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5525 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5526 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5527 (debug (vec<rtx_insn *> *ptr): Remove.
5528 (debug_insn_vector): Remove.
5529 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5531 2017-11-20 Tom de Vries <tom@codesourcery.com>
5533 PR rtl-optimization/82020
5534 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5535 IF_THEN_ELSE condition.
5537 2017-11-19 Jeff Law <law@redhat.com>
5539 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5540 of degenerates resulting from ignoring an edge.
5542 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5545 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5547 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5550 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5551 for uninitialied counts.
5553 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5556 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5558 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5561 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5563 2017-11-19 Tom de Vries <tom@codesourcery.com>
5565 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5568 2017-11-19 Tom de Vries <tom@codesourcery.com>
5570 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5573 2017-11-19 Tom de Vries <tom@codesourcery.com>
5575 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5578 2017-11-19 Tom de Vries <tom@codesourcery.com>
5580 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5582 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5584 2017-11-19 Tom de Vries <tom@codesourcery.com>
5586 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5588 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5589 semicolon after MCORE_EXPORT_NAME call.
5591 2017-11-19 Tom de Vries <tom@codesourcery.com>
5594 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5595 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5597 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5599 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5600 (cgraph_edge::clone): Cleanup updating of profile.
5601 * ipa-cp.c (update_profiling_info): Likewise.
5602 * ipa-inline-transform.c (inline_transform): Likewise.
5603 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5604 * ipa-split.c (execute_split_functions): Do not split when function
5606 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5607 * profile-count.c (profile_count::dump): Add global0.
5608 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5610 (profile_count::to_sreal_scale): Likewise.
5611 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5612 (profile_count::combine_with_ipa_count): New function.
5613 * profile-count.h (profile_guessed_global0adjusted): New.
5614 (profile_count::adjusted_zero): New.
5615 (profile_count::global0adjusted): New.
5616 (profile_count::combine_with_ipa_count): New.
5617 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5618 correct profile of return block of split functions.
5619 (copy_cfg_body): Remove unused profile_count.
5620 (copy_body): Likewise.
5621 (expand_call_inline): Update.
5622 (tree_function_versioning): Update.
5624 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5626 * hash-set.h (hash_set::empty): New.
5627 * tree-ssa-threadbackward.h: Delete.
5628 * tree-ssa-threadbackward.c (class thread_jumps): New.
5629 Move max_threaded_paths into class.
5630 (fsm_find_thread_path): Remove arguments that are now in class.
5631 (profitable_jump_thread_path): Rename to...
5632 (thread_jumps::profitable_jump_thread_path): ...this.
5633 (convert_and_register_jump_thread_path): Rename to...
5634 (thread_jumps::convert_and_register_current_path): ...this.
5635 (check_subpath_and_update_thread_path): Rename to...
5636 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5637 (register_jump_thread_path_if_profitable): Rename to...
5638 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5639 (handle_phi): Rename to...
5640 (thread_jumps::handle_phi): ...this.
5641 (handle_assignment): Rename to...
5642 (thread_jumps::handle_assignment): ...this.
5643 (fsm_find_control_statement_thread_paths): Rename to...
5644 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5645 (find_jump_threads_backwards): Rename to...
5646 (thread_jumps::find_jump_threads_backwards): ...this.
5647 Initialize path local data.
5648 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5649 from within thread_jumps class.
5650 (pass_early_thread_jumps::execute): Same.
5652 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5654 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5656 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5658 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5659 to canonical location.
5661 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5664 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5666 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5668 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5670 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5671 rcrt1.o%s/grcrt1.o%s for -static-pie.
5673 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5675 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5676 ix86_shift_rotate_cost): Break out from ...
5677 (ix86_rtx_costs): ... here.
5678 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5681 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5683 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5684 when entry block was promoted unlikely.
5685 (estimate_bb_frequencies): Increase frequency scale.
5686 * profile-count.h (profile_count): Export precision info.
5688 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5690 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5691 disturbin profile of entry block.
5693 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5695 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5698 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5700 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5701 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5703 2017-11-17 Jeff Law <law@redhat.com>
5705 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5706 from evrp_dom_walker class. Various methods moved into new class.
5707 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5708 (evrp_range_analyzer::enter): New method.
5709 (evrp_range_analyzer::leave): New method.
5710 (evrp_dom_walker): Remove delegators no longer needed by this class.
5711 Replace vr_values data member with evrp_range_analyzer
5713 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5714 method extracted from evrp_dom_walker::before_dom_children.
5715 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5716 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5718 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5719 Add private copy constructor and move assignment operators.
5720 Privatize methods and class data where trivially possible.
5721 (evrp_dom_walker::cleanup): New function, extracted from
5722 execute_early_vrp. Simplify access to class data.
5724 * vr-values.h (get_output_for_vrp): Prototype.
5725 * vr-values.c (get_output_for_vrp): New function extracted from
5726 vrp_visit_assignment_or_call and extract_range_from_stmt.
5727 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5729 2017-11-17 Luis Machado <luis.machado@linaro.org>
5731 * config/aarch64/aarch64.c
5732 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5733 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5734 tune_params::AUTOPREFETCHER_WEAK.
5736 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5739 * config/arm/arm.c (arm_valid_target_attribute_rec):
5740 Parse "arch=" and "+<ext>".
5741 (arm_valid_target_attribute_tree): Re-init global options.
5742 (arm_option_override): Make non-static.
5743 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5744 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5745 (__ARM_FEATURE_CRC32): Support undef.
5746 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5747 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5749 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5751 * gdbinit.in (break-on-diagnostic): New command.
5753 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5755 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5756 return to indirect jump for EH return if control-flow
5757 protection is enabled. Change explicit 'false' argument in
5758 pro_epilogue_adjust_stack with a value of
5760 * config/i386/i386.md (simple_return_indirect_internal):
5761 Remove SImode restriction to support 64-bit.
5763 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5765 * combine.c (added_notes_insn): New.
5766 (try_combine): Handle added_notes_insn like added_links_insn.
5767 Rewrite return value code.
5768 (distribute_notes): Set added_notes_insn to the earliest insn we added
5771 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5773 PR rtl-optimization/82621
5774 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5775 dest of one of those SETs is unused.
5777 2017-11-17 Richard Biener <rguenther@suse.de>
5780 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5781 * tree-pretty-print.c (dump_generic_node): Handle
5782 annot_expr_parallel_kind.
5783 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5784 * gimplify.c (gimple_boolify): Likewise.
5786 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5788 * config.gcc (extra_headers): Add cet.h for x86 targets.
5789 * config/i386/cet.h: New file.
5790 * doc/install.texi: Add --enable-cet/--disable-cet.
5792 2017-11-17 Richard Biener <rguenther@suse.de>
5794 PR tree-optimization/83017
5795 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5796 (gen_parallel_loop): Properly count iterations.
5797 (parallelize_loops): Handle loop->can_be_parallel independent
5798 of flag_loop_parallelize_all. Make static profitability test match
5800 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5801 * invoke.texi (parloops-min-per-thread): Document.
5803 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5805 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5806 upstreaming review comments.
5808 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5810 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5811 with fast unaligned access.
5812 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5814 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5816 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5817 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5819 2017-11-17 Richard Biener <rguenther@suse.de>
5821 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5822 folding of references.
5824 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5827 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5828 of replacing call to strncmp with corresponding call to strcmp when
5831 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5833 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5834 option prefer-avx256 for skylake-avx512 configuration.
5835 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5836 (get_builtin_code_for_version): Ditto.
5838 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5839 Monk Chiang <sh.chiang04@gmail.com>
5841 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5842 (FIXED_REGISTERS): Reserve more register numbers.
5843 (CALL_USED_REGISTERS): Likewise.
5844 (REG_ALLOC_ORDER): Likewise.
5845 (REG_CLASS_CONTENTS): Likewise.
5846 (REGISTER_NAMES): Likewise.
5848 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5849 Kito Cheng <kito.cheng@gmail.com>
5851 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5853 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5855 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5857 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5858 parameter list for vec_splats.
5860 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5862 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5864 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5866 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5869 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5870 with the required version of automake 1.11.6.
5872 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5874 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5876 (DU_C2_3_power9): Correct reservation combinations.
5877 (FP_DIV_power9, VEC_DIV_power9): New.
5878 (power9-alu): Split out rotate/shift...
5879 (power9-rot): ...to here, correct dispatch resource.
5880 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5882 (power9-fp): Correct latency.
5883 (power9-sdiv): Add div/sqrt resource.
5884 (power9-ddiv): Correct latency, add div/sqrt resource.
5885 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5886 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5888 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5890 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5892 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5893 switch statement mapping KF built-ins to TF built-ins if we don't
5894 have the proper ISA 3.0 assembler support.
5896 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5898 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5899 (lower_emutls_function_body): Do not compute it.
5901 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5903 * ipa-split.c (split_bb_info): Turn time to sreal.
5904 (split_point): Likewise.
5905 (dump_split_point): Likewise.
5906 (fine_split_points): Likewise.
5907 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5909 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5911 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5914 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5916 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5918 (expensive_function_p): Remove useless assert.
5919 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5921 2017-11-16 Martin Sebor <msebor@redhat.com>
5923 PR tree-optimization/82588
5924 PR tree-optimization/82583
5925 * tree-vrp.c (check_array_ref): Handle flexible array members,
5926 string literals, and inner indices.
5927 (search_for_addr_array): Add detail to diagnostics.
5929 2017-11-16 Nathan Sidwell <nathan@acm.org>
5933 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5934 (SET_DECL_ASSEMBLER_NAME): Forward to
5935 overwrite_decl_assembler_name.
5936 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5937 (overwrite_decl_assembler_name): Declare.
5938 * tree.c (overwrite_decl_assembler_name): New.
5939 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5940 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5941 (LANG_HOOKS_INITIALIZER): Add it.
5942 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5943 * langhooks.c (lhd_set_decl_assembler_name): Use
5944 SET_DECL_ASSEMBLER_NAME.
5945 (lhd_overwrite_decl_assembler_name): Default implementation.
5947 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5948 Jackson Woodruff <jackson.woodruff@arm.com>
5950 PR tree-optimization/71026
5951 * match.pd: Canonicalize constant multiplies in division.
5953 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5955 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5956 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5957 Remove OPT_fomit_frame_pointer entry.
5958 * common/config/arc/arc-common.c: Likewise.
5959 * common/config/arm/arm-common.c: Likewise.
5960 * common/config/avr/avr-common.c: Likewise.
5961 * common/config/c6x/c6x-common.c: Likewise.
5962 * common/config/cr16/cr16-common.c: Likewise.
5963 * common/config/cris/cris-common.c: Likewise.
5964 * common/config/epiphany/epiphany-common.c: Likewise.
5965 * common/config/fr30/fr30-common.c: Likewise.
5966 * common/config/frv/frv-common.c: Likewise.
5967 * common/config/ia64/ia64-common.c: Likewise.
5968 * common/config/iq2000/iq2000-common.c: Likewise.
5969 * common/config/lm32/lm32-common.c: Likewise.
5970 * common/config/m32r/m32r-common.c: Likewise.
5971 * common/config/mcore/mcore-common.c: Likewise.
5972 * common/config/microblaze/microblaze-common.c: Likewise.
5973 * common/config/mips/mips-common.c: Likewise.
5974 * common/config/mmix/mmix-common.c: Likewise.
5975 * common/config/mn10300/mn10300-common.c: Likewise.
5976 * common/config/nios2/nios2-common.c: Likewise.
5977 * common/config/pa/pa-common.c: Likewise.
5978 * common/config/pdp11/pdp11-common.c: Likewise.
5979 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5980 * common/config/riscv/riscv-common.c: Likewise.
5981 * common/config/rs6000/rs6000-common.c: Likewise.
5982 * common/config/rx/rx-common.c: Likewise.
5983 * common/config/s390/s390-common.c: Likewise.
5984 * common/config/sh/sh-common.c: Likewise.
5985 * common/config/sparc/sparc-common.c: Likewise.
5986 * common/config/tilegx/tilegx-common.c: Likewise.
5987 * common/config/tilepro/tilepro-common.c: Likewise.
5988 * common/config/v850/v850-common.c: Likewise.
5989 * common/config/visium/visium-common.c: Likewise.
5990 * common/config/xstormy16/xstormy16-common.c: Likewise.
5991 * common/config/xtensa/xtensa-common.c: Likewise.
5992 * invoke.texi (-fomit-frame-pointer): Update documentation.
5994 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5996 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5998 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6000 * cfg.c (scale_bbs_frequencies_int,
6001 cale_bbs_frequencies_gcov_type): Remove.
6002 * cfg.h (scale_bbs_frequencies_int,
6003 cale_bbs_frequencies_gcov_type): Remove.
6005 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6007 * tree-ssa-loop-manip.c
6008 (scale_dominated_blocks_in_loop): Update to profile counts.
6009 (tree_transform_and_unroll_loop): Likewise.
6011 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6013 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6014 scale_bbs_frequencies_int.
6016 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6018 * final.c (compute_alignments): Use counts rather than frequencies.
6020 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6022 * cfgloopanal.c: Include sreal.h
6023 (average_num_loop_insns): Use counts and sreal for accounting.
6025 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6027 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6030 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6031 Kito Cheng <kito.cheng@gmail.com>
6033 * config/nds32/constraints.md: Provide more constraints.
6034 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6035 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6036 support constraints usage.
6038 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6040 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6042 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6043 Kito Cheng <kito.cheng@gmail.com>
6045 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6046 * config/nds32/nds32.opt: Refine the layout.
6047 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6048 TARGET_EXT_STRING): Support new options.
6049 * config/nds32/nds32.h: Likewise.
6050 * config/nds32/nds32.md: Likewise.
6051 * config/nds32/nds32-predicates.c: Likewise.
6052 * config/nds32/constraints.md: Likewise.
6053 * common/config/nds32/nds32-common.c: Likewise.
6055 2017-11-16 Julia Koval <julia.koval@intel.com>
6058 * config/i386/gfniintrin.h: Add sse check.
6059 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6061 2017-11-16 Julia Koval <julia.koval@intel.com>
6063 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6064 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6065 (ix86_handle_option): Handle -mavx512vbmi2.
6066 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6067 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6068 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6069 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6070 (ix86_valid_target_attribute_inner_p): Ditto.
6071 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6072 * config/i386/i386.opt (mavx512vbmi2): New option.
6073 * doc/invoke.texi: Add new option.
6075 2017-11-16 Julia Koval <julia.koval@intel.com>
6077 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6078 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6079 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6080 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6081 _mm512_gf2p8mul_epi8): New intrinsics.
6082 * config/i386/i386-builtin-types.def
6083 (V64QI_FTYPE_V64QI_V64QI): New type.
6084 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6085 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6086 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6087 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6088 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6089 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6091 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6093 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6094 explicitly as a stream of bytes.
6096 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6098 * config/rs6000/altivec.h (vec_xst_be): New #define.
6099 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6100 and externalize from *altivec_vperm_<mode>_internal.
6101 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6103 (XL_BE_V8HI): Likewise.
6104 (XL_BE_V4SI): Likewise.
6105 (XL_BE_V4SI): Likewise.
6106 (XL_BE_V2DI): Likewise.
6107 (XL_BE_V4SF): Likewise.
6108 (XL_BE_V2DF): Likewise.
6109 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6110 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6111 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6112 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6113 VSX_BUILTIN_VEC_XST_BE.
6114 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6115 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6117 (altivec_init_builtins): Replace conditional calls to def_builtin
6118 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6119 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6120 with unconditional calls. Remove calls to def_builtin for
6121 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6122 __builtin_vec_xst_be.
6123 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6124 to define_expand, and add alternate RTL generation for P8.
6125 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6126 vsx_ld_elemrev_v8hi.
6127 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6128 add alternate RTL generation for P8.
6129 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6130 vsx_ld_elemrev_v16qi.
6131 (vsx_st_elemrev_v8hi): Convert define_insn
6132 to define_expand, and add alternate RTL generation for P8.
6133 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6134 vsx_st_elemrev_v8hi.
6135 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6136 add alternate RTL generation for P8.
6137 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6138 vsx_st_elemrev_v16qi.
6140 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6143 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6144 TARGET_AVX512ER check.
6145 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6146 neither -mzeroupper nor -mno-zeroupper is used and
6147 TARGET_EMIT_VZEROUPPER is set.
6148 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6149 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6151 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6153 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6154 folding of vector compares.
6155 (fold_build_vec_cmp): New helper function.
6156 (fold_compare_helper): New helper function.
6157 (builtin_function_type): Add compare builtins to the list of functions
6158 having unsigned arguments. Cosmetic updates to comment indentation.
6159 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6160 the not+eq combination.
6162 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6164 PR tree-optimization/82726
6165 PR tree-optimization/70754
6166 * tree-predcom.c (order_drefs_by_pos): New function.
6167 (combine_chains): Move code setting has_max_use_after to...
6168 (try_combine_chains): ...here. New parameter. Sort combined chains
6169 according to position information.
6170 (tree_predictive_commoning_loop): Update call to above function.
6171 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6173 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6175 PR tree-optimization/82726
6177 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6179 PR tree-optimization/70754
6180 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6181 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6182 combined stmt before it if not NULL.
6183 (combine_chains): Process refs reversely and compute dominance point
6187 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6189 PR tree-optimization/79663
6190 * tree-predcom.c (combine_chains): Process refs in reverse order
6191 only for ZERO length chains, and add explaining comment.
6193 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6195 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6196 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6197 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6199 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6201 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6203 2017-11-15 Martin Liska <mliska@suse.cz>
6205 * tree-cfg.c (pass_warn_function_return::execute):
6206 Compare warn_return_type for greater than zero.
6208 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6212 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6213 to return true on Xeon and not on Xeon Phi.
6214 (ix86_check_avx256_register): Changed to ...
6215 (ix86_check_avx_upper_register): ... this. Add extra check for
6216 VALID_AVX512F_REG_OR_XI_MODE.
6217 (ix86_avx_u128_mode_needed): Changed
6218 ix86_check_avx256_register to ix86_check_avx_upper_register.
6219 (ix86_check_avx256_stores): Changed to ...
6220 (ix86_check_avx_upper_stores): ... this. Changed
6221 ix86_check_avx256_register to ix86_check_avx_upper_register.
6222 (ix86_avx_u128_mode_after): Changed
6223 avx_reg256_found to avx_upper_reg_found. Changed
6224 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6225 (ix86_avx_u128_mode_entry): Changed
6226 ix86_check_avx256_register to ix86_check_avx_upper_register.
6227 (ix86_avx_u128_mode_exit): Ditto.
6228 * config/i386/i386.h: (host_detect_local_cpu): New define.
6230 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6232 * config/arm/xgene1.md (xgene1): Split into automatons
6233 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6234 (xgene1_f_load): Adjust reservations and/or types.
6235 (xgene1_f_store): Likewise.
6236 (xgene1_load_pair): Likewise.
6237 (xgene1_store_pair): Likewise.
6238 (xgene1_fp_load1): Likewise.
6239 (xgene1_load1): Likewise.
6240 (xgene1_store1): Likewise.
6241 (xgene1_move): Likewise.
6242 (xgene1_alu): Likewise.
6243 (xgene1_simd): Likewise.
6244 (xgene1_bfm): Likewise.
6245 (xgene1_neon_load1): Likewise.
6246 (xgene1_neon_store1): Likewise.
6247 (xgene1_neon_logic): Likewise.
6248 (xgene1_neon_st1): Likewise.
6249 (xgene1_neon_ld1r): Likewise.
6250 (xgene1_alu_cond): Added.
6251 (xgene1_shift_reg): Likwise.
6252 (xgene1_bfx): Likewise.
6253 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6255 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6258 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6260 (can_widen_mult_without_libcall): New function.
6261 (expand_mul_overflow): If only checking unsigned mul overflow,
6262 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6263 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6264 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6265 (expand_DIVMOD): Formatting fix.
6266 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6267 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6268 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6270 PR tree-optimization/82977
6271 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6272 constructed temporary to strlen_to_stridx.put.
6274 2017-11-15 Martin Liska <mliska@suse.cz>
6276 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6277 * configure: Regenerate.
6279 2017-11-15 Martin Liska <mliska@suse.cz>
6282 * config/sh/sh-mem.cc: Use proper probability for
6285 2017-11-14 Jeff Law <law@redhat.com>
6287 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6288 the red zone for stack_clash_protection_final_dynamic_probe targets
6289 when the total dynamic stack size is zero bytes.
6291 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6292 blocks is post order.
6294 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6296 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6297 * final.c (rest_of_clean_state): Set it for the
6298 -fcompare-debug dump.
6299 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6300 class when TDF_COMPARE_DEBUG is set.
6302 * dwarf2out.c (gen_producer_string): Discard
6305 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6308 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6309 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6310 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6311 (__TGMATH_REAL_2_3): Remove macros.
6312 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6313 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6315 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6317 (remquo): Define using __TGMATH_REAL_3.
6319 2017-11-14 Jeff Law <law@redhat.com>
6321 * vr-values.c: New file with contents extracted from tree-vrp.c.
6322 * Makefile.in (OBJS): Add vr-values.o
6323 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6324 (set_value_range, set_and_canonicalize_value_range): Likewise.
6325 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6326 (value_range_constant_singleton, symbolic_range_p): Likewise.
6327 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6328 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6329 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6330 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6331 (find_case_label_range, find_case_label_index): Likewise.
6332 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6333 (range_int_cst_singleton_p, value_inside_range): Likewise.
6334 (get_single_symbol): Likewise.
6335 (switch_update): Move structure definition here.
6336 (to_remove_edges, to_update_switch_stmts): Provide externs.
6337 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6338 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6339 (vrp_val_is_min, set_value_range): Likewise.
6340 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6341 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6342 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6343 (range_is_nonnull, range_int_cst_p): Likewwise.
6344 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6345 (get_single_symbol, operand_less_p): Likewise
6346 (compare_values_warnv, compare_values): Likewise.
6347 (value_inside_range, value_range_constant_singleton): Likewise.
6348 (zero_nonzero_bitgs_from_vr): Likewise.
6349 (extract_range_from_binary_expr_1): Likewise.
6350 (overflow_comparison_p): Likewise.
6351 (to_remove_edges, to_update_switch_stmts): Likewise.
6352 (find_case_label-index, find_case_label_range): Likewise.
6353 (switch_update, set_value_range_to_nonnegative): Remove.
6354 (set_value_range_to_truthvalue): Likewise.
6355 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6356 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6357 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6358 (find_case_label_ranges, test_for_singularity): Likewise.
6359 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6360 (x_vr_values): Move to its remaining use site.
6362 2017-11-10 Jeff Law <law@redhat.com>
6364 * vr-values.h (VR_INITIALIZER): Move #define here.
6365 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6366 * Makefile.in (OBJS): Add tree-evrp.o
6367 * tree-vrp.h (assert_info): Move structure definition here.
6368 (set_value_range_to_varying): Prototype.
6369 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6370 (infer_value_range, register_edge_assert_for): Likewise.
6371 (stmt_interesting_for_vrp): Likewise.
6372 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6373 (set_value_range_to_varying): No longer static.
6374 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6375 (infer_value_range, register_edge_assert_for): Likewise.
6377 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6379 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6380 generate the XXBRD instruction.
6382 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6383 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6384 as being compatible if -mabi=ieeelongdouble.
6385 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6386 to setup float128 built-ins with hardware support.
6387 (BU_FLOAT128_HW_2): Likewise.
6388 (BU_FLOAT128_HW_3): Likewise.
6389 (BU_FLOAT128_HW_VSX_1): Likewise.
6390 (BU_FLOAT128_HW_VSX_2): Likewise.
6391 (scalar_extract_expq): Change float128 built-in functions to
6392 accommodate having both KFmode and TFmode functions. Use the
6393 KFmode variant as the default.
6394 (scalar_extract_sigq): Likewise.
6395 (scalar_test_neg_qp): Likewise.
6396 (scalar_insert_exp_q): Likewise.
6397 (scalar_insert_exp_qp): Likewise.
6398 (scalar_test_data_class_qp): Likewise.
6399 (sqrtf128_round_to_odd): Delete processing the round to odd
6400 built-in functions as special built-in functions, and define them
6401 as float128 built-ins. Use the KFmode variant as the default.
6402 (truncf128_round_to_odd): Likewise.
6403 (addf128_round_to_odd): Likewise.
6404 (subf128_round_to_odd): Likewise.
6405 (mulf128_round_to_odd): Likewise.
6406 (divf128_round_to_odd): Likewise.
6407 (fmaf128_round_to_odd): Likewise.
6408 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6409 support for KFmode and TFmode xststdcqp calls.
6410 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6411 point, switch the built-in handlers for the get/set float128
6412 exponent, get float128 mantissa, float128 test built-ins, and the
6413 float128 round to odd built-in functions. Eliminate creating the
6414 float128 round to odd built-in functions as special built-ins.
6415 (rs6000_init_builtins): Eliminate special creation of the float128
6416 round to odd built-in functions.
6417 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6418 function insns to support both TFmode and KFmode variants.
6419 (xsxsigqp_<mode>): Likewise.
6420 (xsiexpqpf_<mode>): Likewise.
6421 (xsiexpqp_<mode>): Likewise.
6422 (xststdcqp_<mode>): Likewise.
6423 (xststdcnegqp_<mode>): Likewise.
6424 (xststdcqp_<mode>): Likewise.
6426 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6428 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6429 unused path_in_freq_ptr parameter.
6430 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6432 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6434 * ipa-inline.c (edge_badness): Dump sreal frequency.
6435 (compute_inlined_call_time): Match natural implementaiton ...
6436 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6437 forgotten division by CGRAPH_FREQ_BASE.
6439 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6441 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6442 Solaris 11. Update comment.
6443 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6445 * config/sol2.h (STARTFILE_SPEC): Likewise.
6446 * configure: Regenerate.
6448 2017-11-14 Carl Love <cel@us.ibm.com>
6450 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6451 le_ and be_ prefixes to swap* variables. Remove
6452 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6455 2017-11-14 Jason Merrill <jason@redhat.com>
6457 Support GTY((cache)) on hash_map.
6458 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6459 (ggc_cache_remove): Override it instead of ggc_mx.
6460 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6461 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6462 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6463 (simple_cache_map_traits): Override maybe_mx.
6464 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6465 (hash_map): Friend gt_cleare_cache.
6466 (gt_cleare_cache): New.
6467 * tree.h (tree_cache_traits): New hash_map traits class.
6468 (tree_cache_map): New typedef.
6470 2017-11-14 Richard Biener <rguenther@suse.de>
6472 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6473 paramter and handling.
6474 (cleanup_control_flow_bb): Likewise.
6475 (cleanup_control_flow_pre): New helper performing a DFS walk
6476 to call cleanup_control_flow_bb in PRE order.
6477 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6478 via cleanup_control_flow_pre.
6480 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6482 * config/aarch64/aarch64-simd.md
6483 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6484 (*aarch64_simd_bsl<mode>_alt): Likewise.
6485 (aarch64_simd_bsldi_internal): New.
6486 (aarch64_simd_bsldi_alt): Likewise.
6488 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6490 * tracer.c (better_p): Do not compare frequencies.
6491 * reg-stack.c (better_edge): Likewise.
6492 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6495 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6497 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6498 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6499 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6500 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6501 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6502 * predict.c (maybe_hot_frequency_p): Inline to ...
6503 (maybe_hot_count_p): ... here; rewrite to counts.
6504 (counts_to_freqs): Rename to ...
6505 (update_max_bb_count): ... this one.
6506 (expensive_function_p): Use counts.
6507 (estimate_bb_frequencies): Update.
6508 (rebuild_frequencies): Update.
6509 * predict.h (counts_to_freqs): Rename to ...
6510 (update_max_bb_count): ... this one.
6511 * profile.c (compute_branch_probabilities): Add debug info
6512 * tree-inline.c (expand_call_inline): Update debug info.
6513 (optimize_inline_calls): Use update_max_bb_count..
6514 (tree_function_versioning): Use update_max_bb_count..
6515 * value-prof.c (gimple_value_profile_transformations):
6516 Do not use update_max_bb_count.
6518 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6520 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6521 always use frequencies.
6523 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6525 * bb-reorder.c: Remove frequencies from comments.
6526 (better_edge_p): Use profile counts.
6527 (find_traces): Dump profile counts.
6528 (rotate_loop): Use profile counts.
6529 (find_traces_1_round): Likewise.
6530 (connect_better_edge_p): Use counts instead of probabilities for
6532 (copy_bb_p): Drop early check for non-0 frequency.
6533 (sanitize_hot_paths): Update comments.
6535 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6537 * ipa-split.c (struct split_point): Add count.
6538 (consider_split): Do not compute incoming frequency; compute incoming
6539 count and store it to split_point.
6540 (split_function): Set count of the call to split part correctly.
6542 2017-11-13 Carl Love <cel@us.ibm.com>
6544 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6546 2017-11-13 Tom Tromey <tom@tromey.com>
6548 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6550 2017-11-13 Carl Love <cel@us.ibm.com>
6552 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6553 Add support for builtins:
6554 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6555 vector {un,}signed {char,int,short},
6556 vector {un,}signed {char,int,short}) arguments.
6557 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6558 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6559 Add BU_P9V_AV_2 expansions for the builtins.
6560 * config/rs6000/altivec.h (vec_first_match_index,
6561 vec_first_mismatch_index, vec_first_match_or_eos_index,
6562 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6563 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6564 new extern declaration.
6565 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6566 * config/rs6000/vsx.md (first_match_index_<mode>,
6567 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6568 first_mismatch_or_eos_index_<mode>): Add define expand.
6569 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6570 * doc/extend.texi: Update the built-in documenation file for the new
6573 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6575 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6576 into the min/max operations for _Float<N> and _Float<N>X types.
6578 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6581 * dwarf2out.c (do_eh_frame): New static variable.
6582 (dwarf2out_begin_prologue): Set it.
6583 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6585 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6587 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6589 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6591 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6594 * debug.h (dwarf2out_do_eh_frame): Declare.
6595 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6596 (dwarf2out_do_frame): Use it.
6597 (dwarf2out_do_cfi_asm): Likewise.
6598 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6599 (dwarf2out_assembly_start): Likewise.
6600 (dwarf2out_begin_prologue): Fix comment.
6601 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6602 if the target needs either debug or unwind DWARF2 info.
6603 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6604 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6605 -ftrapping-math, -ftrapv and -fwrapv.
6607 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6609 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6610 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6611 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6612 (estimate_edge_size_and_time): Likewise.
6613 (ipa_merge_fn_summary_after_inlining): Likewise.
6614 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6615 (compute_uninlined_call_time): Use sreal_frequency.
6616 (compute_inlined_call_time): Likewise.
6617 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6618 * profile-count.c: Include sreal.h.
6619 (profile_count::to_sreal_scale): New.
6620 * profile-count.h: Forward declare sreal.
6621 (profile_count::to_sreal_scale): Declare.
6623 2017-11-13 Nathan Sidwell <nathan@acm.org>
6625 * diagnostic.c (maybe_line_and_column): New.
6626 (diagnostic_get_location_text): Use it.
6627 (diagnostic_report_current_module): Likewise.
6628 (test_diagnostic_get_location_text): Add tests.
6630 2017-11-13 Luis Machado <luis.machado@linaro.org>
6632 * doc/md.texi (Specifying processor pipeline description): Fix
6633 incorrect latency for the div instruction example.
6635 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6637 PR tree-optimization/78821
6638 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6639 that bit_not_p is the same.
6640 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6641 (split_group): Count precisely bit_not_p bits in each statement.
6642 (invert_op): New function.
6643 (imm_store_chain_info::output_merged_store): Use invert_op to
6644 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6645 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6647 2017-11-13 Martin Liska <mliska@suse.cz>
6649 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6650 (struct source_info): Likewise.
6651 (add_branch_counts): Likewise.
6652 (add_line_counts): Likewise.
6653 (function_summary): Likewise.
6654 (output_intermediate_line): Likewise.
6655 (generate_results): Likewise.
6657 2017-11-13 Martin Liska <mliska@suse.cz>
6659 * gcov.c (struct block_info): Remove typedef for block_t.
6660 (struct line_info): Likewise.
6661 (line_info::has_block): Likewise.
6662 (EXIT_BLOCK): Likewise.
6663 (unblock): Likewise.
6664 (circuit): Likewise.
6665 (get_cycles_count): Likewise.
6666 (process_file): Likewise.
6667 (read_graph_file): Likewise.
6668 (solve_flow_graph): Likewise.
6669 (find_exception_blocks): Likewise.
6670 (add_line_counts): Likewise.
6671 (accumulate_line_info): Likewise.
6672 (output_line_details): Likewise.
6674 2017-11-13 Martin Liska <mliska@suse.cz>
6676 * gcov.c (struct arc_info): Remove typedef for arc_t.
6677 (struct line_info): Likewise.
6678 (add_branch_counts): Likewise.
6679 (output_branch_count): Likewise.
6680 (function_info::~function_info): Likewise.
6681 (circuit): Likewise.
6682 (output_intermediate_line): Likewise.
6683 (read_graph_file): Likewise.
6684 (solve_flow_graph): Likewise.
6685 (find_exception_blocks): Likewise.
6686 (add_line_counts): Likewise.
6687 (accumulate_line_info): Likewise.
6688 (output_line_details): Likewise.
6689 (output_function_details): Likewise.
6691 2017-11-13 Martin Liska <mliska@suse.cz>
6693 * gcov.c (struct function_info): Remove typedef for function_t.
6694 (struct source_info): Likewise.
6695 (source_info::get_functions_at_location): Likewise.
6696 (solve_flow_graph): Likewise.
6697 (find_exception_blocks): Likewise.
6698 (add_line_counts): Likewise.
6699 (output_intermediate_file): Likewise.
6700 (process_file): Likewise.
6701 (generate_results): Likewise.
6702 (release_structures): Likewise.
6703 (read_graph_file): Likewise.
6704 (read_count_file): Likewise.
6705 (accumulate_line_counts): Likewise.
6706 (output_lines): Likewise.
6708 2017-11-13 Martin Liska <mliska@suse.cz>
6710 * gcov.c (function_info::function_info): Remove num_counts
6711 and add vector<gcov_type>.
6712 (function_info::~function_info): Use the vector.
6713 (process_file): Likewise.
6714 (read_graph_file): Likewise.
6715 (read_count_file): Likewise.
6716 (solve_flow_graph): Likewise.
6718 2017-11-13 Martin Liska <mliska@suse.cz>
6720 * gcov.c (function_info::is_artificial): New function.
6721 (process_file): Erase all artificial early.
6722 (generate_results): Skip as all artificial are already
6725 2017-11-13 Martin Liska <mliska@suse.cz>
6727 * gcov.c (read_graph_file): Store to global vector of functions.
6728 (read_count_file): Iterate the vector.
6729 (process_file): Likewise.
6730 (generate_results): Likewise.
6731 (release_structures): Likewise.
6733 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6735 PR tree-optimization/82954
6736 * gimple-ssa-store-merging.c
6737 (imm_store_chain_info::coalesce_immediate_stores): If
6738 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6740 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6742 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6743 Upddate call to ENDIAN_LANE_N.
6744 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6745 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6746 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6747 and use aarch64_endian_lane_rtx.
6748 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6750 2017-11-12 Tom de Vries <tom@codesourcery.com>
6752 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6754 2017-11-12 Tom de Vries <tom@codesourcery.com>
6756 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6758 2017-11-12 Tom de Vries <tom@codesourcery.com>
6760 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6762 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6763 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6764 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6766 2017-11-11 Martin Sebor <msebor@redhat.com>
6769 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6772 * prefic.c (translate_name): Replace strncpy with memcpy to
6773 avoid -Wstringop-truncation.
6775 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6777 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6779 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6781 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6783 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6785 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6788 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6790 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6791 enable generating XXBRH if the value is in a vector register.
6792 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6793 value is in a vector register.
6794 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6795 register to register bswap64's instead of doing the GPR sequence
6796 used on previous machines.
6797 (bswapdi2_xxbrd): New insn.
6798 (bswapdi2_reg): Disallow on ISA 3.0.
6799 (register to register bswap64 splitter): Do not split the insn on
6800 ISA 3.0 systems that use XXBRD.
6802 2017-11-10 Martin Sebor <msebor@redhat.com>
6805 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6807 (find_subframework_file): Same.
6809 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6811 * auto-profile.c (afdo_indirect_call): Drop frequency.
6812 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6813 (cgraph_node::create_edge): Drop frequency argument.
6814 (cgraph_node::create_indirect_edge): Drop frequency argument.
6815 (cgraph_edge::make_speculative): Drop frequency arguments.
6816 (cgraph_edge::resolve_speculation): Do not update frequencies
6817 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6818 (cgraph_node::dump): Check consistency in IPA mode.
6819 (cgraph_edge::maybe_hot_p): Use IPA counter.
6820 (cgraph_edge::verify_count_and_frequency): Rename to ...
6821 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6822 (cgraph_node::verify_node): Update.
6823 * cgraph.h (struct cgraph_edge): Drop frequency.
6824 (cgraph_edge::frequency): New function.
6825 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6827 (cgraph_edge::rebuild_edges): Likewise.
6828 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6829 (duplicate_thunk_for_node): Do not pass frequency.
6830 (cgraph_node::create_clone): Scale only counts.
6831 (cgraph_node::create_virtual_clone): Do not pass frequency.
6832 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6833 (cgraph_node::create_version_clone): Do not pass frequency.
6834 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6835 (cgraph_node::expand_thunk): Do not pass frequency.
6836 (cgraph_node::create_wrapper): Do not pass frequency.
6837 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6839 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6840 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6841 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6842 (gather_caller_stats): Use frequency function.
6843 (ipcp_cloning_candidate_p): Use frequency function.
6844 (ipcp_propagate_stage): Use frequency function.
6845 (get_info_about_necessary_edges): Use frequency function.
6846 (update_profiling_info): Update only IPA profile.
6847 (update_specialized_profile): Use frequency functoin.
6848 (perhaps_add_new_callers): Update only IPA profile.
6849 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6850 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6851 (dump_ipa_call_summary): Use frequency function.
6852 (estimate_edge_size_and_time): Use frequency function.
6853 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6854 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6855 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6856 (update_noncloned_counts): ... ths one; scale counts only.
6857 (clone_inlined_nodes): Do not scale frequency.
6858 (inline_call): Do not pass frequency.
6859 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6860 (compute_inlined_call_time): Use IPA profile.
6861 (want_inline_small_function_p): Use IPA profile.
6862 (want_inline_self_recursive_call_p): Use IPA profile.
6863 (edge_badness): Use IPA profile.
6864 (lookup_recursive_calls): Use IPA profile.
6865 (recursive_inlining): Do not pass frequency.
6866 (resolve_noninline_speculation): Do not update frequency.
6867 (inline_small_functions): Collect max of IPA profile.
6868 (dump_overall_stats): Dump IPA porfile.
6869 (dump_inline_stats): Dump IPA porfile.
6870 (ipa_inline): Collect IPA stats.
6871 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6872 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6873 (ipa_propagate_frequency): Use frequency function.
6874 (ipa_profile): Cleanup.
6875 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6876 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6877 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6878 (input_node): Do not stream frequency.
6879 (input_edge): Do not stream frequency.
6880 (merge_profile_summaries): Scale only IPA profiles.
6881 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6882 * predict.c (drop_profile): Do not recompute frequency.
6883 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6884 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6885 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6886 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6887 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6888 * tree-inline.c (copy_bb): Do not scale frequency.
6889 (expand_call_inline): Do not scale frequency.
6890 (tree_function_versioning): Do not scale frequency.
6891 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6893 2017-11-10 Julia Koval <julia.koval@intel.com>
6895 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6896 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6897 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6898 (_mm256_maskz_gf2p8affine_epi64_epi8)
6899 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6900 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6901 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6902 (__builtin_ia32_vgf2p8affineqb_v32qi)
6903 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6904 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6906 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6910 (arm_option_override): Refactor.
6911 (arm_option_reconfigure_globals): New.
6912 (arm_options_perform_arch_sanity_checks): New.
6913 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6915 (arm_options_perform_arch_sanity_checks): Likewise
6917 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6919 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6920 (power9-qpmul): New.
6921 * rs6000/rs6000.md ("type" attr): Add qmul.
6922 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6923 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6924 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6926 2017-11-10 Martin Sebor <msebor@redhat.com>
6929 * builtins.c (compute_objsize): Handle arrays that
6930 compute_builtin_object_size likes to fail for. Make extern.
6931 * builtins.h (compute_objsize): Declare.
6932 (check_strncpy_sizes): New function.
6933 (expand_builtin_strncpy): Call check_strncpy_sizes.
6934 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6935 -Wstringop-truncation.
6936 (gimple_fold_builtin_strncat): Same.
6937 * gimple.c (gimple_build_call_from_tree): Set call location.
6938 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6939 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6940 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6941 (handle_builtin_strlen): Use strlen_to_stridx.
6942 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6944 Use strlen_to_stridx.
6945 (pass_strlen::execute): Release strlen_to_stridx.
6946 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6947 (-Wstringop-truncation): Document new option.
6949 2017-11-10 Martin Liska <mliska@suse.cz>
6951 PR gcov-profile/82702
6952 * gcov.c (main): Handle intermediate files in a different
6954 (get_gcov_intermediate_filename): New function.
6955 (output_gcov_file): Remove support of intermediate files.
6956 (generate_results): Allocate intermediate file.
6957 (release_structures): Clean-up properly fn_end.
6958 (output_intermediate_file): Start iterating with line 1.
6960 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6962 PR tree-optimization/82929
6963 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6964 ops_swapped_p non-static data member.
6965 (store_immediate_info::store_immediate_info): Clear it.
6966 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6967 ops set ops_swapped_p.
6968 (count_multiple_uses): Handle ops_swapped_p.
6970 2017-11-10 Martin Liska <mliska@suse.cz>
6972 * coverage.c (coverage_init): Stream information about
6973 support of has_unexecuted_blocks.
6974 * doc/gcov.texi: Document that.
6975 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6976 * gcov.c (read_graph_file): Likewise.
6977 (output_line_beginning): Fix a small issue with
6980 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6982 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6983 reference of trivial component.
6985 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6988 * gimple-ssa-store-merging.c
6989 (pass_store_merging::terminate_all_aliasing_chains): For
6990 gimple_store_p stmts also call refs_output_dependent_p.
6992 PR rtl-optimization/82913
6993 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6996 2017-11-09 Jeff Law <law@redhat.com>
6998 * vr-values.h: New file with vr_values class.
6999 * tree-vrp.c: Include vr-values.h
7000 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7001 data objects into the vr_values class.
7002 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7003 (get_value_range): Make it a member function within vr_values class.
7004 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7005 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7006 (op_with_constant_singleton_value_range): Likewise.
7007 (extract_range_for_var_from_comparison_expr): Likewise.
7008 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7009 (extract_range_from_binary_expr): Likewise.
7010 (extract_range_from_unary_expr): Likewise.
7011 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7012 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7013 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7014 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7015 (compare_name_with_value, compare_names): Likewise.
7016 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7017 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7018 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7019 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7020 (extract_range_from_phi_node): Likewise.
7021 (simplify_truth_ops_using_ranges): Likewise.
7022 (simplify_div_or_mod_using_ranges): Likewise.
7023 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7024 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7025 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7026 (simplify_float_conversion_using_ranges): Likewise.
7027 (simplify_internal_call_using_ranges): Likewise.
7028 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7029 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7030 poitner via x_vr_values for calls into gimple folder.
7031 (vrp_initialize_lattice): Make this the vr_values ctor.
7032 (vrp_free_lattice): Make this the vr_values dtor.
7033 (set_vr_value): New function.
7034 (class vrp_prop): Add vr_values data member. Add various member
7035 functions as well as member functions that delegate to vr_values.
7036 (check_array_ref): Make a member function within vrp_prop class.
7037 (search_for_addr_array, vrp_initialize): Likewise.
7038 (vrp_finalize): Likewise. Revamp to avoid direct access to
7039 vr_value, values_propagated, etc.
7040 (check_array_bounds): Extract vrp_prop class instance pointer from
7041 walk info structure. Use it to call member functions.
7042 (check_all_array_refs): Make a member function within vrp_prop class.
7043 Smuggle class instance pointer via walk info structure.
7044 (x_vr_values): New local static.
7045 (vrp_valueize): Use x_vr_values to get class instance.
7046 (vr_valueize_1): Likewise.
7047 (class vrp_folder): Add vr_values data member. Add various member
7048 functions as well as member functions that delegate to vr_values.
7049 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7050 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7051 class instance from vr_values. Use it to call member functions.
7052 (vrp_dom_walker): Add vr_values data member.
7053 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7054 instance via x_vr_values.
7055 (identify_jump_threads): Accept vr_values as argument. Store
7056 it into the walker structure.
7057 (evrp_dom_walker): Add vr_values class data member. Add various
7059 (evrp_dom_walker::try_find_new_range): Use vr_values data
7060 member to access the memory allocator.
7061 (evrp_dom_walker::before_dom_children): Store vr_values class
7062 instance into the vrp_folder class.
7063 (evrp_dom_walker::push_value_range): Rework to avoid direct
7064 access to num_vr_values and vr_value.
7065 (evrp_dom_walker::pop_value_range): Likewise.
7066 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7067 Use vr_values to get to dump_all_value_ranges member function.
7068 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7069 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7070 Pass vr_values class instance down to identify_jump_threads.
7071 Remove call to vrp_free_lattice.
7072 (debug_all_value_ranges): Remove.
7074 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7075 (vrp_folder): Likewise.
7077 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7078 Get it from the existing bitmap instead.
7079 (vrp_intersect_ranges_1): Likewise.
7081 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7083 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7085 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7086 set bit_not_p to it.
7087 (imm_store_chain_info::coalesce_immediate_stores): Break group
7088 if bit_not_p is different.
7089 (count_multiple_uses, split_group,
7090 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7091 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7092 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7093 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7094 to store_immediate_info ctor.
7096 2017-11-09 Jim Wilson <jimw@sifive.com>
7098 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7099 (scan_prog_file): Likewise.
7101 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7103 * bb-reorder.c (max_entry_frequency): Remove.
7104 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7105 connect_traces, push_to_next_round_p): Remove prototypes.
7106 (find_traces_1_round): Use counts only.
7107 (push_to_next_round_p): Likewise.
7108 (find_traces): Likewise.
7109 (rotate_loop): Likewise.
7110 (find_traces_1_round): Likewise.
7111 (connect_traces): Likewise.
7112 (edge_order): Likewise.
7114 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7116 * config/arm/arm.c (output_return_instruction): Add comments to
7117 indicate requirement for cmse_nonsecure_entry return to account
7118 for the size of clearing instruction output here.
7119 (thumb_exit): Likewise.
7120 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7121 return in hardfloat mode.
7123 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7125 * config/rs6000/rs6000.c (machine_function): Add a bool,
7126 "toc_is_wrapped_separately".
7127 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7128 if it wasn't explicitly set or unset, we are optimizing for speed, and
7129 doing separate shrink-wrapping.
7130 (rs6000_get_separate_components): Enable the TOC component if
7131 saving the TOC register in the prologue.
7132 (rs6000_components_for_bb): Handle the TOC component.
7133 (rs6000_emit_prologue_components): Store the TOC register where needed.
7134 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7135 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7137 2017-11-09 Martin Jambor <mjambor@suse.cz>
7139 * ipa-param-manipulation.c: New file.
7140 * ipa-param-manipulation.h: Likewise.
7141 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7142 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7143 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7144 (ipa_parm_adjustment): Likewise.
7145 (ipa_parm_adjustment_vec): Likewise.
7146 (ipa_get_vector_of_formal_parms): Moved declaration to
7147 ipa-param-manipulation.h.
7148 (ipa_get_vector_of_formal_parm_types): Likewise.
7149 (ipa_modify_formal_parameters): Likewise.
7150 (ipa_modify_call_arguments): Likewise.
7151 (ipa_combine_adjustments): Likewise.
7152 (ipa_dump_param_adjustments): Likewise.
7153 (ipa_modify_expr): Likewise.
7154 (ipa_get_adjustment_candidate): Likewise.
7155 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7156 ipa-param-manipulation.c.
7157 (ipa_get_vector_of_formal_parm_types): Likewise.
7158 (ipa_modify_formal_parameters): Likewise.
7159 (ipa_modify_call_arguments): Likewise.
7160 (ipa_modify_expr): Likewise.
7161 (get_ssa_base_param): Likewise.
7162 (ipa_get_adjustment_candidate): Likewise.
7163 (index_in_adjustments_multiple_times_p): Likewise.
7164 (ipa_combine_adjustments): Likewise.
7165 (ipa_dump_param_adjustments): Likewise.
7166 * tree-sra.c: Also include ipa-param-manipulation.h
7167 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7170 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7171 Alan Hayward <alan.hayward@arm.com>
7172 David Sherwood <david.sherwood@arm.com>
7174 * doc/sourcebuild.texi (vect_masked_store): Document.
7176 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7177 Alan Hayward <alan.hayward@arm.com>
7178 David Sherwood <david.sherwood@arm.com>
7180 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7182 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7183 Alan Hayward <alan.hayward@arm.com>
7184 David Sherwood <david.sherwood@arm.com>
7186 * doc/sourcebuild.texi (vect_variable_length): Document.
7188 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7189 Alan Hayward <alan.hayward@arm.com>
7190 David Sherwood <david.sherwood@arm.com>
7192 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7194 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7195 Alan Hayward <alan.hayward@arm.com>
7196 David Sherwood <david.sherwood@arm.com>
7198 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7200 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7201 Alan Hayward <alan.hayward@arm.com>
7202 David Sherwood <david.sherwood@arm.com>
7204 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7205 previously undocumented selectors.
7206 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7208 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7210 * doc/rtl.texi (const_vector): Say that elements can be
7211 const_wide_ints too.
7212 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7213 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7214 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7215 * optabs.c (expand_vector_broadcast): Likewise.
7217 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7218 Alan Hayward <alan.hayward@arm.com>
7219 David Sherwood <david.sherwood@arm.com>
7221 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7222 scaled index even if the unscaled address was invalid.
7223 Don't increase the complexity of using a scale in that case.
7225 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7226 Alan Hayward <alan.hayward@arm.com>
7227 David Sherwood <david.sherwood@arm.com>
7229 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7230 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7231 UNITS_PER_WORD bytes.
7232 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7233 based on REGMODE_NATURAL_SIZE of the inner mode.
7234 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7235 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7236 * expmed.c (lowpart_bit_field_p): Divide the value up into
7237 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7238 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7239 whether something is likely to occupy more than one register.
7241 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7244 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7245 profile_count::adjust_for_ipa_scaling.
7246 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7247 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7249 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7251 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7253 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7254 (split_group): Add total_orig and total_new arguments, estimate the
7255 number of statements related to the store group without store merging
7256 and with store merging.
7257 (imm_store_chain_info::output_merged_store): Adjust split_group
7258 callers, punt if estimated number of statements with store merging
7259 is not smaller than estimated number of statements without it.
7261 (handled_load): Remove has_single_use checks.
7262 (pass_store_merging::process_store): Likewise.
7264 2017-11-09 Richard Biener <rguenther@suse.de>
7266 PR tree-optimization/82902
7267 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7269 2017-11-09 Martin Liska <mliska@suse.cz>
7272 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7275 2017-11-09 Martin Liska <mliska@suse.cz>
7277 PR tree-optimization/82669
7278 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7280 2017-11-09 Martin Liska <mliska@suse.cz>
7282 PR gcov-profile/48463
7283 * coverage.c (coverage_begin_function): Output also end locus
7284 of a function and information whether the function is
7286 * gcov-dump.c (tag_function): Parse and print the information.
7287 * gcov.c (INCLUDE_MAP): Add include.
7288 (INCLUDE_SET): Likewise.
7289 (struct line_info): Move earlier in the source file because
7290 of vector<line_info> in function_info structure.
7291 (line_info::line_info): Likewise.
7292 (line_info::has_block): Likewise.
7293 (struct source_info): Add new member index.
7294 (source_info::get_functions_at_location): New function.
7295 (function_info::group_line_p): New function.
7296 (output_intermediate_line): New function.
7297 (output_intermediate_file): Use the mentioned function.
7298 (struct function_start): New.
7299 (struct function_start_pair_hash): Likewise.
7300 (process_file): Add code that identifies group functions.
7301 Assign lines either to global or function scope.
7302 (generate_results): Skip artificial functions.
7303 (find_source): Assign index for each source file.
7304 (read_graph_file): Read new flag artificial and end_line.
7305 (add_line_counts): Assign it either to global of function scope.
7306 (accumulate_line_counts): Isolate core of the function to
7307 accumulate_line_info and call it for both function and global
7309 (accumulate_line_info): New function.
7310 (output_line_beginning): Fix GNU coding style.
7311 (print_source_line): New function.
7312 (output_line_details): Likewise.
7313 (output_function_details): Likewise.
7314 (output_lines): Iterate both source (global) scope and function
7316 (struct function_line_start_cmp): New class.
7317 * doc/gcov.texi: Reflect changes in documentation.
7319 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7322 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7323 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7324 and similarly for not instead of neg.
7326 2017-11-08 Andi Kleen <ak@linux.intel.com>
7328 * config/i386/i386.opt: Add -mforce-indirect-call.
7329 * config/i386/predicates.md: Check for flag_force_indirect_call.
7330 * doc/invoke.texi: Document -mforce-indirect-call
7332 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7334 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7336 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7337 config/riscv/riscv.c (predict.h): New include.
7338 (riscv_slow_unaligned_access_p): No longer static.
7339 (riscv_block_move_straight): Add require.
7340 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7342 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7345 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7346 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7347 nonimmediate_operand predicate for operand 1 instead of
7350 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7352 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7354 * config/aarch64/constraints.md (Uml): New constraint.
7355 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7358 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7360 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7361 of two vec_duplicates into a vec_concat.
7363 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7365 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7366 Simplify vec_merge of vec_duplicate and vec_concat.
7367 * config/aarch64/constraints.md (Utq): New constraint.
7368 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7371 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7373 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7374 Simplify vec_merge of vec_duplicate and const_vector.
7375 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7377 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7378 mode iterator. Update predicate on operand 1 to
7379 handle non-const_vec constants. Delete constraints.
7380 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7382 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7384 PR tree-optimization/78821
7385 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7387 (store_operand_info::store_operand_info): Initialize it to false.
7388 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7389 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7390 store in the group, and if chain_info is non-NULL, to ignore altogether
7392 (compatible_load_p): Fail if bit_not_p does not match.
7393 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7394 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7395 (pass_store_merging::process_store): Adjust
7396 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7397 call terminate_all_aliasing_chains newly when adding a store into
7398 a chain with non-NULL chain_info.
7400 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7402 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7404 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7406 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7408 (aarch64_layout_frame): Initialise emit_frame_chain.
7409 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7410 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7412 2017-11-08 Martin Liska <mliska@suse.cz>
7414 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7415 of gimple_call_internal_p.
7417 2017-11-07 Tom de Vries <tom@codesourcery.com>
7419 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7421 2017-11-07 Tom de Vries <tom@codesourcery.com>
7423 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7426 2017-11-08 Martin Liska <mliska@suse.cz>
7429 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7431 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7433 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7435 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7438 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7439 (*cmp<mode>_ccz_1): New insn with $k alternative.
7442 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7443 RTX_COMM_COMPARE as commutative as well.
7444 (ix86_binary_operator_ok): Formatting fix.
7445 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7446 *<code><mode>3<mask_name><round_saeonly_name>,
7447 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7448 *<s>mul<mode>3_highpart<mask_name>,
7449 *vec_widen_umult_even_v16si<mask_name>,
7450 *vec_widen_umult_even_v8si<mask_name>,
7451 *vec_widen_umult_even_v4si<mask_name>,
7452 *vec_widen_smult_even_v16si<mask_name>,
7453 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7454 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7455 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7456 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7457 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7458 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7459 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7460 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7461 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7462 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7463 ix86_binary_operator_ok. Formatting fixes.
7464 (*<plusminus_insn><mode>3<mask_name><round_name>,
7465 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7468 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7470 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7471 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7472 eq and ne if TARGET_ISEL.
7473 (cmp): New code_iterator.
7474 (UNS, UNSU_, UNSIK): New code_attrs.
7475 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7476 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7478 ("eq<mode>3"): ... this.
7479 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7481 ("ne<mode>3"): ... this.
7483 2017-11-07 Julia Koval <julia.koval@intel.com>
7486 * common/config/i386/i386-common.c
7487 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7488 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7489 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7490 * config/i386/i386.opt: Ditto.
7491 * config/i386/i386.c (ix86_target_string): Ditto.
7492 (ix86_option_override_internal): Ditto.
7493 (ix86_init_mpx_builtins): Move MPX to args2.
7494 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7495 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7496 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7497 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7498 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7499 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7500 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7502 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7505 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7506 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7507 (zero-extendsidi peephole2): Remove peephole.
7509 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7512 * stor-layout.c: Include attribs.h.
7513 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7514 explicit lookup of "aligned" attribute.
7516 2017-11-07 Andrew Waterman <andrew@sifive.com>
7518 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7519 (riscv_expand_block_move): Likewise.
7520 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7522 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7523 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7524 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7525 (riscv_adjust_block_mem): Likewise.
7526 (riscv_block_move_loop): Likewise.
7527 (riscv_expand_block_move): Likewise.
7528 * config/riscv/riscv.md (movmemsi): New pattern.
7530 2017-11-07 Michael Clark <michaeljclark@mac.com>
7532 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7533 (MUSL_DYNAMIC_LINKER): Likewise.
7535 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7537 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7538 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7541 2017-11-07 Richard Biener <rguenther@suse.de>
7543 * match.pd: Fix build.
7545 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7546 Jackson Woodruff <jackson.woodruff@arm.com>
7548 PR tree-optimization/71026
7549 * match.pd: Canonicalize negate in division.
7551 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7554 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7556 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7558 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7559 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7561 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7563 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7564 non-scalar integral types.
7565 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7566 (-(A-B), -(~A)): New transformations.
7568 2017-11-07 Tom de Vries <tom@codesourcery.com>
7570 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7571 semicolon after "do {} while (0)".
7572 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7573 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7574 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7575 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7576 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7578 2017-11-07 Tom de Vries <tom@codesourcery.com>
7580 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7581 after "do {} while (0)".
7582 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7583 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7584 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7585 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7586 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7588 2017-11-07 Tom de Vries <tom@codesourcery.com>
7591 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7593 (arm_rtx_costs_internal): Add missing semicolon after
7594 HANDLE_NARROW_SHIFT_ARITH call.
7596 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7598 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7599 disable isel if it was not set explicitly.
7601 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7603 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7604 (add_type_attribute) likewise.
7606 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7608 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7609 of struct ix86_frame.
7610 (ix86_initial_elimination_offset): Likewise.
7611 (ix86_expand_split_stack_prologue): Likewise.
7613 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7615 * tree-vrp.h (enum value_range_type): Update stale comment.
7617 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7618 Alan Hayward <alan.hayward@arm.com>
7619 David Sherwood <david.sherwood@arm.com>
7621 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7622 (aarch64_expand_vec_perm_const): Take the number of units too.
7623 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7624 (aarch64_expand_vec_perm_const): Likewise.
7625 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7626 (vec_perm<mode>): Update accordingly.
7628 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7629 Alan Hayward <alan.hayward@arm.com>
7630 David Sherwood <david.sherwood@arm.com>
7632 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7633 Take the number of units too.
7634 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7635 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7636 but check for a vector mode before rather than after the call.
7637 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7638 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7639 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7640 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7641 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7642 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7643 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7644 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7645 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7646 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7647 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7648 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7649 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7650 (aarch64_sqdmull2_n<mode>): Update accordingly.
7652 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7653 Alan Hayward <alan.hayward@arm.com>
7654 David Sherwood <david.sherwood@arm.com>
7656 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7657 the number of units too.
7658 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7659 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7660 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7661 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7662 (vec_store_lanesxi<mode>): Update accordingly.
7664 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7665 Alan Hayward <alan.hayward@arm.com>
7666 David Sherwood <david.sherwood@arm.com>
7668 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7669 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7670 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7671 of units rather than the mode.
7672 * config/aarch64/iterators.md (nunits): New mode attribute.
7673 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7674 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7675 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7676 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7677 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7678 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7679 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7680 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7681 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7682 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7683 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7684 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7685 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7686 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7687 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7688 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7689 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7690 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7691 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7692 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7693 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7694 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7695 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7696 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7697 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7698 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7699 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7700 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7701 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7702 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7703 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7704 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7705 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7706 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7707 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7708 (aarch64_simd_vec_setv2di): Likewise.
7710 2017-11-06 Carl Love <cel@us.ibm.com>
7712 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7714 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7715 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7717 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7719 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7720 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7721 Add power 8 macro expansions.
7722 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7723 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7724 power 8 instructions. (VSX_XXBR): Add iterator.
7726 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7728 * config/arm/arm.md (predicable_short_it): Change default to "no",
7729 improve documentation, remove uses that are identical to the default.
7730 (enabled_for_depr_it): Rename to enabled_for_short_it.
7731 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7732 Remove default uses.
7733 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7734 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7735 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7736 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7738 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7741 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7742 float128 helper macros, which are no longer used after deleting
7743 the old 'q' built-in functions, and moving the round to odd
7744 built-in functions to being special built-in functions.
7745 (BU_FLOAT128_2): Likewise.
7746 (BU_FLOAT128_1_HW): Likewise.
7747 (BU_FLOAT128_2_HW): Likewise.
7748 (BU_FLOAT128_3_HW): Likewise.
7749 (FABSQ): Delete old 'q' built-in functions.
7750 (COPYSIGNQ): Likewise.
7751 (SQRTF128_ODD): Move round to odd built-in functions to be
7752 special built-in functions, so that we can handle
7753 -mabi=ieeelongdouble.
7754 (TRUNCF128_ODD): Likewise.
7755 (ADDF128_ODD): Likewise.
7756 (SUBF128_ODD): Likewise.
7757 (MULF128_ODD): Likewise.
7758 (DIVF128_ODD): Likewise.
7759 (FMAF128_ODD): Likewise.
7760 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7761 built-in names to 'f128'.
7762 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7763 old 'q' built-in functions, as the machine independent code for
7764 'f128' built-in functions handles this.
7765 (rs6000_expand_builtin): Add expansion for float128 round to odd
7766 functions, keying off on -mabi=ieeelongdouble of whether to use
7767 the KFmode or TFmode variant.
7768 (rs6000_init_builtins): Initialize the _Float128 round to odd
7770 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7771 _Float128 'q' built-in functions are now mapped into the new
7772 'f128' built-in functions.
7774 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7776 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7777 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7779 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7782 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7783 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7784 * toplev.c: Include "ipa-fnsummary.h".
7785 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7787 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7789 PR tree-optimization/82838
7790 * gimple-ssa-store-merging.c
7791 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7792 on a separate gimple_seq which is then appended to seq.
7794 2017-11-06 Jeff Law <law@redhat.com>
7797 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7798 (get_probe_interval): New functions.
7799 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7800 (ix86_adjust_stack_and_probe): Likewise.
7801 (output_adjust_stack_and_probe): Likewise.
7802 (ix86_emit_probe_stack_range): Likewise.
7803 (ix86_expand_prologue): Likewise.
7805 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7807 PR tree-optimization/82816
7808 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7809 if the modes of the two types are the same.
7810 (convert_plusminus_to_widen): Likewise.
7812 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7814 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7816 (usadv16qi): New define_expand.
7817 (usadv8hi): New define_expand.
7819 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7822 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7824 (inline_transform): Likewise.
7825 * predict.c (counts_to_freqs): Remove useless conditional.
7826 * profile-count.h (profile_count::apply_scale): Move sanity check.
7827 * tree-inline.c (copy_bb): Always scale.
7828 (copy_cfg_body): Likewise.
7830 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7833 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7835 (*cmp_ite0): Add enabled_for_depr_it attribute.
7836 (*cmp_ite1): Likewise.
7838 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7840 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7843 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7845 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7846 return the value by pointer.
7847 (extract_range_from_multiplicative_op_1): Update accordingly.
7848 Return as soon as an operation fails.
7850 2017-11-05 Tom de Vries <tom@codesourcery.com>
7853 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7854 (DEF_SANITIZER_BUILTIN): ... here.
7855 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7856 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7858 2017-11-05 Tom de Vries <tom@codesourcery.com>
7861 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7863 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7864 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7865 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7867 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7868 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7869 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7871 2017-11-05 Tom de Vries <tom@codesourcery.com>
7874 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7877 2017-11-04 Michael Clark <michaeljclark@mac.com>
7879 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7880 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7882 (*addsi3_extended): Likewise.
7883 (*addsi3_extended2): Likewise.
7884 (<optab>si3): Likewise.
7885 (<optab>di3): Likewise.
7886 (<optab><mode>3): Likewise.
7887 (<*optabe>si3_internal): Likewise.
7888 (zero_extendqi<SUPERQI:mode>2): Likewise.
7889 (*add<mode>hi3): Likewise.
7890 (*xor<mode>hi3): Likewise.
7891 (<optab>di3): Likewise.
7892 (*<optab>si3_extend): Likewise.
7893 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7894 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7895 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7897 2017-11-04 Andrew Waterman <andrew@sifive.com>
7899 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7900 TARGET_STRICT_ALIGN based upon -mtune argument.
7902 2017-11-04 Andrew Waterman <andrew@sifive.com>
7904 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7906 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7908 * config/i386/i386.c (choose_basereg): Use optional scratch
7909 register and add assertion.
7910 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7911 needed, and don't allocate stack.
7912 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7913 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7914 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7916 2017-11-03 Jeff Law <law@redhat.com>
7918 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7919 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7920 to probe at the start of a noreturn function.
7922 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7924 PR tree-optimization/78821
7925 * gimple-ssa-store-merging.c: Update the file comment.
7926 (MAX_STORE_ALIAS_CHECKS): Define.
7927 (struct store_operand_info): New type.
7928 (store_operand_info::store_operand_info): New constructor.
7929 (struct store_immediate_info): Add rhs_code and ops data members.
7930 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7931 and op1r arguments to the ctor, initialize corresponding data members.
7932 (struct merged_store_group): Add load_align_base and load_align
7934 (merged_store_group::merged_store_group): Initialize them.
7935 (merged_store_group::do_merge): Update them.
7936 (merged_store_group::apply_stores): Pick the constant for
7937 encode_tree_to_bitpos from one of the two operands, or skip
7938 encode_tree_to_bitpos if neither operand is a constant.
7939 (class pass_store_merging): Add process_store method decl. Remove
7940 bool argument from terminate_all_aliasing_chains method decl.
7941 (pass_store_merging::terminate_all_aliasing_chains): Remove
7942 var_offset_p argument and corresponding handling.
7943 (stmts_may_clobber_ref_p): New function.
7944 (compatible_load_p): New function.
7945 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7946 if there is overlap and rhs_code is not INTEGER_CST. For
7947 non-overlapping stores terminate group if rhs is not mergeable.
7948 (get_alias_type_for_stmts): Change first argument from
7949 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7950 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7951 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7953 (get_location_for_stmts): Change first argument from
7954 auto_vec<gimple *> & to vec<gimple *> &.
7955 (struct split_store): Remove orig_stmts data member, add orig_stores.
7956 (split_store::split_store): Create orig_stores rather than orig_stmts.
7957 (find_constituent_stmts): Renamed to ...
7958 (find_constituent_stores): ... this. Change second argument from
7959 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7960 to info structures rather than the statements.
7961 (split_group): Rename ALLOW_UNALIGNED argument to
7962 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7963 it. Adjust find_constituent_stores caller.
7964 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7965 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7966 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7967 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7968 (mem_valid_for_store_merging): New function.
7969 (handled_load): New function.
7970 (pass_store_merging::process_store): New method.
7971 (pass_store_merging::execute): Use process_store method. Adjust
7972 terminate_all_aliasing_chains caller.
7974 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7976 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7977 Return true for more constants, symbols and label references.
7978 (aarch64_valid_floating_const): Remove unused function.
7980 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7983 * config/aarch64/aarch64.c (aarch64_layout_frame):
7984 Undo forcing of LR at bottom of frame.
7986 2017-11-03 Jeff Law <law@redhat.com>
7989 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7990 for int_registers_saved.
7992 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7993 extracted from tree-ssa-dom.c.
7994 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7995 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7996 (record_equivalences_from_incoming_edge): Add additional argument
7997 to single_pred_edge_ignoring_loop_edges call.
7998 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7999 (uncprop_dom_walker::before_dom_children): Add additional argument
8000 to single_pred_edge_ignoring_loop_edges call.
8001 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8002 single_pred_edge_ignoring_loop_edges rather than open coding.
8003 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8005 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8007 * match.pd (-(-A)): Rewrite.
8009 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8011 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8012 (rs6000_emit_int_cmove): New declaration.
8013 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8014 (rs6000_emit_sISEL): Delete.
8015 (rs6000_emit_int_cmove): Make non-static.
8016 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8017 instead of rs6000_emit_sISEL.
8019 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8021 * asan.c (create_cond_insert_point): Maintain profile.
8022 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8024 * basic-block.h (struct basic_block_def): Remove frequency.
8025 (EDGE_FREQUENCY): Use to_frequency
8026 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8028 (find_traces): Update to use to_frequency.
8029 (find_traces_1_round): Likewise; use only IPA counts.
8030 (bb_to_key): Likewise.
8031 (connect_traces): Use IPA counts only.
8032 (copy_bb_p): Update to use to_frequency.
8033 (fix_up_crossing_landing_pad): Likewise.
8034 (sanitize_hot_paths): Likewise.
8035 * bt-load.c (basic_block_freq): Likewise.
8036 * cfg.c (init_flow): Set count_max to uninitialized.
8037 (check_bb_profile): Remove frequencies; check counts.
8038 (dump_bb_info): Do not dump frequencies.
8039 (update_bb_profile_for_threading): Update counts only.
8040 (scale_bbs_frequencies_int): Likewise.
8041 (MAX_SAFE_MULTIPLIER): Remove.
8042 (scale_bbs_frequencies_gcov_type): Update counts only.
8043 (scale_bbs_frequencies_profile_count): Update counts only.
8044 (scale_bbs_frequencies): Update counts only.
8045 * cfg.h (struct control_flow_graph): Add count-max.
8046 (update_bb_profile_for_threading): Update prototype.
8047 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8048 (find_many_sub_basic_blocks): Likewise.
8049 * cfgcleanup.c (try_forward_edges): Likewise.
8050 (try_crossjump_to_edge): Likewise.
8051 * cfgexpand.c (expand_gimple_cond): Likewise.
8052 (expand_gimple_tailcall): Likewise.
8053 (construct_init_block): Likewise.
8054 (construct_exit_block): Likewise.
8055 * cfghooks.c (verify_flow_info): Check consistency of counts.
8056 (dump_bb_for_graph): Do not dump frequencies.
8057 (split_block_1): Do not update frequencies.
8058 (split_edge): Do not update frequencies.
8059 (make_forwarder_block): Do not update frequencies.
8060 (duplicate_block): Do not update frequencies.
8061 (account_profile_record): Do not update frequencies.
8062 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8063 for global heuristics.
8064 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8065 (expected_loop_iterations_unbounded): Use counts only.
8066 * cfgloopmanip.c (scale_loop_profile): Simplify.
8067 (create_empty_loop_on_edge): Simplify
8069 (duplicate_loop_to_header_edge): Simplify
8070 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8071 (update_br_prob_note): Take care of removing note when profile
8073 (relink_block_chain): Do not dump frequency.
8074 (rtl_account_profile_record): Use to_frequency.
8075 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8076 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8077 (cgraph_update_edges_for_call_stmt_node): Likewise.
8078 (cgraph_edge::verify_count_and_frequency): Update.
8079 (cgraph_node::verify_node): Temporarily disable frequency verification.
8080 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8081 to_cgraph_frequency.
8082 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8083 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8085 (cgraph_node::expand_thunk): Update profile.
8086 * except.c (dw2_build_landing_pads): Do not update frequency.
8087 * final.c (compute_alignments): Use to_frequency.
8088 (dump_basic_block_info): Do not dump frequency.
8089 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8090 (dump_gimple_bb_header): Do not dump frequency.
8091 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8093 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8094 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8095 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8096 (init_before_recovery): Do not update frequency.
8097 (sched_create_recovery_edges): Do not update frequency.
8098 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8099 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8100 (ipa_cp_c_finalize): Set max_count to uninitialized.
8101 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8102 (param_change_prob): Use counts.
8103 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8105 * ipa-split.c (consider_split): Use to_frequency.
8106 (split_function): Use to_frequency.
8107 * ira-build.c (loop_compare_func): Likewise.
8108 (mark_loops_for_removal): Likewise.
8109 (mark_all_loops_for_removal): Likewise.
8110 * loop-doloop.c (doloop_modify): Do not update frequency.
8111 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8113 * lto-streamer-in.c (input_function): Update count_max.
8114 * omp-expand.c (expand_omp_taskreg): Update count_max.
8115 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8116 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8117 (maybe_hot_count_p): Use ipa counts only.
8118 (maybe_hot_bb_p): Simplify.
8119 (maybe_hot_edge_p): Simplify.
8120 (probably_never_executed): Do not take frequency argument.
8121 (probably_never_executed_bb_p): Do not pass frequency.
8122 (probably_never_executed_edge_p): Likewise.
8123 (combine_predictions_for_bb): Check that profile is nonzero.
8124 (propagate_freq): Do not set frequency.
8125 (drop_profile): Simplify.
8126 (counts_to_freqs): Simplify.
8127 (expensive_function_p): Use to_frequency.
8128 (propagate_unlikely_bbs_forward): Simplify.
8129 (determine_unlikely_bbs): Simplify.
8130 (estimate_bb_frequencies): Add hack to silence graphite issues.
8131 (compute_function_frequency): Use ipa counts.
8132 (pass_profile::execute): Update.
8133 (rebuild_frequencies): Use counts only.
8134 (force_edge_cold): Use counts only.
8135 * profile-count.c (profile_count::dump): Dump new count types.
8136 (profile_count::differs_from_p): Check compatiblity.
8137 (profile_count::to_frequency): New function.
8138 (profile_count::to_cgraph_frequency): New function.
8139 * profile-count.h (struct function): Declare.
8140 (enum profile_quality): Add profile_guessed_local and
8141 profile_guessed_global0.
8142 (class profile_proability): Decrease number of bits to 29;
8143 update from_reg_br_prob_note and to_reg_br_prob_note.
8144 (class profile_count: Update comment; decrease number of bits
8145 to 61. Check compatibility.
8146 (profile_count::compatible_p): New private member function.
8147 (profile_count::ipa_p): New member function.
8148 (profile_count::operator<): Handle global zero correctly.
8149 (profile_count::operator>): Handle global zero correctly.
8150 (profile_count::operator<=): Handle global zero correctly.
8151 (profile_count::operator>=): Handle global zero correctly.
8152 (profile_count::nonzero_p): New member function.
8153 (profile_count::force_nonzero): New member function.
8154 (profile_count::max): New member function.
8155 (profile_count::apply_scale): Handle IPA scalling.
8156 (profile_count::guessed_local): New member function.
8157 (profile_count::global0): New member function.
8158 (profile_count::ipa): New member function.
8159 (profile_count::to_frequency): Declare.
8160 (profile_count::to_cgraph_frequency): Declare.
8161 * profile.c (OVERLAP_BASE): Delete.
8162 (compute_frequency_overlap): Delete.
8163 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8164 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8165 * sched-ebb.c (rank): Use counts only.
8166 * shrink-wrap.c (handle_simple_exit): Use counts only.
8167 (try_shrink_wrapping): Use counts only.
8168 (place_prologue_for_one_component): Use counts only.
8169 * tracer.c (find_best_predecessor): Use to_frequency.
8170 (find_trace): Use to_frequency.
8171 (tail_duplicate): Use to_frequency.
8172 * trans-mem.c (expand_transaction): Do not update frequency.
8173 * tree-call-cdce.c: Do not update frequency.
8174 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8175 (gimple_merge_blocks): Likewise.
8176 (gimple_split_edge): Likewise.
8177 (gimple_duplicate_sese_region): Likewise.
8178 (gimple_duplicate_sese_tail): Likewise.
8179 (move_sese_region_to_fn): Likewise.
8180 (gimple_account_profile_record): Likewise.
8181 (insert_cond_bb): Likewise.
8182 * tree-complex.c (expand_complex_div_wide): Likewise.
8183 * tree-eh.c (lower_resx): Update profile.
8184 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8186 (initialize_cfun): Do not initialize frequencies
8187 (freqs_to_counts): Delete.
8188 (copy_cfg_body): Ignore count parameter.
8189 (copy_body): Update.
8190 (expand_call_inline): Update count_max.
8191 (optimize_inline_calls): Update count_max.
8192 (tree_function_versioning): Update count_max.
8193 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8194 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8196 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8197 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8198 (try_peel_loop): Likewise.
8199 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8201 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8202 (tree_transform_and_unroll_loop): Do not use frequencies
8203 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8204 Use reliable prediction only.
8205 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8206 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8207 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8208 probability scaling.
8209 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8210 not update frequency
8211 (any_remaining_duplicated_blocks): Likewise.
8212 (update_profile): Likewise.
8213 (estimated_freqs_path): Delete.
8214 (freqs_to_counts_path): Delete.
8215 (clear_counts_path): Delete.
8216 (ssa_fix_duplicate_block_edges): Likewise.
8217 (duplicate_thread_path): Likewise.
8218 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8219 * tree-tailcall.c (decrease_profile): Do not update frequency.
8220 (eliminate_tail_call): Likewise.
8221 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8222 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8223 (optimize_mask_stores): Likewise.
8224 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8225 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8226 (ubsan_expand_ptr_ifn): Update profile.
8227 * value-prof.c (gimple_ic): Simplify.
8228 * value-prof.h (gimple_ic): Update prototype.
8229 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8230 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8232 (want_inline_self_recursive_call_p): Likewise.
8233 (resolve_noninline_speculation): Only cummulate defined counts.
8234 (inline_small_functions): Use nonzero_p.
8235 (ipa_inline): Do not access freed node.
8237 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8239 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8240 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8242 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8244 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8245 non-legitimate address.
8247 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8249 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8250 (*lt0_<mode>di, *lt0_<mode>si): New.
8252 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8254 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8255 TARGET_PAIRED_FLOAT.
8257 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8258 Jim Wilson <jim.wilson@linaro.org>
8260 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8261 * config/aarch64/aarch64-tune.md: Regenerated.
8262 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8263 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8265 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8267 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8268 (arc_expand_prologue): Restore blink for millicode.
8269 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8271 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8274 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8275 gen_vec_duplicate after forcing the scalar into a register.
8277 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8279 * combine (try_combine): Print the insns input to try_combine to the
8282 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8285 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8286 Remove second argument from aarch64_process_target_attr call.
8287 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8289 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8291 (aarch64_handle_attr_arch): Remove second argument.
8292 (aarch64_handle_attr_cpu): Ditto.
8293 (aarch64_handle_attr_tune): Ditto.
8294 (aarch64_handle_attr_isa_flags): Ditto.
8295 (aarch64_process_one_target_attr): Ditto.
8296 (aarch64_process_target_attr): Ditto.
8297 (aarch64_option_valid_attribute_p): Remove second argument.
8298 on aarch64_process_target_attr call.
8300 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8302 * diagnostic.c: Include "selftest-diagnostic.h".
8303 (selftest::assert_location_text): New function.
8304 (selftest::test_diagnostic_get_location_text): New function.
8305 (selftest::diagnostic_c_tests): Call it.
8307 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8309 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8310 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8311 (class selftest::test_diagnostic_context): Move to...
8312 * selftest-diagnostic.c: New file.
8313 * selftest-diagnostic.h: New file.
8315 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8317 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8318 offset range for FT32B.
8319 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8320 * config/ft32/ft32.md: Add TARGET_NOPM.
8321 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8322 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8324 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8326 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8328 2017-11-02 Jeff Law <law@redhat.com>
8330 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8331 virtual keyword on FINAL OVERRIDE members.
8333 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8334 virtuals together. Add virtual destructor.
8335 (substitute_and_fold_engine): Similarly.
8337 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8339 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8341 2017-11-02 Richard Biener <rguenther@suse.de>
8343 PR tree-optimization/82795
8344 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8346 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8348 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8350 (gcc_GAS_CHECK_FEATURE): Remove.
8351 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8352 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8353 * configure: Regenerate.
8355 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8357 * config/arc/arc.c (hwloop_optimize): Account for empty
8360 2017-11-02 Richard Biener <rguenther@suse.de>
8363 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8364 Truncate ARRAY_REF index and element size.
8366 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8368 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8370 2017-11-01 Jeff Law <law@redhat.com>
8372 * tree-ssa-ccp.c (ccp_folder): New class derived from
8373 substitute_and_fold_engine.
8374 (ccp_folder::get_value): New member function.
8375 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8376 (ccp_fold_stmt): Remove prototype.
8377 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8378 * tree-ssa-copy.c (copy_folder): New class derived from
8379 substitute_and_fold_engine.
8380 (copy_folder::get_value): Renamed from get_value.
8381 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8382 * tree-vrp.c (vrp_folder): New class derived from
8383 substitute_and_fold_engine.
8384 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8385 (vrp_folder::get_value): New member function.
8386 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8387 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8388 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8389 provide a class interface to folder/substitute routines.
8390 (ssa_prop_fold_stmt_fn): Remove typedef.
8391 (ssa_prop_get_value_fn): Likewise.
8392 (subsitute_and_fold): Remove prototype.
8393 (replace_uses_in): Likewise.
8394 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8395 Renamed from replace_uses_in. Call the virtual member function
8396 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8397 (substitute_and_fold_dom_walker): Remove initialization of
8398 data member entries for calbacks. Add substitute_and_fold_engine
8399 member and initialize it.
8400 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8401 member functions for get_value, replace_phi_args_in c
8402 replace_uses_in, and fold_stmt calls.
8403 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8404 substitute_and_fold. Remove assert. Update ctor call.
8406 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8407 (ssa_prop_visit_phi_fn): Likewise.
8408 (class ssa_propagation_engine): New class to provide an interface
8410 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8412 (ssa_prop_visit_phi): Likewise.
8413 (ssa_propagation_engine::simulate_stmt): Moved into class.
8414 Call visit_phi/visit_stmt from the class rather than via
8415 file scoped static variables.
8416 (ssa_propagation_engine::simulate_block): Moved into class.
8417 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8418 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8419 set file scoped statics for the visit_stmt/visit_phi callbacks.
8420 * tree-complex.c (complex_propagate): New class derived from
8421 ssa_propagation_engine.
8422 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8423 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8424 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8426 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8427 ssa_propagation_engine.
8428 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8429 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8430 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8431 * tree-ssa-copy.c (copy_prop): New class derived from
8432 ssa_propagation_engine.
8433 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8434 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8435 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8436 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8437 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8438 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8439 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8441 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8443 PR rtl-optimization/82778
8444 PR rtl-optimization/82597
8445 * compare-elim.c (struct comparison): Add in_a_setter field.
8446 (find_comparison_dom_walker::before_dom_children): Remove killed
8447 bitmap and df_simulate_find_defs call, instead walk the defs.
8448 Compute last_setter and initialize in_a_setter. Merge definitions
8449 with first initialization for a few variables.
8450 (try_validate_parallel): Use insn_invalid_p instead of
8451 recog_memoized. Return insn rather than just the pattern.
8452 (try_merge_compare): Fix up comment. Don't uselessly test if
8453 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8455 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8458 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8459 Alan Hayward <alan.hayward@arm.com>
8460 David Sherwood <david.sherwood@arm.com>
8462 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8463 aarch64_hard_regno_nregs to get the number of registers
8466 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8467 Alan Hayward <alan.hayward@arm.com>
8468 David Sherwood <david.sherwood@arm.com>
8470 * config/aarch64/constraints.md (Upl): Rename to...
8472 * config/aarch64/aarch64.md
8473 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8476 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8477 Alan Hayward <alan.hayward@arm.com>
8478 David Sherwood <david.sherwood@arm.com>
8480 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8481 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8484 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8485 Alan Hayward <alan.hayward@arm.com>
8486 David Sherwood <david.sherwood@arm.com>
8488 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8489 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8490 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8492 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8493 of a single element.
8494 * config/aarch64/iterators.md: Add a comment above the permute
8495 unspecs to say that they are generated directly by
8496 aarch64_expand_vec_perm_const.
8497 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8499 2017-11-01 Nathan Sidwell <nathan@acm.org>
8501 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8503 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8505 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8506 and medany code models, and describe what they do.
8508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8510 Revert accidental duplicate:
8512 * combine.c (can_change_dest_mode): Reject changes in
8513 REGMODE_NATURAL_SIZE.
8515 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8517 PR rtl-optimization/64682
8518 PR rtl-optimization/69567
8519 PR rtl-optimization/69737
8520 PR rtl-optimization/82683
8521 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8522 register mentioned in the note, drop the note, unless it came from I3,
8523 in which case it should go to I3 again.
8525 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8527 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8528 and return false if not.
8529 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8531 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8533 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8534 range is known to be empty.
8536 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8537 Alan Hayward <alan.hayward@arm.com>
8538 David Sherwood <david.sherwood@arm.com>
8540 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8541 and CONST_VECTOR_NUNITS instead of computing the number of units from
8542 the byte sizes of the vector and element.
8543 (simplify_binary_operation_1): Likewise.
8544 (simplify_const_binary_operation): Likewise.
8545 (simplify_ternary_operation): Likewise.
8547 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8548 Alan Hayward <alan.hayward@arm.com>
8549 David Sherwood <david.sherwood@arm.com>
8551 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8552 (int_mem_offset): ...this new function.
8553 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8554 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8555 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8558 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8559 Alan Hayward <alan.hayward@arm.com>
8560 David Sherwood <david.sherwood@arm.com>
8562 * lower-subreg.c (interesting_mode_p): New function.
8563 (compute_costs, find_decomposable_subregs, decompose_register)
8564 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8565 (resolve_clobber, dump_choices): Use it.
8567 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8568 Alan Hayward <alan.hayward@arm.com>
8569 David Sherwood <david.sherwood@arm.com>
8571 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8573 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8574 Alan Hayward <alan.hayward@arm.com>
8575 David Sherwood <david.sherwood@arm.com>
8577 * alias.c (find_base_value, find_base_term): Only process integer
8578 truncations. Check the precision rather than the size.
8580 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8581 Alan Hayward <alan.hayward@arm.com>
8582 David Sherwood <david.sherwood@arm.com>
8584 * machmode.h (is_narrower_int_mode): New function
8585 * optabs.c (expand_float, expand_fix): Use it.
8586 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8588 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8589 Alan Hayward <alan.hayward@arm.com>
8590 David Sherwood <david.sherwood@arm.com>
8592 * rtl.h (narrower_subreg_mode): New function.
8593 * ira-color.c (update_costs_from_allocno): Use it.
8595 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8596 Alan Hayward <alan.hayward@arm.com>
8597 David Sherwood <david.sherwood@arm.com>
8599 * optabs-query.h (convert_optab_p): New function, split out from...
8600 (convert_optab_handler): ...here.
8601 (widening_optab_handler): Delete.
8602 (find_widening_optab_handler): Remove permit_non_widening parameter.
8603 (find_widening_optab_handler_and_mode): Likewise. Provide an
8604 override that operates on mode class wrappers.
8605 * optabs-query.c (widening_optab_handler): Delete.
8606 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8607 parameter. Assert that the two modes are the same class and that
8608 the "from" mode is narrower than the "to" mode. Use
8609 convert_optab_handler instead of widening_optab_handler.
8610 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8611 instead of widening_optab_handler.
8612 * expr.c (expand_expr_real_2): Update calls to
8613 find_widening_optab_handler.
8614 * optabs.c (expand_widen_pattern_expr): Likewise.
8615 (expand_binop_directly): Take the insn_code as a parameter.
8616 (expand_binop): Only call find_widening_optab_handler for
8617 conversion optabs; use optab_handler otherwise. Update calls
8618 to find_widening_optab_handler and expand_binop_directly.
8619 Use convert_optab_handler instead of widening_optab_handler.
8620 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8621 find_widening_optab_handler and use scalar_mode rather than
8623 (convert_plusminus_to_widen): Likewise.
8625 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8626 Alan Hayward <alan.hayward@arm.com>
8627 David Sherwood <david.sherwood@arm.com>
8629 * machmode.h (fixed_size_mode): New class.
8630 * rtl.h (get_pool_mode): Return fixed_size_mode.
8631 * gengtype.c (main): Add fixed_size_mode.
8632 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8633 (get_raw_arg_mode): Likewise.
8634 * doc/tm.texi: Regenerate.
8635 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8636 * targhooks.c (default_get_reg_raw_mode): Likewise.
8637 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8638 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8639 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8640 (msp430_get_raw_result_mode): Likewise.
8641 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8642 * dbxout.c (dbxout_parms): Require fixed-size modes.
8643 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8644 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8645 * omp-low.c (lower_oacc_reductions): Likewise.
8646 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8647 (simplify_subreg): Update accordingly.
8648 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8649 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8650 that aren't fixed-size.
8651 (get_pool_mode): Return a fixed_size_mode.
8652 (output_constant_pool_2): Take a fixed_size_mode.
8654 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8655 Alan Hayward <alan.hayward@arm.com>
8656 David Sherwood <david.sherwood@arm.com>
8658 * doc/rtl.texi (vec_series): Document.
8659 (const): Say that the operand can be a vec_series.
8660 * rtl.def (VEC_SERIES): New rtx code.
8661 * rtl.h (const_vec_series_p_1): Declare.
8662 (const_vec_series_p): New function.
8663 * emit-rtl.h (gen_const_vec_series): Declare.
8664 (gen_vec_series): Likewise.
8665 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8666 (gen_vec_series): Likewise.
8667 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8668 * simplify-rtx.c (simplify_unary_operation): Handle negations
8670 (simplify_binary_operation_series): New function.
8671 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8672 (test_vector_ops_series): New function.
8673 (test_vector_ops): Call it.
8674 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8675 gen_const_vec_series.
8676 (altivec_lvsr): Likewise.
8677 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8680 Alan Hayward <alan.hayward@arm.com>
8681 David Sherwood <david.sherwood@arm.com>
8683 * doc/rtl.texi (const): Update description of address constants.
8684 Say that vector constants are allowed too.
8685 * common.md (E, F): Use CONSTANT_P instead of checking for
8687 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8688 checking for CONST_VECTOR.
8689 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8691 * expr.c (expand_expr_real_2): Check for vector modes instead
8692 of checking for CONST_VECTOR.
8693 * rtl.h (const_vec_p): New function.
8694 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8695 (unwrap_const_vec_duplicate): Handle them here too.
8697 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8698 David Malcolm <dmalcolm@redhat.com>
8699 Alan Hayward <alan.hayward@arm.com>
8700 David Sherwood <david.sherwood@arm.com>
8702 * rtl.h (vec_duplicate_p): New function.
8703 * selftest-rtl.c (assert_rtx_eq_at): New function.
8704 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8705 (assert_rtx_eq_at): Declare.
8706 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8707 * selftest-run-tests.c (selftest::run_tests): Call it.
8708 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8709 (simplify_unary_operation_1): Recursively handle vector duplicates.
8710 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8712 (simplify_subreg): Handle subregs of vector duplicates.
8713 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8714 (selftest::simplify_rtx_c_tests): New functions.
8716 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8717 Alan Hayward <alan.hayward@arm.com>
8718 David Sherwood <david.sherwood@arm.com>
8720 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8721 (gen_vec_duplicate): Likewise.
8722 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8724 (gen_const_vector): ...here.
8725 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8726 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8727 whose elements are all equal.
8728 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8729 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8730 (simplify_relational_operation): Likewise.
8731 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8733 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8734 (aarch64_expand_vector_init): Likewise.
8735 * config/arm/arm.c (neon_vdup_constant): Likewise.
8736 (neon_expand_vector_init): Likewise.
8737 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8738 (arm_block_set_unaligned_vect): Likewise.
8739 (arm_block_set_aligned_vect): Likewise.
8740 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8741 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8742 (expand_vec_perm_even_odd_pack): Likewise.
8743 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8744 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8745 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8746 gen_const_vec_duplicate.
8747 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8748 * config/mips/mips.c (mips_gen_const_int_vector): Use
8749 gen_const_vec_duplicate.
8750 (mips_expand_vector_init): Use CONST0_RTX.
8751 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8752 (define_split): Use gen_const_vec_duplicate.
8753 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8754 (define_split): Use gen_const_vec_duplicate.
8755 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8756 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8757 * config/spu/spu.c (spu_const): Likewise.
8759 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8760 Alan Hayward <alan.hayward@arm.com>
8761 David Sherwood <david.sherwood@arm.com>
8763 * combine.c (can_change_dest_mode): Reject changes in
8764 REGMODE_NATURAL_SIZE.
8766 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8768 * configure.ac (--enable-libssp): New.
8769 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8770 trying to determine target-specific default. Adjust indentation.
8771 * configure: Regenerated.
8772 * doc/install.texi (Configuration): Expand --disable-libssp
8775 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8777 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8780 2017-10-31 Martin Jambor <mjambor@suse.cz>
8783 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8785 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8787 * auto-profile.c (autofdo_source_profile::read): Use
8788 UNKNOWN_LOCATION rather than 0.
8789 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8790 (warning_at): ...this overload.
8791 (warning_at_rich_loc_n): Rename to...
8792 (warning_n): ...this overload.
8793 (error_at_rich_loc): Rename to...
8794 (error_at): ...this overload.
8795 (pedwarn_at_rich_loc): Rename to...
8796 (pedwarn): ...this overload.
8797 (permerror_at_rich_loc): Rename to...
8798 (permerror): ...this overload.
8799 (inform_at_rich_loc): Rename to...
8800 (inform): ...this overload.
8801 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8802 (diagnostic_n_impl_richloc): Rename to...
8803 (diagnostic_n_impl): ...this rich_location *-based decl.
8804 (inform_at_rich_loc): Rename to...
8805 (inform): ...this, and add an assertion.
8806 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8807 (warning_at_rich_loc): Rename to...
8808 (warning_at): ...this, and add an assertion.
8809 (warning_at_rich_loc_n): Rename to...
8810 (warning_n): ...this, and add an assertion.
8811 (warning_n): Update location_t-based implementation for removal of
8812 location_t-based diagnostic_n_impl.
8813 (pedwarn_at_rich_loc): Rename to...
8814 (pedwarn): ...this, and add an assertion.
8815 (permerror_at_rich_loc): Rename to...
8816 (permerror): ...this, and add an assertion.
8817 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8818 (error_at_rich_loc): Rename to...
8819 (error_at): ...this, and add an assertion.
8820 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8821 (driver::do_spec_on_infiles): Likewise.
8822 * substring-locations.c (format_warning_va): Update for renaming
8823 of inform_at_rich_loc.
8825 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8827 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8828 _Float<N>X built-in functions so that the variant without the
8829 "__builtin_" prefix is only enabled for the GNU C and Objective C
8830 languages when they are in non-strict ANSI/ISO mode.
8831 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8832 * target.def (floatn_builtin_p): Add a target hook to control
8833 whether _Float<N> and _Float<N>X built-in functions without the
8834 "__builtin_" prefix are enabled, and return true for C and
8835 Objective C in the default hook. Include langhooks.h in
8837 * targhooks.h (default_floatn_builtin_p): Likewise.
8838 * targhooks.c (default_floatn_builtin_p): Likewise.
8839 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8840 floatn_builtin_p target hook.
8841 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8843 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8844 Eric Botcazou <ebotcazou@adacore.com>
8846 PR rtl-optimization/81803
8847 * lra-constraints.c (curr_insn_transform): Also reload the whole
8848 register for a strict subreg no wider than a word if this is for
8849 a WORD_REGISTER_OPERATIONS target.
8851 2017-10-31 Jason Merrill <jason@redhat.com>
8853 * gdbinit.in: Skip over inlines from timevar.h.
8855 2017-10-31 Martin Liska <mliska@suse.cz>
8857 * doc/gcov.texi: Document new option.
8858 * gcov.c (print_usage): Likewise print it.
8859 (process_args): Support the argument.
8860 (format_count): New function.
8861 (format_gcov): Use the function.
8863 2017-10-31 Martin Liska <mliska@suse.cz>
8865 * gcov.c (struct name_map): do not use typedef.
8866 Define operator== and operator<.
8867 (name_search): Remove.
8868 (name_sort): Remove.
8869 (main): Do not allocate names.
8870 (process_file): Add vertical space.
8871 (generate_results): Use std::find.
8872 (release_structures): Do not release memory.
8873 (find_source): Use std::find.
8875 2017-10-31 Martin Liska <mliska@suse.cz>
8877 * gcov.c (struct line_info): Remove it's typedef.
8878 (line_info::line_info): Add proper ctor.
8879 (line_info::has_block): Do not use a typedef.
8880 (struct source_info): Do not use typedef.
8881 (circuit): Likewise.
8882 (get_cycles_count): Likewise.
8883 (output_intermediate_file): Iterate via vector iterator.
8884 (add_line_counts): Use std::vector methods.
8885 (accumulate_line_counts): Likewise.
8886 (output_lines): Likewise.
8888 2017-10-31 Martin Liska <mliska@suse.cz>
8890 * gcov.c (struct source_info): Remove typedef.
8891 (source_info::source_info): Add proper ctor.
8892 (accumulate_line_counts): Use struct, not it's typedef.
8893 (output_gcov_file): Likewise.
8894 (output_lines): Likewise.
8895 (main): Do not allocate an array.
8896 (output_intermediate_file): Use size of vector container.
8897 (process_file): Resize the vector.
8898 (generate_results): Do not preallocate, use newly added vector
8900 (release_structures): Do not release sources.
8901 (find_source): Use vector methods.
8902 (add_line_counts): Do not use typedef.
8904 2017-10-31 Martin Liska <mliska@suse.cz>
8906 * doc/gcov.texi: Document that.
8907 * gcov.c (add_line_counts): Mark lines with a non-executed
8909 (output_line_beginning): Handle such lines.
8910 (output_lines): Pass new argument.
8911 (output_intermediate_file): Print it in intermediate format.
8913 2017-10-31 Martin Liska <mliska@suse.cz>
8915 * color-macros.h: New file.
8916 * diagnostic-color.c: Factor out color related to macros to
8918 * doc/gcov.texi: Document -k option.
8919 * gcov.c (INCLUDE_STRING): Include string.h.
8920 (print_usage): Add -k option.
8921 (process_args): Parse it.
8922 (pad_count_string): New function.
8923 (output_line_beginning): Likewise.
8924 (DEFAULT_LINE_START): New macro.
8925 (output_lines): Support color output.
8927 2017-10-31 Martin Liska <mliska@suse.cz>
8929 PR gcov-profile/82633
8930 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8931 their interaction with GCOV infrastructure.
8932 * configure.ac: Add -fkeep-{inline,static}-functions to
8934 * configure: Regenerate.
8936 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8939 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8941 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8944 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8945 into a register if it does not fit into an immediate offset field.
8947 2017-10-31 Olivier Hainque <hainque@adacore.com>
8949 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8951 2017-10-31 Julia Koval <julia.koval@intel.com>
8953 * config.gcc: Add gfniintrin.h.
8954 * config/i386/gfniintrin.h: New.
8955 * config/i386/i386-builtin-types.def
8956 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8957 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8958 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8959 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8960 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8961 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8962 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8963 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8964 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8965 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8966 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8967 * config/i386/immintrin.h: Include gfniintrin.h.
8968 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8970 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8972 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8974 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8976 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8977 (arm_ashldi3_1bit): Remove pattern.
8978 (ashrdi3): Remove shift by 1 expansion.
8979 (arm_ashrdi3_1bit): Remove pattern.
8980 (lshrdi3): Remove shift by 1 expansion.
8981 (arm_lshrdi3_1bit): Remove pattern.
8982 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8983 cost of ashldi3 by 1.
8984 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8985 (<shift>di3_neon): Likewise.
8987 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8989 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8990 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8991 and (*aarch64_simd_mov<VQ:mode>).
8992 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8993 pattern alternative.
8994 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8995 attributes to match pattern alternative.
8997 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8999 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9000 * config/rs6000/emmintrin.h: New file.
9001 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9003 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9005 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9006 * (movdi_vfp_cortexa8): Remove pattern.
9008 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9010 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9012 (Specific, alpha*-dec-osf5.1): Remove.
9013 (Specific, mips-sgi-irix5): Remove.
9014 (Specific, mips-sgi-irix6): Remove.
9016 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9019 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9020 arguments to clear_bit_region_be.
9022 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9024 * gimplify.c: Include memmodel.h.
9026 2017-10-30 Martin Jambor <mjambor@suse.cz>
9028 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9029 condition whether loop should be executed at all.
9031 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9033 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9034 gimple folding of vec_madd() intrinsics.
9035 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9036 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9037 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9039 2017-10-30 Richard Biener <rguenther@suse.de>
9041 PR tree-optimization/82762
9043 2017-10-23 Richard Biener <rguenther@suse.de>
9045 PR tree-optimization/82129
9047 2017-08-01 Richard Biener <rguenther@suse.de>
9049 PR tree-optimization/81181
9050 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9051 (compute_antic): ... end of iteration here.
9053 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9055 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9056 -std=iso9899:2017 and -std=gnu17.
9057 * doc/standards.texi (C Language): Document C17 support.
9058 * doc/cpp.texi (Overview): Mention -std=c17.
9059 (Standard Predefined Macros): Document C11 and C17 values of
9060 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9061 * doc/extend.texi (Inline): Do not list individual options for
9062 standards newer than C99.
9063 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9065 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9068 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9070 * asan.c (asan_finish_file): Align asan globals array by shadow
9073 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9076 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9077 (struct store_immediate_info): Add bitregion_start and bitregion_end
9079 (store_immediate_info::store_immediate_info): Add brs and bre
9080 arguments and initialize bitregion_{start,end} from those.
9081 (struct merged_store_group): Add bitregion_start, bitregion_end,
9082 align_base and mask fields. Drop unnecessary struct keyword from
9083 struct store_immediate_info. Add do_merge method.
9084 (clear_bit_region_be): Use memset instead of loop storing zeros.
9085 (merged_store_group::do_merge): New method.
9086 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9087 stores as long as the surrounding bitregions have no gaps.
9088 (merged_store_group::merge_overlapping): Use do_merge.
9089 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9090 is byte aligned, rather than requiring that start and width are
9091 byte aligned. Drop unnecessary struct keyword from
9092 struct store_immediate_info. Allocate and populate also mask array.
9093 Make start of the arrays relative to bitregion_start rather than
9094 start and size them according to bitregion_{end,start} difference.
9095 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9096 struct store_immediate_info.
9097 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9098 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9099 struct keyword from struct store_immediate_info.
9100 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9101 between stores as long as the surrounding bitregions have no gaps.
9103 (struct split_store): Add orig non-static data member.
9104 (split_store::split_store): Initialize orig to false.
9105 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9106 if there is exactly a single original stmt. Change stmts argument
9107 to pointer from reference, if NULL, don't push anything to it. Add
9108 first argument, use it to optimize skipping over orig stmts that
9109 are known to be before bitpos already. Simplify.
9110 (split_group): Return unsigned int count how many stores are or
9111 would be needed rather than a bool. Add allow_unaligned argument.
9112 Change split_stores argument from reference to pointer, if NULL,
9113 only do a dry run computing how many stores would be produced.
9114 Rewritten algorithm to use both alignment and misalign if
9115 !allow_unaligned and handle bitfield stores with gaps.
9116 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9117 from bitregion_start instead of start. Compute allow_unaligned
9118 here, if true, do 2 split_group dry runs to compute which one
9119 produces fewer stores and prefer aligned if equal. Punt if
9120 new count is bigger or equal than original before emitting any
9121 statements, rather than during that. Remove no longer needed
9122 new_ssa_names tracking. Replace num_stmts with
9123 split_stores.length (). Use 32-bit stack allocated entries
9124 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9125 if possible. Handle bitfields with gaps.
9126 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9127 Compute bitregion_{start,end} for the stores and construct
9128 store_immediate_info with that. Formatting fixes.
9130 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9133 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9134 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9136 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9138 * gimplify.c: Include tm_p.h.
9140 * common.opt (gcoff): Re-add as ignored option.
9141 (gcoff1, gcoff2, gcoff3): Likewise.
9143 * Makefile.in (OBJS): Delete sdbout.o.
9144 (GTFILES): Delete $(srcdir)/sdbout.c.
9145 * debug.h: Delete sdb_debug_hooks.
9146 * final.c: Delete sdbout.h include.
9147 (final_scan_insn): Delete SDB_DEBUG check.
9148 (rest_of_clean_state): Likewise.
9149 * output.h: Delete sdb_begin_function_line.
9152 * toplev.c: Delete sdbout.h include.
9153 (process_options): Delete SDB_DEBUG check.
9154 * tree-core.h (tree_type_common): Delete pointer field of
9156 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9157 TYPE_SYMTAB_POINTER.
9158 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9159 (TYPE_SYMTAB_IS_POINTER): Delete.
9160 (TYPE_SYMTAB_IS_DIE): Renumber.
9161 * xcoffout.c: Refer to former sdbout.c file.
9162 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9164 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9165 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9166 (Debugging Options): Delete -gcoff.
9167 (-gstabs): Delete SDB reference.
9169 (-gcoff@var{level}): Delete.
9170 * doc/passes.texi (Debugging information output): Delete SDB and
9171 sdbout.c references.
9172 * doc/tm.texi: Regenerate.
9173 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9174 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9176 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9177 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9178 (SDB_DEBUGGING_INFO): Delete.
9179 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9180 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9181 * target.def (output_source_filename): Delete COFF reference.
9183 * common.opt (gcoff): Delete.
9184 (gxcoff+): Update Negative chain.
9185 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9187 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9188 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9189 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9191 (expand_function_start): Change sdb reference to past tense.
9192 (expand_function_end): Change sdb reference to past tense.
9193 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9194 * opts.c (debug_type_names): Delete coff entry.
9195 (common_handle_option): Delete OPT_gcoff case.
9196 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9198 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9199 * config/cris/cris.h: Delete SDB reference in comment.
9200 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9201 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9202 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9203 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9205 (ix86_expand_prologue): Likewise.
9206 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9207 * config/ia64/ia64.h: Likewise.
9208 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9209 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9211 * config/mmix/mmix.h: Likewise.
9212 * config/nds32/nds32.c: Likewise.
9213 * config/stormy/storym16.h: Likewise.
9214 * config/visium/visium.h: Likewise.
9215 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9217 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9219 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9220 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9221 FRAME_POINTER_REGNUM point at high end of local var area.
9223 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9225 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9226 Move comment around. Do not reset best_edge for a copiable
9227 destination if the copy would cause a partition change.
9228 (better_edge_p): Remove redundant check.
9230 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9232 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9234 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9236 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9237 for math functions that have _Float<N> and _Float<N>X variants.
9238 (mathfn_built_in_2): Add support for math functions that have
9239 _Float<N> and _Float<N>X variants.
9240 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9241 (expand_builtin_mathfn_ternary): Add support for fma with
9242 _Float<N> and _Float<N>X variants.
9243 (expand_builtin): Likewise.
9244 (fold_builtin_3): Likewise.
9245 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9246 create math function _Float<N> and _Float<N>X variants as external
9248 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9249 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9250 the __builtin_ prefix and if not strict ansi, without the prefix.
9251 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9252 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9253 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9254 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9255 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9256 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9257 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9258 function signatures for fma _Float<N> and _Float<N>X variants.
9259 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9260 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9261 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9262 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9263 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9264 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9265 * gencfn-macros.c (print_case_cfn): Add support for math functions
9266 that have _Float<N> and _Float<N>X variants.
9267 (print_define_operator_list): Likewise.
9268 (fltfn_suffixes): Likewise.
9270 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9271 for math functions that have _Float<N> and _Float<N>X variants.
9272 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9273 and _Float<N>X variants.
9274 (COPYSIGN): Likewise.
9277 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9278 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9280 (integer_valued_read_call_p): Likewise.
9281 * fold-const-call.c (fold_const_call_ss): Likewise.
9282 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9283 _Float<N> and _Float<N>X variants.
9284 (fold_const_call_ssss): Add support for fma _Float<N> and
9285 _Float<N>X variants.
9286 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9287 support for copysign and fma _Float<N> and _Float<N>X variants.
9288 (backprop::process_builtin_call_use): Likewise.
9289 * tree-call-cdce.c (can_test_argument_range); Add support for
9290 sqrt _Float<N> and _Float<N>X variants.
9291 (edom_only_function): Likewise.
9292 (get_no_error_domain): Likewise.
9293 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9294 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9295 copysign _Float<N> and _Float<N>X variants.
9296 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9297 handled by machine independent code.
9298 (FMAF128): Likewise.
9299 * doc/cpp.texi (Common Predefined Macros): Document defining
9300 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9301 fma _Float<N> and _Float<N>X variants.
9303 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9306 * config/i386/i386-modes.def (CCFPU): Remove definition.
9307 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9308 (ix86_cc_modes_compatible): Ditto.
9309 (ix86_expand_carry_flag_compare): Ditto.
9310 (ix86_expand_int_movcc): Ditto.
9311 (ix86_expand_int_addcc): Ditto.
9312 (ix86_reverse_condition): Ditto.
9313 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9314 Return true/false for unordered/ordered fp comparisons.
9315 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9316 (ix86_prepare_fp_compare_args): Update for rename.
9317 (ix86_expand_fp_compare): Update for rename. Generate unordered
9318 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9319 (ix86_expand_sse_compare_and_jump): Ditto.
9320 * config/i386/predicates.md (fcmov_comparison_operator):
9321 Remove CCFPU mode handling.
9322 (ix86_comparison_operator): Ditto.
9323 (ix86_carry_flag_operator): Ditto.
9324 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9325 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9326 (*cmpu<mode>_cc_i387): Ditto.
9327 (FPCMP): Remove mode iterator.
9328 (unord): Remove mode attribute.
9329 (unord_subst): New define_subst transformation
9330 (unord): New define_subst attribute.
9332 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9333 (*cmpi<unord>xf_i387): Ditto.
9334 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9335 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9336 using unord_subst transformation.
9337 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9338 (round_saeonly): Also handle CCFP mode.
9339 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9340 Remove UNSPEC_SAHF unspec handling.
9342 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9344 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9346 2017-10-27 Jeff Law <law@redhat.com>
9348 * gimple-ssa-sprintf.c: Include domwalk.h.
9349 (class sprintf_dom_walker): New class, derived from dom_walker.
9350 (sprintf_dom_walker::before_dom_children): New function.
9351 (struct call_info): Moved into sprintf_dom_walker class
9352 (compute_formath_length, handle_gimple_call): Likewise.
9353 (sprintf_length::execute): Call the dominator walker rather
9354 than walking the statements.
9356 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9357 gimple statement locations.
9358 (check_array_bounds): Corresponding changes. Get the statement's
9359 location directly from wi->stmt.
9361 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9364 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9366 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9368 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9369 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9371 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9374 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9375 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9376 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9377 maybe_get_pool_constant.
9378 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9381 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9383 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9384 2.26 caveat. Update gas and gld versions.
9385 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9388 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9390 * cgraph.h (set_malloc_flag): Declare.
9391 * cgraph.c (set_malloc_flag_1): New function.
9392 (set_malloc_flag): Likewise.
9393 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9394 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9396 (read_ipa_call_summary): Add support for reading is_return_callee.
9397 (write_ipa_call_summary): Stream is_return_callee.
9398 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9399 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9400 ipa-prop.h, ipa-fnsummary.h.
9401 (pure_const_names): Change to static.
9402 (malloc_state_e): Define.
9403 (malloc_state_names): Define.
9404 (funct_state_d): Add field malloc_state.
9405 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9406 (check_retval_uses): New function.
9407 (malloc_candidate_p): Likewise.
9408 (analyze_function): Add support for malloc attribute.
9409 (pure_const_write_summary): Stream malloc_state.
9410 (pure_const_read_summary): Add support for reading malloc_state.
9411 (dump_malloc_lattice): New function.
9412 (propagate_malloc): New function.
9413 (warn_function_malloc): New function.
9414 (ipa_pure_const::execute): Call propagate_malloc and
9415 ipa_free_fn_summary.
9416 (pass_local_pure_const::execute): Add support for malloc attribute.
9417 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9418 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9420 2017-10-27 Martin Liska <mliska@suse.cz>
9422 PR gcov-profile/82457
9423 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9424 for fork-like functions to be properly instrumented.
9426 2017-10-27 Richard Biener <rguenther@suse.de>
9429 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9430 info when we redirected EH.
9432 2017-10-26 Michael Collison <michael.collison@arm.com>
9434 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9436 (<optab>_trunchf<GPI:mode>2: New pattern.
9437 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9438 * config/aarch64/iterators.md (wv): New mode attribute.
9439 (vf, VF): New mode attributes.
9440 (vgp, VGP): New mode attributes.
9441 (s): Update attribute with SImode and DImode prefixes.
9443 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9445 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9446 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9447 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9448 (nios2_option_overide): Initialize it. Don't allow R0-relative
9449 addressing with PIC.
9450 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9451 (nios2_symbolic_constant_p): Likewise.
9452 (nios2_legitimate_address_p): Likewise.
9453 (nios2_r0rel_section_name_p): New.
9454 (nios2_symbol_ref_in_r0rel_data_p): New.
9455 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9456 (r0rel_constant_p): New.
9457 (nios2_print_operand_address): Handle r0rel_constant_p.
9458 (nios2_cdx_narrow_form_p): Likewise.
9459 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9460 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9461 (Nios II Options): Document -mr0rel-sec.
9463 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9465 * config/nios2/nios2.c: Include xregex.h.
9466 (nios2_gprel_sec_regex): New.
9467 (nios2_option_overide): Initialize it. Don't allow GP-relative
9468 addressing with PIC.
9469 (nios2_small_section_name_p): Check for regex match.
9470 * config/nios2/nios2.opt (mgprel-sec=): New option.
9471 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9472 (Nios II Options): Document -mgprel-sec.
9474 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9476 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9478 2017-10-26 Tom de Vries <tom@codesourcery.com>
9480 PR tree-optimization/82707
9481 * gimple.c (gimple_copy): Fix unsharing of
9482 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9484 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9486 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9487 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9488 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9489 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9490 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9491 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9492 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9493 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9494 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9495 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9496 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9497 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9498 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9499 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9500 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9501 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9502 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9504 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9506 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9508 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9509 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9510 warning to rs6000.c. Remove the Undocumented flag, since it has
9512 (-mabi=ibmlongdouble): Likewise.
9513 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9514 already set, set the default format for long double.
9515 (rs6000_debug_reg_global): Print whether long double is IBM or
9517 (rs6000_option_override_internal): Rework setting long double
9518 format. Only warn if the user is changing the long double default
9519 and they did not use -Wno-psabi.
9520 * doc/invoke.texi (PowerPC options): Update the documentation for
9521 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9523 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9524 Alan Hayward <alan.hayward@arm.com>
9525 David Sherwood <david.sherwood@arm.com>
9527 * rtl.h (wider_subreg_mode): New function.
9528 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9529 rather than an unsigned int *.
9530 * ira-color.c (regno_max_ref_width): Replace with...
9531 (regno_max_ref_mode): ...this new variable.
9532 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9533 Use wider_subreg_mode.
9534 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9535 rather than an unsigned int *.
9536 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9537 (process_alt_operands): Likewise.
9538 (invariant_p): Likewise.
9539 * lra-spills.c (assign_mem_slot): Likewise.
9540 (add_pseudo_to_slot): Likewise.
9541 * lra.c (collect_non_operand_hard_regs): Likewise.
9542 (add_regs_to_insn_regno_info): Likewise.
9543 * reload1.c (regno_max_ref_width): Replace with...
9544 (regno_max_ref_mode): ...this new variable.
9545 (reload): Update accordingly. Update call to
9546 ira_sort_regnos_for_alter_reg.
9547 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9548 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9549 (scan_paradoxical_subregs): Likewise.
9551 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9553 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9554 (aarch64_frame): Add emit_frame_chain boolean.
9555 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9556 Move eh_return case to aarch64_layout_frame.
9557 (aarch64_layout_frame): Initialize emit_frame_chain.
9558 (aarch64_expand_prologue): Use emit_frame_chain.
9560 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9562 * config/aarch64/aarch64.c (aarch64_layout_frame):
9563 Ensure LR is always stored at the bottom of the callee-saves.
9564 Remove rarely used frame layout which saves callee-saves at top of
9565 frame, so the store of LR can be used as a valid probe in all cases.
9567 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9569 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9570 Improve unaligned TImode/TFmode base/offset split.
9572 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9577 * combine.c (find_single_use_1): Likewise.
9578 (expand_field_assignment): Likewise.
9579 (move_deaths): Likewise.
9580 * lra-constraints.c (simplify_operand_subreg): Likewise.
9581 (curr_insn_transform): Likewise.
9582 * lra.c (collect_non_operand_hard_regs): Likewise.
9583 (add_regs_to_insn_regno_info): Likewise.
9584 * rtlanal.c (reg_referenced_p): Likewise.
9585 (covers_regno_no_parallel_p): Likewise.
9587 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9589 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9590 Don't print any bits outside the precision of the value.
9591 * wide-int.cc (test_printing): Add some new tests.
9593 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9595 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9596 supports -xbrace_comment option.
9597 * configure: Regenerate.
9598 * config.in: Regenerate.
9599 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9600 (ASM_CPU_SPEC): Use it.
9602 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9604 * target.def (static_rtx_alignment): New hook.
9605 * targhooks.h (default_static_rtx_alignment): Declare.
9606 * targhooks.c (default_static_rtx_alignment): New function.
9607 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9608 * doc/tm.texi: Regenerate.
9609 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9610 instead of targetm.constant_alignment. Remove call to
9612 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9613 (cris_preferred_mininum_alignment): New function, split out from...
9614 (cris_constant_alignment): ...here.
9615 (cris_static_rtx_alignment): New function.
9616 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9618 (ix86_constant_alignment): ...here.
9619 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9620 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9621 (mmix_static_rtx_alignment): New function.
9622 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9623 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9625 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9628 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9629 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9631 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9633 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9634 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9635 mark as a sign-extending load.
9636 (local_pic_load_u): Define.
9638 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9641 * fold-const.c (operand_equal_for_comparison_p): Also return true
9642 if ARG0 is a simple variant of ARG1 with narrower precision.
9643 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9645 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9647 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9649 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9650 scatter_static, scatter_per_elt.
9651 * x86-tune-costs.h: Add new cost entries.
9653 2017-10-25 Richard Biener <rguenther@suse.de>
9655 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9656 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9657 class pass_fre): Move to ...
9658 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9659 class pass_fre): ... here and adjust for statistics.
9661 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9664 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9665 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9666 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9667 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9670 2017-10-25 Richard Biener <rguenther@suse.de>
9672 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9673 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9674 eliminate_push_avail, eliminate_insert): Move inside...
9675 (class eliminate_dom_walker): ... this class in preparation
9677 (fini_eliminate): Remove by merging with ...
9678 (eliminate): ... this function. Adjust for class changes.
9679 (pass_pre::execute): Remove fini_eliminate call.
9680 (pass_fre::execute): Likewise.
9682 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9685 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9686 (VPERMI2, VPERMI2I): New mode iterators.
9687 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9688 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9690 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9691 mode iterator. Remove 3 old define_insn patterns.
9692 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9693 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9694 VPERMI2 mode iterator, remove the other two expanders.
9695 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9696 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9697 instructions, remove the other two patterns.
9698 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9699 mode iterator, remove the other two patterns.
9700 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9701 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9702 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9703 and adjust argument order accordingly.
9704 (ix86_expand_vec_perm): Adjust caller.
9705 (expand_vec_perm_1): Likewise.
9706 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9707 (expand_vec_perm_vpermt2_vpshub2): ... this.
9708 (ix86_expand_vec_perm_const_1): Adjust caller.
9709 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9712 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9713 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9714 (vec_shl_<mode>): Remove unused expander.
9715 (avx512bw_<shift_insn><mode>3): New define_insn.
9716 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9717 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9719 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9722 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9725 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9727 PR rtl-optimization/82396
9728 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9729 (autopref_multipass_init): Simplify initialization.
9730 (autopref_rank_data): Simplify sort order.
9731 * gcc/sched-int.h (autopref_multipass_data_): Remove
9732 multi_mem_insn_p, min_offset and max_offset.
9734 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9737 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9738 Check special value of flag_omit_frame_pointer.
9739 (aarch64_can_eliminate): Likewise.
9740 (aarch64_override_options_after_change_1): Simplify handling of
9741 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9743 2017-10-24 Richard Biener <rguenther@suse.de>
9745 PR tree-optimization/82697
9746 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9747 zero for conditional load and unconditional store.
9749 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9751 * doc/install.texi: Document bootstrap-cet.
9753 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9756 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9757 ENDBR instruction at function entrance if function is only
9760 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9763 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9764 patterns to better describe from which operation the CF is computed.
9765 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9766 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9767 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9768 is 0, use _0 suffixed expanders instead of emitting a comparison
9771 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9773 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9774 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9776 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9779 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9780 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9781 * loop-iv.c (iv_get_reaching_def): Likewise.
9782 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9783 variable is promoted and the partition contains undefined values.
9785 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9787 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9789 (nios2_address_cost): Define.
9790 (nios2_legitimize_address): Recognize (exp + constant) directly.
9791 (TARGET_ADDRESS_COST): Define.
9793 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9795 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9796 (nios2_symbolic_memory_operand_p): Declare.
9797 (nios2_split_large_constant): Declare.
9798 (nios2_split_symbolic_memory_operand): Declare.
9799 * config/nios2/nios2.c: Adjust includes.
9800 (nios2_symbolic_constant_allowed): New.
9801 (nios2_symbolic_constant_p): New.
9802 (nios2_plus_symbolic_constant_p): New.
9803 (nios2_valid_addr_expr_p): Recognize addresses involving
9805 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9806 (nios2_symbolic_memory_operand_p): New.
9807 (nios2_large_constant_p): New.
9808 (nios2_split_large_constant): New.
9809 (nios2_split_plus_large_constant): New.
9810 (nios2_split_symbolic_memory_operand): New.
9811 (nios2_legitimize_address): Code refactoring. Handle addresses
9812 involving symbolic constants.
9813 (nios2_emit_move_sequence): Likewise.
9814 (nios2_print_operand): Improve error output.
9815 (nios2_print_operand_address): Handle LO_SUM.
9816 (nios2_cdx_narrow_form_p): Likewise.
9817 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9818 operands involving symbolic constants.
9819 (movhi_internal, movsi_internal): Likewise.
9820 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9821 (extendhisi2, extendqi<mode>2): Likewise.
9823 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9825 * tree-pass.h (PROP_rtl_split_insns): Define.
9826 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9828 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9830 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9832 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9835 * target.def (const_not_ok_for_debug_p): Default to
9836 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9837 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9838 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9839 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9840 which targetm.const_not_ok_for_debug_p returned true.
9841 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9843 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9845 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9846 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9847 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9849 (ix86_const_not_ok_for_debug_p): New function.
9850 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9851 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9853 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9856 * system.h: Conditionally include "unique-ptr.h" if
9857 INCLUDE_UNIQUE_PTR is defined.
9858 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9859 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9861 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9863 * config/rl78/rl78.md: New define_expand "subdi3".
9865 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9868 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9869 DF_REF_INSN if DF_REF_INSN_INFO is false.
9871 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9873 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9874 xmm_move instead of sse_move.
9875 (sse_store_index): New function.
9876 (ix86_register_move_cost): Be more sensible about mismatch stall;
9877 model AVX moves correctly; make difference between sse->integer and
9879 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9880 moves; make difference between SSE and AVX.
9881 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9882 and zmm_move. Increase size of sse load and store tables;
9883 add unaligned load and store tables; add ssemmx_to_integer.
9884 * x86-tune-costs.h: Update all entries according to real
9885 move latencies from Agner Fog's manual and chip documentation.
9887 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9890 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9891 * config/i386/constraints.md (Wf): New constraint.
9892 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9893 (cmp<dwi>_doubleword): Removed.
9894 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9895 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9896 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9897 expand with cmp<dwi>_doubleword. For LTU and GEU use
9898 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9900 * common.opt (gcolumn-info): Enable by default.
9901 * doc/invoke.texi (gcolumn-info): Document new default.
9903 2017-10-23 Richard Biener <rguenther@suse.de>
9905 PR tree-optimization/82672
9906 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9907 Fold the stmt if we propagated into it.
9909 2017-10-23 Richard Biener <rguenther@suse.de>
9911 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9912 (bitmap_remove_expr_from_set): ... this. All callers call this
9913 for non-constant values.
9914 (bitmap_set_subtract): Rename to...
9915 (bitmap_set_subtract_expressions): ... this. Adjust and
9917 (bitmap_set_contains_value): Remove superfluous check.
9918 (bitmap_set_replace_value): Inline into single caller ...
9919 (bitmap_value_replace_in_set): ... here and simplify.
9920 (dependent_clean): Merge into ...
9921 (clean): ... this using an overload. Adjust.
9922 (prune_clobbered_mems): Adjust.
9923 (compute_antic_aux): Likewise.
9924 (compute_partial_antic_aux): Likewise.
9926 2017-10-23 Richard Biener <rguenther@suse.de>
9928 PR tree-optimization/82129
9930 2017-08-01 Richard Biener <rguenther@suse.de>
9932 PR tree-optimization/81181
9933 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9934 (compute_antic): ... end of iteration here.
9936 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9938 * target.def (starting_frame_offset): New hook.
9939 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9940 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9941 * doc/tm.texi.in: Regenerate.
9942 * hooks.h (hook_hwi_void_0): Declare.
9943 * hooks.c (hook_hwi_void_0): New function.
9944 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9945 STARTING_FRAME_OFFSET.
9946 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9947 * reload1.c (reload): Likewise.
9948 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9949 instead of STARTING_FRAME_OFFSET.
9950 * function.c (try_fit_stack_local): Likewise.
9951 (assign_stack_local_1): Likewise
9952 (instantiate_virtual_regs): Likewise.
9953 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9954 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9955 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9956 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9957 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9958 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9959 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9960 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9961 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9962 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9963 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9964 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9965 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9966 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9967 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9968 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9969 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9970 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9971 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9972 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9973 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9974 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9975 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9976 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9977 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9978 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9979 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9980 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9981 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9982 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9983 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9984 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9985 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9986 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9987 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9988 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9989 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9990 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9991 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9992 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9993 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9994 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9995 return a HOST_WIDE_INT.
9996 (avr_builtin_setjmp_frame_value): Use it instead of
9997 STARTING_FRAME_OFFSET.
9998 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9999 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10000 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10002 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10003 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10004 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10005 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10006 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10007 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10008 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10009 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10010 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10011 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10012 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10013 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10015 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10016 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10017 * config/mips/mips.c (mips_compute_frame_info): Refer to
10018 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10019 (mips_starting_frame_offset): New function.
10020 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10021 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10022 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10023 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10024 and return a HOST_WIDE_INT.
10025 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10026 (mmix_initial_elimination_offset): Refer to
10027 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10028 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10029 * config/pa/pa.c (pa_starting_frame_offset): New function.
10030 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10031 (pa_expand_prologue): Likewise.
10032 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10033 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10034 !FRAME_GROWS_DOWNWARD handling to...
10035 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10036 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10037 !FRAME_GROWS_DOWNWARD handling to...
10038 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10039 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10040 !FRAME_GROWS_DOWNWARD handling to...
10041 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10042 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10044 (rs6000_starting_frame_offset): New function.
10045 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10046 !FRAME_GROWS_DOWNWARD handling to...
10047 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10048 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10049 !FRAME_GROWS_DOWNWARD handling to...
10050 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10051 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10052 !FRAME_GROWS_DOWNWARD handling to...
10053 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10054 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10055 (rs6000_starting_frame_offset): New function.
10056 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10057 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10058 * config/vax/vax.c (vax_starting_frame_offset): New function.
10059 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10060 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10061 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10062 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10063 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10064 * system.h (STARTING_FRAME_OFFSET): Poison.
10066 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10068 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10069 SCALAR_TYPE_MODE instead of TYPE_MODE.
10071 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10072 Alan Hayward <alan.hayward@arm.com>
10073 David Sherwood <david.sherwood@arm.com>
10075 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10077 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10078 Alan Hayward <alan.hayward@arm.com>
10079 David Sherwood <david.sherwood@arm.com>
10081 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10083 2017-10-23 Richard Biener <rguenther@suse.de>
10085 PR tree-optimization/82129
10086 * tree-ssa-pre.c (bitmap_set_and): Remove.
10087 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10088 canonicalizing expressions in the set to those with lowest
10089 ID rather than taking that from the first edge.
10091 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10093 * combine.c (rtx_equal_for_field_assignment_p): Use
10094 byte_lowpart_offset.
10096 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10097 Alan Hayward <alan.hayward@arm.com>
10098 David Sherwood <david.sherwood@arm.com>
10100 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10101 to a SUBREG_PROMOTED_VAR.
10103 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10104 Alan Hayward <alan.hayward@arm.com>
10105 David Sherwood <david.sherwood@arm.com>
10107 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10108 (expand_debug_source_expr): Likewise.
10109 * combine.c (combine_simplify_rtx): Likewise.
10110 * cse.c (fold_rtx): Likewise.
10111 * optabs.c (expand_float): Likewise.
10112 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10113 (simplify_binary_operation_1): Likewise.
10115 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10116 Alan Hayward <alan.hayward@arm.com>
10117 David Sherwood <david.sherwood@arm.com>
10119 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10120 (record_promoted_value): Likewise.
10121 * expr.c (expand_expr_real_2): Likewise.
10122 * ree.c (update_reg_equal_equiv_notes): Likewise.
10123 (combine_set_extension): Likewise.
10124 * rtlanal.c (low_bitmask_len): Likewise.
10125 * simplify-rtx.c (neg_const_int): Likewise.
10126 (simplify_binary_operation_1): Likewise.
10128 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10129 Alan Hayward <alan.hayward@arm.com>
10130 David Sherwood <david.sherwood@arm.com>
10132 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10133 * regcprop.c (maybe_mode_change): Likewise.
10134 * reload1.c (alter_reg): Likewise.
10136 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10138 * inchash.h (inchash::hash::add_wide_int): New function.
10139 * lto-streamer-out.c (hash_tree): Use it.
10141 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10143 * inchash.h (inchash::hash::add_wide_int): Rename to...
10144 (inchash::hash::add_hwi): ...this.
10145 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10146 (polymorphic_call_target_hasher::hash): Likewise.
10147 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10148 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10149 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10150 * lto-streamer-out.c (hash_tree): Likewise.
10151 * optc-save-gen.awk: Likewise.
10152 * tree.c (add_expr): Likewise.
10154 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10157 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10158 for ordered inequality comparisons even with TARGET_IEEE_FP.
10160 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10163 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10164 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10165 Expand with cmp<dwi>_doubleword.
10167 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10169 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10170 List CET intrinsics.
10171 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10172 specific to -fcf-protection option.
10174 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10176 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10177 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10178 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10179 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10180 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10181 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10182 (extra_objs): Add cet.o for Linux/x86 targets.
10183 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10184 * config/i386/cet.c: New file.
10185 * config/i386/cetintrin.h: Likewise.
10186 * config/i386/t-cet: Likewise.
10187 * config/i386/cpuid.h (bit_SHSTK): New.
10188 (bit_IBT): Likewise.
10189 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10190 pass IBT and SHSTK bits.
10191 * config/i386/i386-builtin-types.def
10192 (VOID_FTYPE_UNSIGNED_PVOID): New.
10193 (VOID_FTYPE_UINT64_PVOID): Likewise.
10194 * config/i386/i386-builtin.def: Add CET intrinsics.
10195 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10196 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10197 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10198 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10200 * config/i386/i386.c (rest_of_insert_endbranch): New.
10201 (pass_data_insert_endbranch): Likewise.
10202 (pass_insert_endbranch): Likewise.
10203 (make_pass_insert_endbranch): Likewise.
10204 (ix86_notrack_prefixed_insn_p): Likewise.
10205 (ix86_target_string): Add -mibt, -mshstk flags.
10206 (ix86_option_override_internal): Add flag_cf_protection
10208 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10209 (ix86_print_operand): Add 'notrack' prefix output.
10210 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10211 (ix86_expand_builtin): Expand CET intrinsics.
10212 (x86_output_mi_thunk): Add 'endbranch' instruction.
10213 * config/i386/i386.h (TARGET_IBT): New.
10214 (TARGET_IBT_P): Likewise.
10215 (TARGET_SHSTK): Likewise.
10216 (TARGET_SHSTK_P): Likewise.
10217 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10218 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10219 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10220 (builtin_setjmp_setup): New pattern.
10221 (builtin_longjmp): Likewise.
10222 (rdssp<mode>): Likewise.
10223 (incssp<mode>): Likewise.
10224 (saveprevssp): Likewise.
10225 (rstorssp): Likewise.
10226 (wrss<mode>): Likewise.
10227 (wruss<mode>): Likewise.
10228 (setssbsy): Likewise.
10229 (clrssbsy): Likewise.
10230 (nop_endbr): Likewise.
10231 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10233 * config/i386/immintrin.h: Include <cetintrin.h>.
10234 * config/i386/linux-common.h
10235 (file_end_indicate_exec_stack_and_cet): New prototype.
10236 (TARGET_ASM_FILE_END): New.
10238 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10240 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10241 latencies instead of having separate table; make difference between
10242 integer and float costs.
10243 * i386.h (processor_costs): Remove scalar_stmt_cost,
10244 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10245 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10247 * x86-tune-costs.h: Remove entries which has been removed in
10248 procesor_costs from all tables; make cond_taken_branch_cost
10249 and cond_not_taken_branch_cost COST_N_INSNS based.
10251 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10253 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10255 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10257 * config/i386/i386.md (isa): Remove fma_avx512f.
10258 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10259 <avx512>_fmadd_<mode>_mask3<round_name>,
10260 <avx512>_fmsub_<mode>_mask<round_name>,
10261 <avx512>_fmsub_<mode>_mask3<round_name>,
10262 <avx512>_fnmadd_<mode>_mask<round_name>,
10263 <avx512>_fnmadd_<mode>_mask3<round_name>,
10264 <avx512>_fnmsub_<mode>_mask<round_name>,
10265 <avx512>_fnmsub_<mode>_mask3<round_name>,
10266 <avx512>_fmaddsub_<mode>_mask<round_name>,
10267 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10268 <avx512>_fmsubadd_<mode>_mask<round_name>,
10269 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10270 (*vec_widen_umult_even_v16si<mask_name>,
10271 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10272 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10274 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10276 * extend.texi: Add 'nocf_check' documentation.
10277 * gimple.texi: Add second parameter to
10278 gimple_build_call_from_tree.
10279 * invoke.texi: Add -fcf-protection documentation.
10280 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10282 2017-10-20 Richard Biener <rguenther@suse.de>
10284 PR tree-optimization/82473
10285 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10286 the largest input type.
10288 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10290 * c-attribs.c (handle_nocf_check_attribute): New function.
10291 (c_common_attribute_table): Add 'nocf_check' handling.
10292 * gimple-parser.c: Add second argument NULL to
10293 gimple_build_call_from_tree.
10294 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10295 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10297 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10298 * common.opt: Add fcf-protection flag.
10299 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10300 * flag-types.h: Add enum cf_protection_level.
10301 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10302 Add 'nocf_check' attribute propagation to gimple call.
10303 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10304 (gimple_build_call_from_tree): Update prototype.
10305 (gimple_call_nocf_check_p): New function.
10306 (gimple_call_set_nocf_check): Likewise.
10307 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10308 * ipa-icf.c: Add nocf_check attribute in statement hash.
10309 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10310 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10311 * toplev.c (process_options): Add flag_cf_protection handling.
10313 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10315 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10317 2017-10-20 Richard Biener <rguenther@suse.de>
10319 PR tree-optimization/82603
10320 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10321 remove false predicated stores.
10323 2017-10-20 Richard Biener <rguenther@suse.de>
10325 * graphite-isl-ast-to-gimple.c
10326 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10327 Remove return value and simplify, dump copied stmt after lhs
10329 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10330 Reduce dump verbosity.
10331 (gsi_insert_earliest): Likewise.
10332 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10333 * graphite.c (print_global_statistics): Adjust dumping.
10334 (print_graphite_scop_statistics): Likewise.
10335 (print_graphite_statistics): Do not dump loops here.
10336 (graphite_transform_loops): But here.
10338 2017-10-20 Nicolas Roche <roche@adacore.com>
10340 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10341 * configure: Regenerate.
10343 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10346 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10347 functions when optimizing replace GIMPLE_RETURN stmts with
10348 calls to __builtin_unreachable ().
10351 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10352 for -fsanitize=thread link of executables.
10353 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10354 link of executables.
10357 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10358 New mode iterators.
10359 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10360 define_insns for logical vector shifts to use VI248_AVX512BW
10361 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10362 condition, useless isa and prefix attributes. Change the first
10363 2 of these define_insns to ...
10364 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10365 define_insn for avx512vl.
10366 (<shift_insn><mode>3): ... and this, new define_insn without
10367 masking for non-avx512vl.
10370 * config/i386/sse.md (*andnot<mode>3,
10371 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10372 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10373 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10374 not applied use empty suffix even for TARGET_AVX512VL.
10375 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10376 is applied, supply evex,evex or evex,evex,evex instead of just
10379 2017-10-20 Julia Koval <julia.koval@intel.com>
10381 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10382 (OPTION_MASK_ISA_GFNI_UNSET): New.
10383 (ix86_handle_option): Handle OPT_mgfni.
10384 * config/i386/cpuid.h (bit_GFNI): New.
10385 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10386 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10387 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10388 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10389 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10390 * config/i386/i386.opt: Add mgfni.
10392 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10394 * config/msp430/msp430.c (msp430_option_override): Disable
10395 -fdelete-null-pointer-checks.
10396 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10398 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10400 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10401 of x87 and SSE instructions.
10403 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10405 * asan.c (create_cond_insert_point): Do not update edge count.
10406 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10407 (afdo_propagate_circuit): Likewise.
10408 (afdo_calculate_branch_prob): Likewise.
10409 (afdo_annotate_cfg): Likewise.
10410 * basic-block.h (struct edge_def): Remove count.
10411 (edge_def::count): New accessor.
10412 * bb-reorder.c (rotate_loop): Update.
10413 (find_traces_1_round): Update.
10414 (connect_traces): Update.
10415 (sanitize_hot_paths): Update.
10416 * cfg.c (unchecked_make_edge): Update.
10417 (make_single_succ_edge): Update.
10418 (check_bb_profile): Update.
10419 (dump_edge_info): Update.
10420 (update_bb_profile_for_threading): Update.
10421 (scale_bbs_frequencies_int): Update.
10422 (scale_bbs_frequencies_gcov_type): Update.
10423 (scale_bbs_frequencies_profile_count): Update.
10424 (scale_bbs_frequencies): Update.
10425 * cfganal.c (connect_infinite_loops_to_exit): Update.
10426 * cfgbuild.c (compute_outgoing_frequencies): Update.
10427 (find_many_sub_basic_blocks): Update.
10428 * cfgcleanup.c (try_forward_edges): Update.
10429 (try_crossjump_to_edge): Update
10430 * cfgexpand.c (expand_gimple_cond): Update
10431 (expand_gimple_tailcall): Update
10432 (construct_exit_block): Update
10433 * cfghooks.c (verify_flow_info): Update
10434 (redirect_edge_succ_nodup): Update
10435 (split_edge): Update
10436 (make_forwarder_block): Update
10437 (duplicate_block): Update
10438 (account_profile_record): Update
10439 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10440 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10441 * cfgloopmanip.c (scale_loop_profile): Update.
10443 (lv_adjust_loop_entry_edge): Update.
10444 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10445 (force_nonfallthru_and_redirect): Update.
10446 (purge_dead_edges): Update.
10447 (rtl_flow_call_edges_add): Update.
10448 * cgraphunit.c (init_lowered_empty_function): Update.
10449 (cgraph_node::expand_thunk): Update.
10450 * gimple-pretty-print.c (dump_probability): Update.
10451 (dump_edge_probability): Update.
10452 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10453 * haifa-sched.c (sched_create_recovery_edges): Update.
10454 * hsa-gen.c (convert_switch_statements): Update.
10455 * ifcvt.c (dead_or_predicable): Update.
10456 * ipa-inline-transform.c (inline_transform): Update.
10457 * ipa-split.c (split_function): Update.
10458 * ipa-utils.c (ipa_merge_profiles): Update.
10459 * loop-doloop.c (add_test): Update.
10460 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10461 * lto-streamer-in.c (input_cfg): Update.
10462 (input_function): Update.
10463 * lto-streamer-out.c (output_cfg): Update.
10464 * modulo-sched.c (sms_schedule): Update.
10465 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10466 * predict.c (maybe_hot_edge_p): Update.
10467 (unlikely_executed_edge_p): Update.
10468 (probably_never_executed_edge_p): Update.
10469 (dump_prediction): Update.
10470 (drop_profile): Update.
10471 (propagate_unlikely_bbs_forward): Update.
10472 (determine_unlikely_bbs): Update.
10473 (force_edge_cold): Update.
10474 * profile.c (compute_branch_probabilities): Update.
10475 * reg-stack.c (better_edge): Update.
10476 * shrink-wrap.c (handle_simple_exit): Update.
10477 * tracer.c (better_p): Update.
10478 * trans-mem.c (expand_transaction): Update.
10479 (split_bb_make_tm_edge): Update.
10480 * tree-call-cdce.c: Update.
10481 * tree-cfg.c (gimple_find_sub_bbs): Update.
10482 (gimple_split_edge): Update.
10483 (gimple_duplicate_sese_region): Update.
10484 (gimple_duplicate_sese_tail): Update.
10485 (gimple_flow_call_edges_add): Update.
10486 (insert_cond_bb): Update.
10487 (execute_fixup_cfg): Update.
10488 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10489 * tree-complex.c (expand_complex_div_wide): Update.
10490 * tree-eh.c (lower_resx): Update.
10491 (unsplit_eh): Update.
10492 (cleanup_empty_eh_move_lp): Update.
10493 * tree-inline.c (copy_edges_for_bb): Update.
10494 (freqs_to_counts): Update.
10495 (copy_cfg_body): Update.
10496 * tree-ssa-dce.c (remove_dead_stmt): Update.
10497 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10498 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10499 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10500 (unloop_loops): Update.
10501 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10502 * tree-ssa-loop-split.c (connect_loops): Update.
10503 (split_loop): Update.
10504 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10505 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10506 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10507 * tree-ssa-reassoc.c (branch_fixup): Update.
10508 * tree-ssa-tail-merge.c (replace_block_by): Update.
10509 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10510 (compute_path_counts): Update.
10511 (update_profile): Update.
10512 (recompute_probabilities): Update.
10513 (update_joiner_offpath_counts): Update.
10514 (estimated_freqs_path): Update.
10515 (freqs_to_counts_path): Update.
10516 (clear_counts_path): Update.
10517 (ssa_fix_duplicate_block_edges): Update.
10518 (duplicate_thread_path): Update.
10519 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10520 (case_bit_test_cmp): Update.
10521 (collect_switch_conv_info): Update.
10522 (gen_inbound_check): Update.
10523 (do_jump_if_equal): Update.
10524 (emit_cmp_and_jump_insns): Update.
10525 * tree-tailcall.c (decrease_profile): Update.
10526 (eliminate_tail_call): Update.
10527 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10528 (vect_do_peeling): Update.
10529 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10530 * ubsan.c (ubsan_expand_null_ifn): Update.
10531 (ubsan_expand_ptr_ifn): Update.
10532 * value-prof.c (gimple_divmod_fixed_value): Update.
10533 (gimple_mod_pow2): Update.
10534 (gimple_mod_subtract): Update.
10535 (gimple_ic): Update.
10536 (gimple_stringop_fixed_value): Update.
10538 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10541 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10543 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10545 PR rtl-optimization/82395
10546 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10547 based on non_spilled_static_chain_regno_p.
10549 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10551 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10552 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10553 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10555 2017-10-19 Martin Sebor <msebor@redhat.com>
10557 PR tree-optimization/82596
10558 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10560 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10562 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10563 (get_mem_refs_of_builtin_call): Likewise.
10564 * builtins.c (expand_builtin_apply): Adjust call to
10565 allocate_dynamic_stack_space.
10566 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10567 the third argument to allocate_dynamic_stack_space, otherwise -1.
10568 (expand_builtin): Deal with all alloca variants.
10569 (is_inexpensive_builtin): Likewise.
10570 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10571 * calls.c (special_function_p): Deal with all alloca variants.
10572 (initialize_argument_information): Adjust call to
10573 allocate_dynamic_stack_space.
10574 (expand_call): Likewise.
10575 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10576 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10577 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10578 use it for the stack usage computation.
10579 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10580 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10581 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10582 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10583 (in_loop_p): Remove first argument and useless check.
10584 (pass_walloca::execute): Remove useless test and adjust call to above.
10585 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10586 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10587 (gimplify_call_expr): Deal with all alloca variants.
10588 * hsa-gen.c (gen_hsa_alloca): Likewise.
10589 (gen_hsa_insns_for_call): Likewise.
10590 * ipa-pure-const.c (special_builtin_state): Likewise.
10591 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10592 * tree-object-size.c (alloc_object_size): Likewise.
10593 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10594 (call_may_clobber_ref_p_1): Likewise.
10595 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10596 (ccp_fold_stmt): Likewise.
10597 (optimize_stack_restore): Likewise.
10598 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10599 (mark_all_reaching_defs_necessary_1): Likewise.
10600 (propagate_necessity): Likewise.
10601 (eliminate_unnecessary_stmts): Likewise.
10602 * tree.c (build_common_builtin_nodes): Build
10603 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10604 (build_alloca_call_expr): New function.
10605 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10606 (CASE_BUILT_IN_ALLOCA): Likewise.
10607 (build_alloca_call_expr): Declare.
10608 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10610 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10613 * dwarf2out.c (new_die_raw): New static inline function.
10614 (new_die): Use it to create the DIE.
10615 (add_AT_external_die_ref): Likewise.
10616 (clone_die): Likewise.
10617 (clone_as_declaration): Likewise.
10618 (dwarf2out_vms_debug_main_pointer): Likewise.
10619 (base_type_die): Likewise. Remove early return for corner cases.
10620 Do not call add_pubtype on the DIE here.
10621 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10622 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10623 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10624 native order exists for base types, attach the DIE manually and call
10625 add_pubtype on it. Do not equate a reverse order DIE to the type.
10627 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10629 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10630 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10632 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10634 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10636 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10637 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10638 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10639 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10640 * opts.c (sanitizer_opts): Add builtin.
10641 * ubsan.c (instrument_builtin): New function.
10642 (pass_ubsan::execute): Call it.
10643 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10644 * doc/invoke.texi: Document -fsanitize=builtin.
10646 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10647 builtins, store max (log2 (align), 0) into uchar field instead of
10648 align into uptr field.
10649 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10650 store uchar 0 field instead of uptr 0 field.
10651 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10652 instead of passing one address of struct with 2 locations pass
10653 two addresses of structs with 1 location each.
10654 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10655 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10656 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10657 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10658 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10659 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10660 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10661 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10663 2017-10-19 Martin Liska <mliska@suse.cz>
10666 * file-find.c (remove_prefix): Remove.
10667 * file-find.h (remove_prefix): Likewise.
10668 * gcc-ar.c: Remove smartness of lookup.
10670 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10672 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10673 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10674 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10676 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10679 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10680 (setcc + and to xor + setcc): New peephole2.
10682 2017-10-19 Tom de Vries <tom@codesourcery.com>
10684 * doc/sourcebuild.texi (Test Directives, Variants of
10685 dg-require-support): Add dg-require-stack-size.
10687 2017-10-19 Martin Liska <mliska@suse.cz>
10690 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10691 that have a large alignment.
10692 (gimplify_target_expr): Likewise.
10694 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10696 PR rtl-optimization/82602
10697 * ira.c (rtx_moveable_p): Return false for volatile asm.
10699 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10702 * config/i386/i386-modes.def (CCGZ): New CC mode.
10703 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10704 * config/i386/predicates.md (ix86_comparison_operator):
10706 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10707 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10708 with double-word subtraction.
10709 (put_condition_code): Handle CCGZmode.
10711 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10713 * wide-int.cc (debug (const wide_int &)): New.
10714 (debug (const wide_int *)): New.
10715 (debug (const widest_int &)): New.
10716 (debug (const widest_int *)): New.
10718 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10720 PR middle-end/82556
10721 * lra-constraints.c (curr_insn_transform): Use non-input operand
10722 instead of output one for matched reload.
10724 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10726 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10727 (tree-ssa-loop-ivopts.h): New header file.
10728 (struct builtin_info): New fields.
10729 (classify_builtin_1): Compute and record base and offset parts for
10730 memset builtin partition by calling strip_offset.
10731 (offset_cmp, fuse_memset_builtins): New functions.
10732 (finalize_partitions): Fuse adjacent memset partitions by calling
10734 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10735 Expose the interface.
10736 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10738 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10740 PR tree-optimization/82574
10741 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10742 that data reference must be executed exactly once per iteration
10743 against the outermost loop in nest.
10744 (classify_partition): Update call to above function.
10746 2017-10-18 Richard Biener <rguenther@suse.de>
10748 PR tree-optimization/82591
10749 * graphite.c (graphite_transform_loops): Move code gen message
10751 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10752 Here. Handle scop_to_isl_ast failing.
10753 (scop_to_isl_ast): Limit the number of ISL operations.
10755 2017-10-18 Richard Biener <rguenther@suse.de>
10757 * graphite-isl-ast-to-gimple.c
10758 (translate_isl_ast_to_gimple::set_rename): Simplify.
10759 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10760 (graphite_copy_stmts_from_block): ... here.
10761 (copy_bb_and_scalar_dependences): Simplify.
10762 (add_parameters_to_ivs_params): Canonicalize.
10763 (generate_entry_out_of_ssa_copies): Simplify.
10764 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10765 by passing in ISL dimension.
10766 (parameter_index_in_region_1): Rename to ...
10767 (parameter_index_in_region): ... this.
10768 (extract_affine): Adjust assert, pass down parameter index.
10769 (add_param_constraints): Use range-info when available.
10770 (build_scop_context): Adjust.
10771 * sese.c (new_sese_info): Adjust.
10772 (free_sese_info): Likewise.
10773 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10774 Remove unused typedefs.
10775 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10777 2017-10-18 Martin Liska <mliska@suse.cz>
10779 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10781 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10783 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10784 (s390_sched_init): Do not reset s390_sched_state if we entered the
10785 current basic block via a fallthru edge and all others are unlikely.
10787 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10789 * config/s390/s390.c (NUM_SIDES): New variable.
10790 (LONGRUNNING_THRESHOLD): New variable.
10791 (LATENCY_FACTOR): New variable.
10792 (s390_sched_score): Decrease score for long-running instructions on
10794 (s390_sched_variable_issue): Perform bookkeeping for long-running
10797 2017-10-18 Richard Biener <rguenther@suse.de>
10799 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10800 Simplify with removal of the parameter rename map.
10801 (set_rename): Likewise.
10802 (should_copy_to_new_region): Likewise.
10803 (graphite_copy_stmts_from_block): Likewise.
10804 (copy_bb_and_scalar_dependences): Remove initialization of
10805 unused copied_bb_map.
10806 (copy_def): Remove.
10807 (copy_internal_parameters): Likewise.
10808 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10809 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10810 Use INTEGRAL_TYPE_P.
10811 (parameter_index_in_region_1): Rename to ...
10812 (assign_parameter_index_in_region): ... this. Assert we have
10813 a parameter we handle.
10814 (scan_tree_for_params): Adjust.
10815 * sese.h (parameter_rename_map_t): Remove.
10816 (struct sese_info_t): Remove unused parameter_rename_map and
10817 copied_bb_map members.
10818 * sese.c (new_sese_info): Adjust.
10819 (free_sese_info): Likewise.
10821 2017-10-18 Martin Liska <mliska@suse.cz>
10824 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10825 on an abnormal edge.
10827 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10829 * doc/invoke.texi (ffunction-sections and fdata-sections):
10832 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10834 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10835 the use statement can throw internally.
10837 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10839 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10840 any RTX present on the RHS of a SET.
10841 * compare-elim.c (try_eliminate_compare): Restore comment.
10843 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10845 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10846 may be also called on tcc_type nodes.
10847 * langhooks.c (lhd_tree_size): Likewise.
10849 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10851 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10852 format_warning_at_substring.
10853 (maybe_warn): Convert source_range * param to a location_t. Pass
10854 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10855 (format_directive): Remove code to extract source_ranges and
10856 source_range * in favor of just a location_t.
10857 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10859 * substring-locations.c (format_warning_va): Convert
10860 source_range * param to a location_t.
10861 (format_warning_at_substring): Likewise.
10862 * substring-locations.h (format_warning_va): Likewise.
10863 (format_warning_at_substring): Likewise.
10865 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10867 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10869 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10870 and scatter/gather ops.
10872 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10873 vec_gather_load and vec_scatter_store.
10874 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10875 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10876 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10877 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10878 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10879 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10881 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10883 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10884 Detect FTST instruction and handle its register pops. Only pop
10885 second operand if can_pop_second_op is true.
10886 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10887 set can_pop_second_op to false in the compare_for_stack_reg call.
10889 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10890 output_fp_compare for stack register operands.
10891 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10892 instructions here. Do not emit stack register pops here. Assert
10893 that FCOMPP pops next to top stack register. Rewrite function.
10895 2017-10-17 Nathan Sidwell <nathan@acm.org>
10897 PR middle-end/82577
10898 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10899 use DECL_ASSEMBLER_NAME_RAW.
10901 PR middle-end/82546
10902 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10905 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10906 Wilco Dijkstra <wilco.dijkstra@arm.com>
10908 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10909 converstion to Pmode from the buf_addr.
10911 2017-10-17 Richard Biener <rguenther@suse.de>
10913 * graphite-dependences.c (scop_get_reads_and_writes): Change
10914 output parameters to references.
10916 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10918 PR 71026/tree-optimization
10919 * fold-const.c (distribute_real_division): Removed.
10920 (fold_binary_loc): Remove calls to distribute_real_divison.
10922 2017-10-17 Richard Biener <rguenther@suse.de>
10924 * graphite-scop-detection.c
10925 (scop_detection::stmt_has_simple_data_refs_p): Always use
10926 the full nest as region.
10927 (try_generate_gimple_bb): Likewise.
10928 * sese.c (scalar_evolution_in_region): Simplify now that
10929 SCEV can handle instantiation in regions.
10930 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10931 in the non-loop part of a function if requested.
10933 2017-10-17 Richard Biener <rguenther@suse.de>
10935 PR tree-optimization/82563
10936 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10938 (graphite_regenerate_ast_isl): Call it.
10939 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10941 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10943 PR tree-optimization/82549
10944 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10945 Formatting fixes. Instead of calling make_bit_field_ref with negative
10948 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10950 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10951 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10952 _mm_maskz_reduce_ss): New.
10953 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10954 __builtin_ia32_reducess_mask): Ditto..
10955 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10956 * config/i386/sse.md (reduces<mode>): Renamed to ...
10957 (reduces<mode><mask_scalar_name>): ... this.
10958 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10960 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10961 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10963 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10965 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10966 * selftest-run-tests.c (selftest::run_tests): Call
10967 selftest::unique_ptr_tests_cc_tests.
10968 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10969 * unique-ptr-tests.cc: New file.
10971 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10974 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10976 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10977 (make_hard_regno_born, make_hard_regno_dead): Update
10978 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10980 2017-10-16 Jeff Law <law@redhat.com>
10982 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10984 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10986 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10988 2017-10-16 Olivier Hainque <hainque@adacore.com>
10990 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10991 with_cpu if we were configured for an e500v2 target cpu name.
10993 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10995 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10996 * doc/invoke.texi: Document +nodsp as a valid extension for
10999 2017-10-16 Martin Liska <mliska@suse.cz>
11001 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11002 (test_set_range): Likewise.
11003 (test_range_functions): Rename to ...
11004 (test_bit_in_range): ... this.
11005 (sbitmap_c_tests): Add new test.
11007 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11009 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11011 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11012 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11014 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11016 * config/aarch64/aarch64-builtins.c
11017 (aarch64_types_quadopu_lane_qualifiers): New.
11018 (TYPES_QUADOPU_LANE): New.
11019 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11020 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11021 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11022 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11023 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11024 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11025 (Vdottype, DOTPROD): New.
11026 (sur): Add SDOT and UDOT.
11028 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11030 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11031 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11032 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11033 Add TARGET_DOTPROD.
11034 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11035 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11036 Enable TARGET_DOTPROD.
11037 (cortex-a75.cortex-a55): Likewise.
11038 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11040 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11042 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11043 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11045 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11047 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11048 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11049 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11050 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11051 * config/arm/types.md (neon_dot, neon_dot_q): New.
11052 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11054 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11056 * config/arm/arm.h (TARGET_DOTPROD): New.
11057 * config/arm/arm.c (arm_arch_dotprod): New.
11058 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11059 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11060 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11061 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11062 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11063 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11064 * doc/invoke.texi (armv8.2-a): Document dotprod
11066 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11068 * i386.c (ix86_vec_cost): New function.
11069 (ix86_rtx_costs): Handle vector operations better.
11070 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11071 * x86-tune-costs.h: Add new costs to all tables.
11073 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11075 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11077 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11078 divsd, sqrtss and sqrtsd
11079 * x86-tune-costs.h: Add new entries to all costs.
11080 (znver1_cost): Fix to match real instruction latencies.
11082 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11083 Michael Collison <michael.collison@arm.com>
11085 * compare-elim.c: Include emit-rtl.h.
11086 (can_merge_compare_into_arith): New function.
11087 (try_validate_parallel): Likewise.
11088 (try_merge_compare): Likewise.
11089 (try_eliminate_compare): Call the above when no previous clobber
11091 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11094 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11096 PR middle-end/62263
11097 PR middle-end/82498
11098 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11099 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11101 PR middle-end/62263
11102 PR middle-end/82498
11103 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11104 to be any operand_equal_p operands. For & (B - 1) require
11105 B to be power of 2. Recognize
11106 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11108 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11111 * optabs.c (expand_memory_blockage): Fix call of
11112 targetm.have_memory_blockage.
11114 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11117 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11118 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11119 objects to extra_objs instead of overwriting it.
11121 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11123 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11124 Use any_fp_register_operand as operand[3] predicate. Simplify
11125 equality test for operands[2] and operands[4] memory location.
11126 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11127 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11128 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11129 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11130 any_fp_register_operand as operand[1] predicate. Simplify
11131 equality test for operands[0] and operands[3] memory location.
11132 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11133 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11134 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11136 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11138 * target-insns.def: Add memory_blockage.
11139 * optabs.c (expand_memory_blockage): New function.
11140 (expand_asm_memory_barrier): Rename ...
11141 (expand_asm_memory_blockage): ... to this.
11142 (expand_mem_thread_fence): Call expand_memory_blockage
11143 instead of expand_asm_memory_barrier.
11144 (expand_mem_singnal_fence): Ditto.
11145 (expand_atomic_load): Ditto.
11146 (expand_atomic_store): Ditto.
11147 * doc/md.texi (Standard Pattern Names For Generation):
11148 Document memory_blockage instruction pattern.
11150 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11152 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11153 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11154 * config/rl78/rl78.md: New define_expand "adddi3".
11156 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11158 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11161 2017-10-13 Jeff Law <law@redhat.com>
11163 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11165 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11168 * internal-fn.c (expand_mul_overflow): If both operands have
11169 the same highpart of -1 or 0 and the topmost bit of lowpart
11170 is different, overflow is if res <= 0 rather than res < 0.
11172 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11174 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11175 TARGET_P9_VECTOR code for unaligned_load case.
11177 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11179 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11181 2017-10-13 Nathan Sidwell <nathan@acm.org>
11183 * tree-core.h (tree_contains_struct): Make bool.
11184 * tree.c (tree_contains_struct): Likewise.
11185 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11186 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11187 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11188 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11189 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11191 2017-10-13 Richard Biener <rguenther@suse.de>
11193 * graphite-isl-ast-to-gimple.c
11194 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11195 parameters and dominance check.
11196 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11197 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11198 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11199 Do not update SSA form here or do intermediate IL verification.
11200 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11201 (graphite_initialize): Remove check on the number of loops in
11202 the function and inline into graphite_transform_loops.
11203 (graphite_finalize): Inline into graphite_transform_loops.
11204 (graphite_transform_loops): Perform SSA update and IL verification
11206 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11208 2017-10-13 Richard Biener <rguenther@suse.de>
11210 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11211 graphite_expression_type_precision): Avoid global constructor
11213 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11214 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11215 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11216 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11217 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11218 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11220 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11223 * config/i386/i386.h (ix86_red_zone_size): New.
11224 * config/i386/i386.md (push peephole2s): Replace
11225 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11227 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11228 Alan Hayward <alan.hayward@arm.com>
11229 David Sherwood <david.sherwood@arm.com>
11231 * combine.c (can_change_dest_mode): Reject changes in
11232 REGMODE_NATURAL_SIZE.
11234 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11235 Alan Hayward <alan.hayward@arm.com>
11236 David Sherwood <david.sherwood@arm.com>
11238 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11239 (expand_debug_source_expr): Likewise.
11240 * combine.c (combine_simplify_rtx): Likewise.
11241 * cse.c (fold_rtx): Likewise.
11242 * fwprop.c (canonicalize_address): Likewise.
11243 * targhooks.c (default_shift_truncation_mask): Likewise.
11245 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11246 Alan Hayward <alan.hayward@arm.com>
11247 David Sherwood <david.sherwood@arm.com>
11249 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11250 (widened_mode): Likewise.
11251 (expand_unop): Likewise.
11252 * ree.c (transform_ifelse): Likewise.
11253 (merge_def_and_ext): Likewise.
11254 (combine_reaching_defs): Likewise.
11255 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11257 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11258 Alan Hayward <alan.hayward@arm.com>
11259 David Sherwood <david.sherwood@arm.com>
11261 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11262 * combine.c (gen_lowpart_for_combine): Likewise.
11263 * dwarf2out.c (rtl_for_decl_location): Likewise.
11264 * final.c (alter_subreg): Likewise.
11265 * rtlhooks.c (gen_lowpart_general): Likewise.
11266 (gen_lowpart_if_possible): Likewise.
11268 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11269 Alan Hayward <alan.hayward@arm.com>
11270 David Sherwood <david.sherwood@arm.com>
11272 * calls.c (expand_call): Use subreg_lowpart_offset.
11273 * cse.c (cse_insn): Likewise.
11274 * regcprop.c (copy_value): Likewise.
11275 (copyprop_hardreg_forward_1): Likewise.
11277 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11280 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11281 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11282 =Q constraints to +Q and into insn condition add check
11283 that operands[0] and operands[1] are equal.
11284 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11285 =Q constraints to +Q and into insn condition add check
11286 that operands[0] is equal to either operands[1] or operands[2].
11289 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11290 instead of handling MINUS_EXPR twice (once for each argument),
11291 canonicalize operand order and handle just once, use rtype where
11292 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11295 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11296 any values of __C while still being pattern recognizable as a simple
11297 rotate instruction.
11299 2017-10-13 Richard Biener <rguenther@suse.de>
11301 PR tree-optimization/82451
11303 2017-10-02 Richard Biener <rguenther@suse.de>
11305 PR tree-optimization/82355
11306 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11307 a mapping for the enclosing loop but avoid generating one for
11308 the loop tree root.
11309 (copy_bb_and_scalar_dependences): Remove premature codegen
11310 error on PHIs in blocks duplicated into multiple places.
11311 * graphite-scop-detection.c
11312 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11313 in the region use it as loop and nest to analyze the DR in.
11314 (try_generate_gimple_bb): Likewise.
11315 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11316 (add_loop_constraints): For blocks in a loop not in the region
11317 create a dimension with a single iteration.
11318 * sese.h (gbb_loop_at_index): Remove assert.
11320 * cfgloop.c (loop_preheader_edge): For the loop tree root
11321 return the single successor of the entry block.
11322 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11323 Reset the SCEV hashtable and niters.
11324 * graphite-scop-detection.c
11325 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11326 assert that we only have POLYNOMIAL_CHREC that vary in loops
11327 contained in the region.
11328 (scop_detection::graphite_can_represent_expr): Adjust.
11329 (scop_detection::stmt_has_simple_data_refs_p): For loops
11330 not in the region set loop to NULL. The nest is now the
11331 entry edge to the region.
11332 (try_generate_gimple_bb): Likewise.
11333 * sese.c (scalar_evolution_in_region): Adjust for
11334 instantiate_scev change.
11335 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11336 Make nest parameter the edge into the region.
11337 (create_data_ref): Likewise.
11338 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11339 entry edge into a region and adjust instantiate_scev calls.
11340 (create_data_ref): Likewise.
11341 (graphite_find_data_references_in_stmt): Likewise.
11342 (find_data_references_in_stmt): Pass the loop preheader edge
11343 from the nest argument.
11344 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11345 parameter the edge into the region.
11346 (instantiate_parameters): Use the loop preheader edge as entry.
11347 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11349 (get_instantiated_value_entry): Make instantiate_below parameter
11350 the edge into the region.
11351 (instantiate_scev_name): Likewise. Adjust dominance checks,
11352 when we cannot use loop-based instantiation instantiate by
11353 walking use-def chains.
11354 (instantiate_scev_poly): Adjust.
11355 (instantiate_scev_binary): Likewise.
11356 (instantiate_scev_convert): Likewise.
11357 (instantiate_scev_not): Likewise.
11358 (instantiate_array_ref): Remove.
11359 (instantiate_scev_3): Likewise.
11360 (instantiate_scev_2): Likewise.
11361 (instantiate_scev_1): Likewise.
11362 (instantiate_scev_r): Do not blindly handle N-operand trees.
11363 Do not instantiate array-refs. Handle all constants and invariants.
11364 (instantiate_scev): Make instantiate_below parameter
11365 the edge into the region.
11366 (resolve_mixers): Use the loop preheader edge for the region
11367 parameter to instantiate_scev_r.
11368 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11370 2017-10-13 Richard Biener <rguenther@suse.de>
11372 PR tree-optimization/82525
11373 * graphite-isl-ast-to-gimple.c
11374 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11376 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11377 Fail code generation when we cannot represent the isl integer.
11378 (binary_op_to_tree): Elide modulo operations that are no-ops
11379 in the type we code generate. Remove now superfluous code
11382 2017-10-13 Richard Biener <rguenther@suse.de>
11384 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11385 (scop_detection::harmful_loop_in_region): Remove premature
11386 IV type restriction.
11387 (scop_detection::graphite_can_represent_scev): We can handle
11388 pointer IVs just fine.
11390 2017-10-13 Alan Modra <amodra@gmail.com>
11392 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11393 "Clobbers and Scratch Registers". Add paragraph on
11394 alternative to clobbers for scratch registers and OpenBLAS
11397 2017-10-13 Alan Modra <amodra@gmail.com>
11399 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11400 example of a memory input for a string of known length. Move
11401 commentary out of table. Add a number of new examples
11402 covering array memory inputs.
11404 2017-10-12 Martin Liska <mliska@suse.cz>
11406 PR tree-optimization/82493
11407 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11408 (test_range_functions): New function.
11409 (sbitmap_c_tests): Likewise.
11410 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11411 * selftest.h (sbitmap_c_tests): New function.
11413 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11415 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11417 * config/rs6000/amo.h: Fix spacing issue.
11419 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11422 * config/i386/i386.md (*ashl<mode>3_mask_1,
11423 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11424 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11427 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11429 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11430 (profile_probability): Set max_probability
11431 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11433 * profile-count.c (profile_probability::differs_from_p): Do not
11434 rely on max_probaiblity == 10000
11436 2017-10-12 Jeff Law <law@redhat.com>
11438 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11441 2017-10-12 Martin Sebor <msebor@redhat.com>
11445 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11446 (handle_alias_pairs): Call it.
11447 * common.opt (-Wattribute-alias): New option.
11448 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11449 * doc/invoke.texi (-Wattribute-alias): Document.
11451 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11454 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11456 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11458 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11459 (make_hard_regno_born, make_hard_regno_dead): Update
11460 bb_killed_pseudos and bb_gen_pseudos.
11462 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11464 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11466 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11468 * config/alpha/alpha.c (alpha_split_conditional_move):
11469 Use std::swap instead of manually swapping.
11470 (alpha_stdarg_optimize_hook): Ditto.
11471 (alpha_canonicalize_comparison): Ditto.
11473 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11475 * tree-loop-distribution.c (struct builtin_info): New struct.
11476 (struct partition): Refactor fields into struct builtin_info.
11477 (partition_free): Free struct builtin_info.
11478 (build_size_arg_loc, build_addr_arg_loc): Delete.
11479 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11480 information from struct builtin_info.
11481 (find_single_drs): New function refactored from classify_partition.
11482 Also moved builtin validity checks to this function.
11483 (compute_access_range, alloc_builtin): New functions.
11484 (classify_builtin_st, classify_builtin_ldst): New functions.
11485 (classify_partition): Refactor code into functions find_single_drs,
11486 classify_builtin_st and classify_builtin_ldst.
11487 (distribute_loop): Don't do runtime alias check when distributing
11489 (find_seed_stmts_for_distribution): New function.
11490 (pass_loop_distribution::execute): Refactor code finding seed
11491 stmts into above function. Support distribution for the innermost
11492 two-level loop nest. Adjust dump information.
11494 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11496 * tree-loop-distribution.c: Adjust the general comment.
11497 (NUM_PARTITION_THRESHOLD): New macro.
11498 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11499 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11500 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11501 in call to build_partition_graph.
11502 (finalize_partitions): New parameter. Make loop distribution more
11503 conservative by fusing more partitions.
11504 (distribute_loop): Don't do runtime alias check in case of loop nest
11506 (find_seed_stmts_for_distribution): New function.
11507 (prepare_perfect_loop_nest): New function.
11508 (pass_loop_distribution::execute): Refactor code finding seed stmts
11509 and loop nest into above functions. Support loop nest distribution.
11510 Adjust dump information accordingly.
11512 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11514 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11515 and set PTYPE_SEQUENTIAL for merged partition.
11517 2017-10-12 Richard Biener <rguenther@suse.de>
11519 PR tree-optimization/69728
11521 2017-09-19 Richard Biener <rguenther@suse.de>
11523 PR tree-optimization/69728
11524 * graphite-sese-to-poly.c (schedule_error): New global.
11525 (add_loop_schedule): Handle empty domain by failing the
11527 (build_original_schedule): Handle schedule_error.
11529 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11530 domain by returning an unchanged schedule.
11532 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11534 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11535 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11537 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11539 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11542 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11545 * expr.c (store_field): Don't optimize away bitsize == 0 store
11546 from CALL_EXPR with addressable return type.
11548 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11550 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11551 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11552 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11553 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11554 TARGET_ISEL instead of TARGET_ISEL<sel>.
11556 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11558 * config/rs6000/rs6000.c
11559 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11561 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11563 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11565 (reg_or_cint_operand): Fix comment.
11566 (reg_or_zero_operand): New predicate.
11567 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11568 * config/rs6000/rs6000.c (output_isel): Delete.
11569 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11570 instead of reg_or_cint_operand. Output instruction directly (not via
11572 (isel_unsigned_<mode>): Ditto.
11573 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11574 gpc_reg_operand. Add an instruction alternative for this. Output
11575 instruction directly.
11576 (*isel_reversed_unsigned_<mode>): Ditto.
11578 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11580 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11581 (TARGET_CANONICALIZE_COMPARISON): Define.
11583 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11586 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11587 Check whether the dest is REG before adding REG_EQUIV note.
11589 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11592 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11594 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11595 (make_hard_regno_born, make_hard_regno_dead): Update
11596 bb_killed_pseudos and bb_gen_pseudos.
11598 2017-10-11 Nathan Sidwell <nathan@acm.org>
11600 * incpath.h (enum incpath_kind): Name enum, prefix values.
11601 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11602 * incpath.c (heads, tails): Use INC_MAX.
11603 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11604 (merge_include_chains, split_quote_chain,
11605 register_include_chains): Update incpath_kind names.
11606 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11607 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11609 (add_framework_path, darwin_register_objc_includes): Likewise.
11610 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11612 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11614 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11615 Do not use float_operator operator predicate.
11616 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11617 * config/i386/predicates.md (float_operator): Remove predicate.
11619 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11621 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11622 (*jccxf_i387): Ditto.
11623 (*jcc<mode>_i387): Ditto.
11624 (*jccu<mode>_i387): Ditto.
11625 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11626 (*jcc_*_i387 splitters): Remove.
11627 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11628 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11629 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11632 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11634 * profile-count.h (slow_safe_scale_64bit): New function.
11635 (safe_scale_64bit): New inline.
11636 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11637 * profile-count.c: Include wide-int.h
11638 (slow_safe_scale_64bit): New.
11640 2017-10-11 Nathan Sidwell <nathan@acm.org>
11642 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11643 HAS_DECL_ASSEMBLER_NAME_P.
11644 * gimple-expr.c (gimple_decl_printable_name: Check
11645 HAS_DECL_ASSEMBLER_NAME_P too.
11646 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11647 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11648 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11649 * passes.c (rest_of_decl_compilation): Check
11650 HAS_DECL_ASSEMBLER_NAME_P too.
11651 * recog.c (verify_changes): Likewise.
11652 * tree-pretty-print.c (dump_decl_name): Likewise.
11653 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11655 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11656 (SET_DECL_ASSEMBLER_NAME): Use it.
11657 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11658 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11659 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11661 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11663 * config.gcc (i386, x86_64): Add extra objects.
11664 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11665 (ix86_min_insn_size): Declare.
11666 (ix86_issue_rate): Declare.
11667 (ix86_adjust_cost): Declare.
11668 (ia32_multipass_dfa_lookahead): Declare.
11669 (ix86_macro_fusion_p): Declare.
11670 (ix86_macro_fusion_pair_p): Declare.
11671 (ix86_bd_has_dispatch): Declare.
11672 (ix86_bd_do_dispatch): Declare.
11673 (ix86_core2i7_init_hooks): Declare.
11674 (ix86_atom_sched_reorder): Declare.
11675 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11676 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11677 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11678 (rip_relative_addr_p): Rename to ...
11679 (ix86_rip_relative_addr_p): ... this one; export.
11680 (memory_address_length): Update.
11681 (ix86_issue_rate): Move to x86-tune-sched.c.
11682 (ix86_flags_dependent): Move to x86-tune-sched.c.
11683 (ix86_agi_dependent): Move to x86-tune-sched.c.
11684 (exact_dependency_1): Move to x86-tune-sched.c.
11685 (exact_store_load_dependency): Move to x86-tune-sched.c.
11686 (ix86_adjust_cost): Move to x86-tune-sched.c.
11687 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11688 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11689 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11690 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11691 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11692 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11693 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11694 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11695 (min_insn_size): Rename to ...
11696 (ix86_min_insn_size): ... this one; export.
11697 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11698 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11699 (core2i7_first_cycle_multipass_backtrack): Move to
11700 x86-tune-sched-core.c.
11701 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11702 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11703 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11704 (ix86_avoid_jump_mispredicts): Update.
11705 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11706 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11707 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11708 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11709 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11710 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11711 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11712 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11713 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11714 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11715 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11716 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11717 (BIG): Move to ix86-tune-sched-bd.c.
11718 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11719 (enum insn_path): Move to ix86-tune-sched-bd.c.
11720 (get_mem_group): Move to ix86-tune-sched-bd.c.
11721 (is_cmp): Move to ix86-tune-sched-bd.c.
11722 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11723 (is_branch): Move to ix86-tune-sched-bd.c.
11724 (is_prefetch): Move to ix86-tune-sched-bd.c.
11725 (init_window): Move to ix86-tune-sched-bd.c.
11726 (allocate_window): Move to ix86-tune-sched-bd.c.
11727 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11728 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11729 (process_end_window): Move to ix86-tune-sched-bd.c.
11730 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11731 (find_constant): Move to ix86-tune-sched-bd.c.
11732 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11733 (has_immediate): Move to ix86-tune-sched-bd.c.
11734 (get_insn_path): Move to ix86-tune-sched-bd.c.
11735 (get_insn_group): Move to ix86-tune-sched-bd.c.
11736 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11737 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11738 (add_insn_window): Move to ix86-tune-sched-bd.c.
11739 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11740 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11741 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11742 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11743 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11744 (do_dispatch): Move to ix86-tune-sched-bd.c.
11745 (has_dispatch): Move to ix86-tune-sched-bd.c.
11746 * config/i386/t-i386: Add new object files.
11747 * config/i386/x86-tune-costs.h: New file.
11748 * config/i386/x86-tune-sched-atom.c: New file.
11749 * config/i386/x86-tune-sched-bd.c: New file.
11750 * config/i386/x86-tune-sched-core.c: New file.
11751 * config/i386/x86-tune-sched.c: New file.
11753 2017-10-11 Liu Hao <lh_mouse@126.com>
11755 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11756 the generic version below instead.
11757 (should_colorize): Recognize Windows consoles as terminals
11759 * pretty-print.c [__MINGW32__] (write_all): New function.
11760 [__MINGW32__] (find_esc_head): Likewise.
11761 [__MINGW32__] (find_esc_terminator): Likewise.
11762 [__MINGW32__] (eat_esc_sequence): Likewise.
11763 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11765 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11768 2017-10-11 Richard Biener <rguenther@suse.de>
11770 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11771 Properly call analyze_scalar_evolution with the loop of the stmt.
11773 2017-10-11 Richard Biener <rguenther@suse.de>
11775 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11776 * tree-core.h (tree_base): Add chrec_var union member.
11777 * tree.h (CHREC_VAR): Remove.
11778 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11779 * tree-chrec.h (build_polynomial_chrec): Adjust.
11780 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11781 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11783 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11785 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11786 * match.pd: ... here.
11787 ((T) X == (T) Y): Relax condition.
11789 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11791 PR tree-optimization/82472
11792 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11794 (break_alias_scc_partitions): Update postorder number.
11796 2017-10-11 Martin Liska <mliska@suse.cz>
11799 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11801 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11803 2017-10-11 Martin Liska <mliska@suse.cz>
11805 * print-rtl.c (print_insn): Move declaration of idbuf
11806 to same scope as name.
11808 2017-10-11 Martin Liska <mliska@suse.cz>
11813 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11814 volatile arguments.
11816 2017-10-11 Martin Liska <mliska@suse.cz>
11819 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11820 volatile arguments.
11822 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11824 * config.gcc (default_gnu_indirect_function): Default to yes for
11825 arm*-*-linux* with glibc.
11827 2017-10-11 Richard Biener <rguenther@suse.de>
11829 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11830 default-defs and types we do not want to analyze.
11831 (interpret_loop_phi): Replace unreachable code with an assert.
11832 (compute_scalar_evolution_in_loop): Remove and inline ...
11833 (analyze_scalar_evolution_1): ... here, replacing condition with
11834 what makes the intent clearer. Remove handling of cases
11835 get_scalar_evolution now handles.
11837 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11839 PR rtl-optimization/81434
11840 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11841 comment for main loop. In sched_group_found if, also add checks for
11842 pass and min_cost_group.
11844 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11846 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11847 (rs6000_insn_cost): New function.
11848 * config/rs6000/rs6000.md (cost): New attribute.
11850 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11851 H.J. Lu <hongjiu.lu@intel.com>
11855 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11856 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11857 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11858 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11859 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11860 OPTION_MASK_ISA_MMX and some other bit set require both
11861 mmx and the ISAs without the mmx bit.
11862 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11863 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11864 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11865 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11866 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11867 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11868 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11869 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11870 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11871 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11872 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11873 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11874 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11875 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11876 Add OPTION_MASK_ISA_MMX.
11878 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11880 * config.gcc (armv7*-*-freebsd*): New target.
11881 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11883 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11885 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11886 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11889 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11891 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11892 a template parameter.
11893 (WIDE_INT_REF_FOR): Update accordingly.
11894 * tree.h (wi::int_traits <const_tree>): Delete.
11895 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11896 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11897 (wi::tree_to_wide_ref): New typedef.
11898 (wi::to_wide): New function.
11899 * calls.c (get_size_range): Use wi::to_wide when operating on
11900 trees as wide_ints.
11901 * cgraph.c (cgraph_node::create_thunk): Likewise.
11902 * config/i386/i386.c (ix86_data_alignment): Likewise.
11903 (ix86_local_alignment): Likewise.
11904 * dbxout.c (stabstr_O): Likewise.
11905 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11906 * expr.c (const_vector_from_tree): Likewise.
11907 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11908 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11909 (fold_negate_expr_1, int_const_binop_1, const_binop)
11910 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11911 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11912 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11913 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11914 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11915 (fold_not_const, round_up_loc): Likewise.
11916 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11917 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11918 (alloca_call_type): Likewise.
11919 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11920 * godump.c (go_output_typedef): Likewise.
11921 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11922 * internal-fn.c (get_min_precision): Likewise.
11923 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11924 * ipa-polymorphic-call.c
11925 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11926 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11927 (ipa_modify_call_arguments): Likewise.
11928 * match.pd: Likewise.
11929 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11930 * print-tree.c (print_node_brief, print_node): Likewise.
11931 * stmt.c (expand_case): Likewise.
11932 * stor-layout.c (layout_type): Likewise.
11933 * tree-affine.c (tree_to_aff_combination): Likewise.
11934 * tree-cfg.c (group_case_labels_stmt): Likewise.
11935 * tree-data-ref.c (dr_analyze_indices): Likewise.
11936 (prune_runtime_alias_test_list): Likewise.
11937 * tree-dump.c (dequeue_and_dump): Likewise.
11938 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11939 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11940 * tree-pretty-print.c (dump_generic_node): Likewise.
11941 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11942 (simple_iv_with_niters): Likewise.
11943 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11944 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11945 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11946 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11947 (refine_value_range_using_guard, number_of_iterations_ne_max)
11948 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11949 (get_cst_init_from_scev, record_nonwrapping_iv)
11950 (scev_var_range_cant_overflow): Likewise.
11951 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11952 * tree-ssa-pre.c (compute_avail): Likewise.
11953 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11954 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11955 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11956 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11957 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11958 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11959 (dump_case_nodes, try_switch_expansion): Likewise.
11960 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11961 (vect_do_peeling): Likewise.
11962 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11963 * tree-vect-stmts.c (vectorizable_load): Likewise.
11964 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11965 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11966 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11967 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11968 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11969 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11970 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11971 (evrp_dom_walker::before_dom_children): Likewise.
11972 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11973 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11974 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11975 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11976 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11977 (get_range_pos_neg): Likewise.
11978 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11979 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11980 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11981 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11982 * config/avr/avr.c (avr_fold_builtin): Likewise.
11983 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11984 * config/msp430/msp430.c (msp430_attr): Likewise.
11985 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11986 * config/powerpcspe/powerpcspe-c.c
11987 (altivec_resolve_overloaded_builtin): Likewise.
11988 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11989 (rs6000_expand_ternop_builtin): Likewise.
11990 * config/rs6000/rs6000-c.c
11991 (altivec_resolve_overloaded_builtin): Likewise.
11992 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11993 (rs6000_expand_ternop_builtin): Likewise.
11994 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11996 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11998 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11999 when copying loop nest with only one inner loop.
12001 2017-10-10 Richard Biener <rguenther@suse.de>
12003 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12004 blocks if SCEV is active.
12005 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12007 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12008 (scev_initialize): Assert we are not yet initialized.
12010 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12012 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12013 inner loop's exit stmt by making it always exit the loop, otherwise
12014 we would generate an infinite empty loop.
12016 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12018 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12019 renaming variables in new preheader if it's deleted.
12021 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12023 * tree-loop-distribution.c (struct partition): Remove unused field
12024 loops of the structure.
12025 (partition_alloc, partition_free): Ditto.
12026 (build_rdg_partition_for_vertex): Ditto.
12028 2017-10-09 Jeff Law <law@redhat.com>
12030 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12031 return type to match prototype and documentation.
12033 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12035 * config/rs6000/rs6000.c (processor_costs): Move to ...
12036 * config/rs6000/rs6000.h: ... here.
12037 (rs6000_cost): Declare.
12039 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12041 * except.c (setjmp_fn): New global variable.
12042 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12043 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12044 if DONT_USE_BUILTIN_SETJMP is defined.
12046 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12048 * target.def (insn_cost): New hook.
12049 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12050 * doc/tm.texi: Regenerate.
12051 * rtlanal.c (insn_cost): Use the new hook.
12053 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12055 * combine.c (combine_validate_cost): Compute the new insn_cost,
12056 not just pattern_cost.
12057 (try_combine): Adjust comment.
12059 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12061 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12063 * combine.c (uid_insn_cost): Adjust comment.
12064 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12066 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12067 * dse.c (find_shift_sequence): Ditto.
12068 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12069 (bb_valid_for_noce_process_p): Use pattern_cost.
12070 * rtl.h (insn_rtx_cost): Delete.
12071 (pattern_cost): New prototype.
12072 (insn_cost): New prototype.
12073 * rtlanal.c (insn_rtx_cost): Rename to...
12074 (pattern_cost): ... this.
12077 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12079 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12080 (*jcc<mode>_0_r_i387): Ditto.
12081 (*jccxf_r_i387): Ditto.
12082 (*jcc<mode>_r_i387): Ditto.
12083 (*jccu<mode>_r_i387): Ditto.
12084 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12085 (*jcc): Rename from *jcc_1.
12087 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12089 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12090 deferred rescans after the lvx/stvx recombination pre-pass.
12092 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12094 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12095 memory operation instruction support.
12096 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12097 (rs6000-ibm-aix[789]*): Likewise.
12098 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12099 Document new functions.
12101 2017-10-09 Richard Biener <rguenther@suse.de>
12103 PR tree-optimization/82397
12104 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12105 equality only for semantically equal trees.
12107 2017-10-09 Richard Biener <rguenther@suse.de>
12109 PR tree-optimization/82449
12110 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12111 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12112 allow constant addresses.
12113 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12116 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12118 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12121 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12124 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12127 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12130 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12132 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12135 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12136 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12138 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12140 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12141 (WI_BINARY_PREDICATE_RESULT): Likewise.
12142 (wi::binary_traits::operator_result): New type.
12143 (wi::binary_traits::predicate_result): Likewise.
12144 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12145 (generic_wide_int::operator==, generic_wide_int::operator!=)
12146 (generic_wide_int::operator&, generic_wide_int::and_not)
12147 (generic_wide_int::operator|, generic_wide_int::or_not)
12148 (generic_wide_int::operator^, generic_wide_int::operator+
12149 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12151 (operator~, unary operator-, operator==, operator!=, operator&)
12152 (operator|, operator^, operator+, binary operator-, operator*): New
12154 * expr.c (get_inner_reference): Use wi::bit_and_not.
12155 * fold-const.c (fold_binary_loc): Likewise.
12156 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12157 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12158 (bit_value_binop): Likewise.
12159 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12160 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12161 (extract_range_from_binary_expr_1): Likewise.
12162 (masked_increment): Likewise.
12163 (simplify_bit_ops_using_ranges): Likewise.
12165 2017-10-09 Martin Jambor <mjambor@suse.cz>
12168 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12169 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12170 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12172 (hsa_fixup_mov_insn_type): New function.
12173 (hsa_op_with_type::get_in_type): Use it.
12174 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12175 immediates in an assert.
12176 (hsa_op_with_type::extend_int_to_32bit): New method.
12177 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12178 types. Convert to dest type if necessary.
12179 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12180 (reg_for_gimple_ssa): Pass false as min32int to
12181 hsa_type_for_scalar_tree_type.
12182 (gen_hsa_addr): Fixup type when creating addresable temporary.
12183 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12184 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12185 necessary. Call hsa_fixup_mov_insn_type.
12186 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12187 extend operands and convert to dest type if necessary.
12188 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12189 to dest type if necessary.
12190 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12191 if conversion nt necessary and size matches.
12192 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12193 to dest type if necessary.
12194 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12195 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12197 (gen_hsa_clrsb): Likewise.
12198 (gen_hsa_ffs): Likewise.
12199 (gen_hsa_divmod): Extend operands and convert to dest type if
12201 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12203 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12205 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12206 Remove empty default arguments. Use a brace block as output
12208 (conditional return): Ditto.
12210 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12211 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12213 (group_ending_nop): Ditto.
12214 (doloop_end): Ditto.
12215 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12216 (splitters for those): Ditto.
12218 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12220 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12221 a conditional jump (and the compare for it) so that pc_rtx is the
12223 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12224 for the deleted and renamed ctr<mode>_internal[234] patterns.
12225 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12226 Delete second conditional return pattern.
12227 (ctr<mode>_internal2): Delete this second bdnz pattern.
12228 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12229 (ctr<mode>_internal4): Delete this second bdz pattern.
12231 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12233 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12234 (always_initialized_rtx_for_ssa_name_p): New predicate.
12235 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12236 (finish_out_of_ssa): Free new field of SA.
12237 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12238 * tree-ssa-coalesce.c: Include tree-ssa.h.
12239 (get_parm_default_def_partitions): Remove extern keyword.
12240 (get_undefined_value_partitions): New function.
12241 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12242 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12243 uninitialized bits.
12244 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12246 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12248 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12250 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12252 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12253 for targets that preffer 128bit.
12255 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12257 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12259 2017-10-08 Olivier Hainque <hainque@adacore.com>
12261 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12262 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12264 (thumb_set_return_address): Likewise.
12266 2017-10-08 Olivier Hainque <hainque@adacore.com>
12268 * common/config/arm/arm-common.c (arm_except_unwind_info):
12269 Handle DWARF2_UNWIND_INFO.
12271 2017-10-07 Michael Collison <michael.collison@arm.com>
12273 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12276 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12278 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12279 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12280 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12281 defined, force the creation of a new block for a dispatch label.
12283 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12285 * invoke.texi (Wsuggest-attribute=cold): Document.
12286 * common.opt (Wsuggest-attribute=cold): New
12287 * ipa-pure-const.c (warn_function_cold): New function.
12288 * predict.c (compute_function_frequency): Use it.
12289 * predict.h (warn_function_cold): Declare.
12291 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12293 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12296 2017-10-06 Martin Liska <mliska@suse.cz>
12298 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12299 keyword for member functions.
12300 (struct sanopt_tree_couple): New struct.
12301 (struct sanopt_tree_couple_hash): New function.
12302 (struct sanopt_ctx): Add new hash_map.
12303 (has_dominating_ubsan_ptr_check): New function.
12304 (record_ubsan_ptr_check_stmt): Likewise.
12305 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12306 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12307 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12309 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12312 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12313 aarch64_simd_valid_immediate on CONST_VECTORs.
12314 (aarch64_reg_or_bic_imm): Likewise.
12316 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12318 PR rtl-optimization/82396
12319 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12321 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12323 * graphite-dependences.c (scop_get_reads): Move code to...
12324 (scop_get_must_writes): Move code to...
12325 (scop_get_may_writes): Move code to...
12326 (scop_get_reads_and_writes): ... here.
12327 (scop_get_dependences): Call scop_get_reads_and_writes.
12329 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12331 PR tree-optimization/82434
12332 * fold-const.h (can_native_encode_type_p,
12333 can_native_encode_string_p): Remove.
12334 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12335 don't encode anything, just return what would be otherwise returned.
12336 (native_encode_fixed, native_encode_complex, native_encode_vector):
12338 (native_encode_string): Likewise. Inline by hand
12339 can_native_encode_string_p.
12340 (can_native_encode_type_p): Remove.
12341 (can_native_encode_string_p): Remove.
12342 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12343 STRING_CSTs using can_native_encode_string_p, test all
12344 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12345 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12346 argument from native_encode_expr.
12347 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12348 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12351 2017-10-06 Richard Biener <rguenther@suse.de>
12353 PR tree-optimization/82397
12354 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12355 operand_equal_p but rely on data_ref_compare_tree for detecting
12357 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12358 to match up with dr_group_sort_cmp.
12360 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12363 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12365 * config/s390/s390-builtin-types.def: Regenerate.
12367 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12370 * config/s390/s390-builtin-types.def: Regenerate.
12371 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12372 Change flag from B_VXE to B_VX.
12373 (s390_vec_min_dbl): Remove B_VXE flag.
12375 2017-10-06 Richard Biener <rguenther@suse.de>
12377 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12378 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12379 translate_isl_ast_to_gimple::is_valid_rename,
12380 translate_isl_ast_to_gimple::get_rename,
12381 translate_isl_ast_to_gimple::get_def_bb_for_const,
12382 translate_isl_ast_to_gimple::get_new_name,
12383 translate_isl_ast_to_gimple::collect_all_ssa_names,
12384 translate_isl_ast_to_gimple::copy_loop_phi_args,
12385 translate_isl_ast_to_gimple::collect_all_ssa_names,
12386 translate_isl_ast_to_gimple::copy_loop_phi_args,
12387 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12388 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12389 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12390 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12391 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12392 translate_isl_ast_to_gimple::copy_cond_phi_args,
12393 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12394 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12395 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12396 translate_isl_ast_to_gimple::rename_uses,
12397 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12398 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12399 (set_rename_for_each_def): Likewise.
12400 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12401 here. Handle rewriting SCEV analyzable uses here.
12402 (copy_bb_and_scalar_dependences): Generate code for PHI
12404 (graphite_regenerate_ast_isl): Adjust.
12405 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12406 (add_write, add_read): New functions.
12407 (build_cross_bb_scalars_def): Use it and simplify.
12408 (build_cross_bb_scalars_use): Likewise.
12409 (graphite_find_cross_bb_scalar_vars): Inline into...
12410 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12411 simulating out-of-SSA. Compute liveout and add dependencies.
12412 (build_scops): Force an empty entry block.
12413 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12415 (sese_build_liveouts): Declare.
12416 (sese_trivially_empty_bb_p): Likewise.
12417 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12418 compute liveout and debug_liveout.
12419 (sese_bad_liveouts_use): Remove.
12420 (sese_reset_debug_liveouts_bb): Likewise.
12421 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12422 (sese_build_liveouts): Build liveout and debug_liveout and store
12424 (new_sese_info): Adjust.
12425 (free_sese_info): Likewise.
12426 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12427 do not build liveout here.
12428 (move_sese_in_condition): Adjust region entry.
12429 (scev_analyzable_p): Match up with chrec_apply requirements.
12430 (sese_trivially_empty_bb_p): New.
12431 * tree-into-ssa.c (get_reaching_def): Properly support generating
12432 default-defs for incremental rewrite of anonymous names.
12434 2017-10-06 Richard Biener <rguenther@suse.de>
12436 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12437 precision do not perform modulo reduction.
12439 2017-10-06 Richard Biener <rguenther@suse.de>
12441 PR tree-optimization/82436
12442 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12443 conservatively choose the vectorization factor when checking
12444 whether we can perform the required load permutation.
12445 (vect_transform_slp_perm_load): Assert when we may not fail.
12447 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12449 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12450 message for incompatible -msdata=* and -mcall-* options.
12452 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12454 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12455 rate for post-reload scheduling.
12457 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12459 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12461 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12463 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12464 to improve monte carlo in scimark.
12466 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12468 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12469 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12470 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12471 pentium4_cost, nocona_cost): Set reassociation width to 1.
12472 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12473 width to 2 for fp operations and 1 otherwise.
12474 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12476 (atom_cost): Set reassociation width to 2.
12477 (slm_cost, generic_cost): Set fp reassociation width
12478 to 2 and 1 otherwise.
12479 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12480 (core_cost): Set fp reassociation width to 4 and vector to 2.
12481 (ix86_reassociation_width): Rewrite using cost table; special case
12482 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12483 and TARGET_AVX128_OPTIMAL.
12484 * config/i386/i386.h (processor_costs): Add
12485 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12486 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12487 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12488 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12489 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12490 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12492 2017-10-05 Nathan Sidwell <nathan@acm.org>
12494 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12496 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12498 * config/arm/arm.c (arm_test_fpu_data): New.
12499 (arm_run_selftests): Call arm_test_fpu_data.
12501 2017-10-04 Nathan Sidwell <nathan@acm.org>
12503 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12505 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12507 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12509 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12510 check type for aarch64_simd_valid_immediate.
12511 (aarch64_output_simd_mov_immediate): Update prototype.
12512 (aarch64_simd_valid_immediate): Update prototype.
12513 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12514 support for ORR-immediate.
12515 (and<mode>3): modified pattern to add support for BIC-immediate.
12516 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12517 now checks for valid immediate for BIC and ORR based on new enum
12519 (aarch64_output_simd_mov_immediate): Function now used to output
12520 BIC/ORR imm as well based on new enum argument.
12521 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12523 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12524 (aarch64_reg_or_bic_imm): Likewise.
12526 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12528 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12529 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12531 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12535 PR rtl-optimization/82396
12536 * haifa-sched.c (autopref_multipass_init): Simplify
12538 (autopref_rank_data): Simplify sort order.
12539 * sched-int.h (autopref_multipass_data_): Remove
12540 multi_mem_insn_p, min_offset and max_offset.
12542 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12544 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12546 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12548 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12549 vect_doubleint_cvt.
12551 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12553 * doc/sourcebuild.texi: Document vect_long_mult.
12555 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12557 PR tree-optimization/82413
12558 * fold-const.c (build_range_check): Use widest_int when comparing
12559 the maximum ETYPE value with HIGH.
12561 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12563 PR rtl-optimization/82396
12564 * haifa-sched.c (autopref_multipass_init): Simplify
12566 (autopref_rank_data): Simplify sort order.
12567 * sched-int.h (autopref_multipass_data_): Remove
12568 multi_mem_insn_p, min_offset and max_offset.
12570 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12572 PR tree-optimization/82381
12573 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12574 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12577 PR tree-optimization/82374
12578 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12579 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12580 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12581 current_function_decl to the new decl.
12583 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12585 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12586 helper macro for IEEE float128 hardware built-in functions.
12587 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12589 (TRUNCF128_ODD): Likewise.
12590 (ADDF128_ODD): Likewise.
12591 (SUBF128_ODD): Likewise.
12592 (MULF128_ODD): Likewise.
12593 (DIVF128_ODD): Likewise.
12594 (FMAF128_ODD): Likewise.
12595 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12596 UNSPEC_TRUNC_ROUND_TO_ODD.
12597 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12598 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12599 floating point round to odd instructions.
12600 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12601 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12602 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12603 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12604 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12605 (trunc<mode>sf2_hw): Change the truncate with round to odd
12606 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12607 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12608 to odd hardware instructions.
12609 (sub<mode>3_odd): Likewise.
12610 (mul<mode>3_odd): Likewise.
12611 (div<mode>3_odd): Likewise.
12612 (sqrt<mode>2_odd): Likewise.
12613 (fma<mode>4_odd): Likewise.
12614 (fms<mode>4_odd): Likewise.
12615 (nfma<mode>4_odd): Likewise.
12616 (nfms<mode>4_odd): Likewise.
12617 (trunc<mode>df2_odd): Change the truncate with round to odd
12618 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12620 * doc/extend.texi (PowerPC built-in functions): Update documentation
12621 for existing IEEE float128-bit built-in functions. Add built-in
12622 functions that generate the IEEE 128-bit floating point round to
12625 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12627 PR rtl-optimization/77729
12628 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12629 to (X&(C1&~C2))|C2 transformations.
12631 2017-10-03 Martin Jambor <mjambor@suse.cz>
12633 PR tree-optimization/82363
12634 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12635 mismatch, mark lacc written regardless of racc.
12637 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12639 PR tree-optimization/82381
12640 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12641 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12642 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12643 fallthrough into reassoc_stmt_dominates_stmt_p.
12646 * combine.c (combine_instructions): Don't combine in unreachable
12649 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12652 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12653 function to not use the have_cpu variable. Do not set cpu_index,
12654 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12655 or the default cpu.
12656 (rs6000_valid_attribute_p): Remove duplicate initializations of
12657 old_optimize and func_optimize.
12658 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12659 (rs6000_activate_target_options): Make global.
12660 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12663 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12665 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12666 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12667 Set *poffset to 0 and *psize and *pmax_size to -1 if
12668 *poffset + *psize overflows in HOST_WIDE_INT.
12670 PR tree-optimization/82387
12671 PR tree-optimization/82388
12672 PR tree-optimization/82389
12673 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12674 instead of live_bytes non-NULL.
12676 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12679 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12682 2017-10-02 Richard Biener <rguenther@suse.de>
12684 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12685 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12686 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12688 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12690 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12691 requested precision matches the type's.
12692 * calls.c (alloc_max_size): Calculate the new candidate size as
12693 a widest_int and use wi::to_widest when comparing it with the
12694 current candidate size.
12695 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12696 zero rather than integer_zero_node.
12697 * match.pd: Check for a no-op conversion before using wi::add
12698 rather than after. Use tree_to_uhwi when summing small shift
12699 counts into an unsigned int.
12701 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12702 Alan Hayward <alan.hayward@arm.com>
12703 David Sherwood <david.sherwood@arm.com>
12706 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12707 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12708 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12709 POINTER_AND_FP_REGS.
12711 2017-10-02 Richard Biener <rguenther@suse.de>
12713 PR tree-optimization/82355
12714 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12715 a mapping for the enclosing loop but avoid generating one for
12716 the loop tree root.
12717 (copy_bb_and_scalar_dependences): Remove premature codegen
12718 error on PHIs in blocks duplicated into multiple places.
12719 * graphite-scop-detection.c
12720 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12721 in the region use it as loop and nest to analyze the DR in.
12722 (try_generate_gimple_bb): Likewise.
12723 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12724 (add_loop_constraints): For blocks in a loop not in the region
12725 create a dimension with a single iteration.
12726 * sese.h (gbb_loop_at_index): Remove assert.
12728 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12730 * omp-expand.c (adjust_context_scope): New function.
12731 (expand_parallel_call): Call adjust_context_scope.
12733 2017-10-01 Jeff Law <law@redhat.com>
12735 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12736 dom_opt_dom_walker class with direct access to private members.
12737 Add comments. Call test_for_singularity.
12738 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12739 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12740 m_dummy_cond anymore.
12741 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12743 (pass_dominator:execute): Build the dummy_cond here and pass it
12744 to the dom_opt_dom_walker ctor.
12745 (test_for_singularity): New function.
12747 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12748 Maya Rashish <coypu@sdf.org>
12750 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12751 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12752 (alpha*-*-netbsd*) Use nbsd_tm_file.
12753 (arm*-*-netbsdelf*) Likewise.
12754 (i[34567]86-*-netbsdelf*) Likewise.
12755 (x86_64-*-netbsd*) Likewise.
12756 (mips*-*-netbsd*) Likewise.
12757 (powerpc-*-netbsd*) Likewise.
12758 (sh*-*-netbsd*) Likewise.
12759 (sparc-*-netbsdelf*) Likewise.
12760 (sparc64-*-netbsd*) Likewise.
12761 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12763 (vax-*-netbsdelf*) Likewise.
12764 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12765 (UINT_FAST8_TYPE) Likewise.
12766 (INT_FAST16_TYPE) Check CHAR_FAST16.
12767 (UINT_FAST16_TYPE) Likewise.
12769 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12772 * config/i386/i386.md
12773 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12774 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12775 *divmodsi4_zext_2): New define_insn_and_split.
12776 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12777 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12778 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12779 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12780 New define_insn_and_split.
12781 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12782 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12783 operands[1] having DImode when mode is SImode.
12785 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12786 always SImode for DIV and MOD in REG_EQUAL notes.
12788 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12790 PR middle-end/82319
12791 * match.pd: Fix handling of NaNs in pattern.
12793 2017-09-29 Jeff Law <law@redhat.com>
12795 * sbitmap.c (bitmap_bit_in_range_p): New function.
12796 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12797 * tree-ssa-dse.c (live_bytes_read): New function.
12798 (dse_classify_store): Ignore reads of dead bytes.
12800 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12801 typos and whitespace errors.
12802 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12803 * config/s390/s390.c (s390_emit_prologue): Likewise.
12805 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12808 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12809 with a symbol for LRA.
12811 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12813 PR rtl-optimization/82338
12814 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12816 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12818 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12819 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12820 (qsort_chk): Declare.
12821 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12822 (qsort_chk): New function.
12824 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12826 PR tree-optimization/82337
12827 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12828 phi definition if the PHI result appears in an abnormal PHI.
12829 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12830 basis appears in an abnormal PHI.
12832 2017-09-29 Richard Biener <rguenther@suse.de>
12834 * graphite-isl-ast-to-gimple.c
12835 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12836 (binary_op_to_tree): Use it.
12837 (get_rename_from_scev): Likewise.
12838 (copy_loop_phi_nodes): Likewise.
12839 (copy_bb_and_scalar_dependences): Likewise.
12840 (translate_pending_phi_nodes): Likewise.
12842 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12845 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12846 for movabsq $(i32 << shift), r64.
12848 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12850 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12851 index when encoding %esp as %rsp to avoid 0x67 prefix.
12853 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12855 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12856 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12858 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12860 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12861 Extensions with more than 16 double VFP registers.
12862 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12863 to_clear_mask and all code related to it. Replace the remaining
12864 entry by a sbitmap and adapt code accordingly.
12866 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12868 * brig-builtins.def: Change pure attributes to const.
12870 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12872 * config.gcc (default_gnu_indirect_function): Default to yes for
12873 sparc*-*-linux* with glibc.
12875 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12877 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12878 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12879 when creating .init_array and .fini_array sections with priority
12882 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12885 * config/aarch64/aarch64.c
12886 (aarch64_builtin_support_vector_misalignment): Always return false
12887 when misalignment is unknown.
12889 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12891 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12892 this function to return false if the definition used by the swap
12893 instruction is artificial, or if the memory address from which the
12894 constant value is loaded is not represented by a base address held
12895 in a register or if the base address register is a frame or stack
12896 pointer. Additionally, return false if the base address of the
12897 loaded constant is a SYMBOL_REF but is not considered to be a
12899 (replace_swapped_load_constant): New function.
12900 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12901 loaded constant vector with a load of a swapped constant vector.
12903 2017-09-27 Carl Love <cel@us.ibm.com>
12905 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12906 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12907 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12910 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12912 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12913 first, always call autopref_rank_data otherwise.
12915 2017-09-27 Richard Biener <rguenther@suse.de>
12917 * graphite-scop-detection.c (find_scop_parameters): Move
12918 loop bound handling ...
12919 (gather_bbs::before_dom_children): ... here, avoiding the need
12920 to build scop_info->loop_nest.
12921 (record_loop_in_sese): Remove.
12922 * sese.h (sese_info_t::loop_nest): Remove.
12923 * sese.c (new_sese_info): Do not allocate loop_nest.
12924 (free_sese_info): Do not free loop_nest.
12926 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12929 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12930 lhs from calls if the lhs has addressable type.
12932 2017-09-27 Richard Biener <rguenther@suse.de>
12934 * graphite.h (scop::max_alias_set): New member.
12935 * graphite-scop-detection.c: Remove references to non-existing
12936 --param in comments.
12937 (build_alias_sets): Record the maximum alias set used for drs.
12938 (build_scops): Support zero as unlimited for
12939 --param graphite-max-arrays-per-scop.
12940 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12941 and inline into ...
12942 (build_poly_sr_1): ... here. Compute alias set based on the
12943 maximum alias set used for drs rather than
12944 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12946 2017-09-27 Richard Biener <rguenther@suse.de>
12948 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12949 --param loop-block-tile-size=0 to disable tiling.
12951 2017-09-27 Richard Biener <rguenther@suse.de>
12953 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12954 (graphite-max-nb-scop-params): Document special value zero.
12955 * domwalk.h (dom_walker::STOP): New symbolical constant.
12956 (dom_walker::dom_walker): Add optional parameter for bb to
12958 (dom_walker::~dom_walker): Declare.
12959 (dom_walker::before_dom_children): Document STOP return value.
12960 (dom_walker::m_user_bb_to_rpo): New member.
12961 (dom_walker::m_bb_to_rpo): Likewise.
12962 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12963 mapping here if not provided by the user.
12964 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12965 provided by the user.
12966 (dom_walker::STOP): Define.
12967 (dom_walker::walk): Do not compute bb to RPO mapping here.
12968 Support STOP return value from before_dom_children to stop
12970 * graphite-optimize-isl.c (optimize_isl): If the schedule
12971 is the same still generate code if -fgraphite-identity
12972 or -floop-parallelize-all are given.
12973 * graphite-scop-detection.c: Include cfganal.h.
12974 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12976 (gather_bbs::before_dom_children): Return STOP for BBs
12978 (build_scops): Compute bb to RPO mapping and pass it to
12979 the domwalk. Treat --param graphite-max-nb-scop-params=0
12980 as not limiting the number of params.
12981 * graphite.c (graphite_initialize): Remove limit on the
12982 number of basic-blocks in a function.
12983 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12984 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12985 default value of 10.
12987 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12989 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12990 Adjust code to eliminate needing to do the shift right 32-bits
12991 operation after XSCVDPSPN.
12993 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12995 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12996 ((X / Y) != 0 -> X >= Y): Likewise.
12998 2017-09-26 Carl Love <cel@us.ibm.com>
13000 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13001 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13002 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13003 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13004 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13005 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13006 definitions and overloading.
13007 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13008 statement for P9V_BUILTIN_XST_LEN_R.
13009 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13010 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13011 define_expand and define_insn for the instructions and builtins.
13012 * doc/extend.texi: Update the built-in documentation file for the new
13013 built-in functions.
13014 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13015 define_insn for the instructions
13017 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13020 * gcc/config/netbsd-protos.h: New file.
13021 * gcc/config/netbsd.c: New file.
13022 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13023 * gcc/config/t-netbsd: New file.
13024 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13025 (tmake_file) Add t-netbsd.
13026 (extra_objs) Add netbsd.o.
13028 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13032 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13034 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13036 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13037 sign extension from a vector register to a GPR by doing a 32-bit
13038 direct move and then an EXTSW.
13039 (extendsi<mode>2 splitter): Likewise.
13040 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13041 right or vector extract after doing XSCVDPSPN. Use
13042 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13044 (movdi_from_sf_zero_ext): Likewise.
13045 (reload_gpr_from_vsxsf): Likewise.
13046 (p8_mfvsrd_4_disf): Delete, no longer used.
13047 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13048 then needing to move the SFmode to a GPR to use the XSCVDPSP
13049 instruction instead of FRSP and XSCVDPSPN.
13050 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13051 it is adjacent to the other XSCVSPDP insns.
13052 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13053 SFmode to be in traditional Altivec registers.
13054 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13055 (vsx_xscvspdpn): Likewise.
13056 (vsx_xscvspdpn_scalar): Likewise.
13058 2017-09-26 Martin Jambor <mjambor@suse.cz>
13060 * tree-sra.c (compare_access_positions): Put integral types first,
13061 stabilize sorting of integral types, remove conditions putting
13062 non-full-precision integers last.
13063 (sort_and_splice_var_accesses): Disable scalarization if a
13064 non-integert would be represented by a non-full-precision integer.
13066 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13068 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13069 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13070 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13071 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13072 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13073 conditionals inside the function instead of around it. Call
13074 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13075 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13077 2017-09-26 Richard Biener <rguenther@suse.de>
13079 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13081 (scop_detection::build_scop_breadth): ... this. Removed.
13082 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13083 (scop_detection::harmful_stmt_in_bb): Likewise.
13084 (scop_detection::graphite_can_represent_stmt): Likewise.
13085 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13086 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13087 (scop_detection::can_represent_loop_1): ... this. Removed.
13088 (scop_detection::harmful_loop_in_region): Simplify after inlining
13089 the above and remove more quadraticness.
13090 (build_scops): Adjust.
13091 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13094 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13097 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13098 REGNO (base) == SP_REG if base is a REG.
13100 PR middle-end/35691
13101 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13102 if it is different SSA_NAME.
13103 (optimize_range_tests_cmp_bitwise): New function.
13104 (optimize_range_tests): Call it.
13106 2017-09-26 Richard Biener <rguenther@suse.de>
13108 PR tree-optimization/82321
13109 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13110 for the def being inside the loop.
13112 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13114 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13116 * config/s390/s390-builtins.def: Fix constraint on op4.
13118 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13120 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13121 independent expanders.
13122 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13123 ("vec_ordered", "vec_unordered"): New expanders.
13125 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13127 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13130 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13132 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13133 vec_unpacks_lo_v16qi.
13134 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13136 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13138 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13139 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13140 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13142 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13144 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13146 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13148 ("*vec_slb<mode>"): New insn pattern.
13149 ("vec_shr_<mode>"): New expander.
13150 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13151 and force the shift count operand to V16QImode.
13152 ("vec_srb<mode>"): Set shift count mode to V16QI.
13154 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13156 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13157 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13158 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13160 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13163 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13165 2017-09-26 Richard Biener <rguenther@suse.de>
13167 PR tree-optimization/82320
13168 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13171 2017-09-25 Jeff Law <law@redhat.com>
13173 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13174 prototype for new argument.
13175 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13176 mostly extracted from rs6000_emit_allocate_stack.
13177 (rs6000_emit_probe_stack_range_stack_clash): New function.
13178 (rs6000_emit_allocate_stack): Call
13179 rs6000_emit_probe_stack_range_stack_clash as needed.
13180 (rs6000_emit_probe_stack_range): Add additional argument
13181 to call to gen_probe_stack_range{si,di}.
13182 (output_probe_stack_range): New.
13183 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13184 (output_probe_stack_range_stack_clash): New.
13185 (rs6000_emit_prologue): Emit notes into dump file as requested.
13186 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13187 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13188 Add additional operand and pass it to output_probe_stack_range.
13190 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13192 PR tree-optimization/82163
13193 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13194 (checking_verify_loop_closed_ssa): New parameter.
13195 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13196 (check_loop_closed_ssa_stmt): Delete.
13197 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13198 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13199 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13202 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13204 * brig-builtins.def: Treat HSAIL barrier builtins as
13205 setjmp/longjump style functions.
13207 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13209 * target.def (constant_alignment): New hook.
13210 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13211 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13212 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13213 * doc/tm.texi: Regenerate.
13214 * targhooks.h (default_constant_alignment): Declare.
13215 (constant_alignment_word_strings): Likewise.
13216 * targhooks.c (default_constant_alignment): New function.
13217 (constant_alignment_word_strings): Likewise.
13218 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13219 instead of CONSTANT_ALIGNMENT.
13220 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13221 (force_const_mem): Likewise.
13222 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13223 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13224 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13225 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13226 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13228 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13229 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13230 constant_alignment_word_strings.
13231 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13232 (CONSTANT_ALIGNMENT): Likewise.
13233 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13234 (arm_constant_alignment): New function.
13235 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13236 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13237 constant_alignment_word_strings.
13238 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13239 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13240 constant_alignment_word_strings.
13241 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13242 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13243 (cris_constant_alignment): New function.
13244 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13245 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13246 (epiphany_constant_alignment): New function.
13247 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13248 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13249 constant_alignment_word_strings.
13250 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13251 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13252 constant_alignment_word_strings.
13253 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13254 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13255 constant_alignment_word_strings.
13256 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13257 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13258 * config/i386/i386.c (ix86_constant_alignment): Make static.
13259 Use the same interface as the target hook.
13260 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13261 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13262 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13263 constant_alignment_word_strings.
13264 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13265 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13266 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13267 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13268 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13269 constant_alignment_word_strings.
13270 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13271 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13272 constant_alignment_word_strings.
13273 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13274 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13275 constant_alignment_word_strings.
13276 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13277 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13279 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13280 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13281 * config/mips/mips.c (mips_constant_alignment): New function.
13282 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13283 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13284 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13285 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13286 (mmix_constant_alignment): Make static. Use the same interface
13287 as the target hook.
13288 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13289 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13290 constant_alignment_word_strings.
13291 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13292 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13293 constant_alignment_word_strings.
13294 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13295 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13296 constant_alignment_word_strings.
13297 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13298 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13299 (rs6000_constant_alignment): New function.
13300 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13301 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13302 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13303 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13304 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13305 (rs6000_constant_alignment): New function.
13306 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13307 * config/s390/s390.c (s390_constant_alignment): New function.
13308 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13309 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13310 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13311 constant_alignment_word_strings.
13312 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13313 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13314 (sparc_constant_alignment): New function.
13315 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13316 * config/spu/spu.c (spu_constant_alignment): New function.
13317 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13318 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13319 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13320 constant_alignment_word_strings.
13321 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13322 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13323 constant_alignment_word_strings.
13324 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13325 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13326 constant_alignment_word_strings.
13327 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13328 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13329 (visium_constant_alignment): New function.
13330 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13331 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13332 (xtensa_constant_alignment): New function.
13333 * system.h (CONSTANT_ALIGNMENT): Poison.
13335 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13338 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13339 (rs6000_builtin_valid_without_lhs): New helper function.
13340 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13341 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13343 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13345 * target.h (vec_perm_indices): Use unsigned short rather than
13347 (auto_vec_perm_indices): Likewise.
13348 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13349 Use unsigned int rather than unsigned char.
13350 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13352 2017-09-25 Richard Biener <rguenther@suse.de>
13354 * cfgloop.h (sort_sibling_loops): Declare.
13355 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13356 (sort_sibling_loops): New function sorting the sibling loop list
13358 * graphite.c (graphite_transform_loops): Sort sibling loops.
13360 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13362 * target.def (vec_perm_const_ok): Change sel parameter to
13364 * optabs-query.c (can_vec_perm_p): Update accordingly.
13365 * doc/tm.texi: Regenerate.
13366 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13367 auto_vec_perm_indices and remove separate nelt field.
13368 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13369 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13370 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13371 (aarch64_expand_vec_perm_const): Update accordingly.
13372 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13373 to vec_perm_indices.
13374 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13375 auto_vec_perm_indices and remove separate nelt field.
13376 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13377 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13378 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13380 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13381 to vec_perm_indices.
13382 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13383 sel to vec_perm_indices.
13384 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13385 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13386 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13388 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13391 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13394 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13395 on the FUNCTION_DECL function context if it has a DIE that is a
13398 2017-09-25 Richard Biener <rguenther@suse.de>
13400 PR tree-optimization/82285
13401 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13404 2017-09-25 Tom de Vries <tom@codesourcery.com>
13408 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13411 2017-09-25 Richard Biener <rguenther@suse.de>
13413 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13414 ISL errors other than isl_error_quota happen. Dump if the
13415 schedule is the same.
13416 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13417 errors instead of aborting inside ISL.
13419 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13422 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13423 of libgcc_eh for m64.
13424 * config/i386/darwin64.h: Likewise.
13426 2017-09-25 Richard Biener <rguenther@suse.de>
13428 PR middle-end/82144
13429 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13430 attribute for incomplete types nor twice for complete ones.
13432 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13435 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13436 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13439 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13442 * config/i386/i386.opt (mprefer-avx256): Use
13443 ix86_target_flags variable.
13444 * config/i386/i386.c (ix86_target_string): Move
13445 -mprefer-avx256 to flag2_opts.
13447 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13449 PR middle-end/35691
13450 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13451 and x != -1 | y != -1 into (x & y) != -1.
13453 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13455 * config.gcc: Add new case statement to set
13456 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13457 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13458 s390x-*-linux* case statements. Added aarch64 to the list of
13459 supported architectures.
13461 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13463 PR tree-optimization/82289
13464 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13465 STMT_VINFO_RELEVANT_P.
13467 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13468 Alan Hayward <alan.hayward@arm.com>
13469 David Sherwood <david.sherwood@arm.com>
13471 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13472 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13473 (extract_range_from_binary_expr_1): Don't call
13474 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13476 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13477 Alan Hayward <alan.hayward@arm.com>
13478 David Sherwood <david.sherwood@arm.com>
13480 * target.def (preferred_vector_alignment): New hook.
13481 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13483 * doc/tm.texi: Regenerate.
13484 * targhooks.h (default_preferred_vector_alignment): Declare.
13485 * targhooks.c (default_preferred_vector_alignment): New function.
13486 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13488 (DR_TARGET_ALIGNMENT): New macro.
13489 (aligned_access_p): Update commentary.
13490 (vect_known_alignment_in_bytes): New function.
13491 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13493 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13494 Calculate the misalignment based on the target alignment rather than
13496 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13497 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13498 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13499 the target alignment, rather than masking the element misalignment
13500 with the number of elements in a vector. Also use the target
13501 alignment when calculating the maximum number of peels.
13502 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13503 instead of TYPE_ALIGN_UNIT.
13504 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13505 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13506 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13507 (vect_create_data_ref_ptr): Likewise.
13508 (vect_setup_realignment): Realign by ANDing with
13509 -DR_TARGET_MISALIGNMENT.
13510 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13511 the number of peels based on DR_TARGET_ALIGNMENT.
13512 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13513 with the guaranteed alignment boundary when deciding whether
13515 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13516 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13517 (ensure_base_align): Remove stmt_info parameter. Get the
13518 target base alignment from DR_TARGET_ALIGNMENT.
13519 (vectorizable_store): Update call accordingly. Interpret
13520 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13522 (vectorizable_load): Likewise.
13524 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13525 Alan Hayward <alan.hayward@arm.com>
13526 David Sherwood <david.sherwood@arm.com>
13528 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13529 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13530 (vect_enhance_data_refs_alignment): Likewise.
13532 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13534 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13535 error. Only quit immediately if parsing is complete.
13536 (BEGIN): Initialize fatal_err and parse_done.
13537 (begin fpu, end fpu): Check number of arguments.
13538 (begin arch, end arch): Likewise.
13539 (begin cpu, end cpu): Likewise.
13540 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13541 (optalias): Likewise.
13543 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13545 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13546 * config/arm/arm-isa.h: Delete. Move definitions to ...
13547 * arm-cpus.in: ... here. Use new feature and fgroup values.
13548 * config/arm/arm.c (arm_option_override): Use lower case for feature
13550 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13551 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13552 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13554 (print_isa_bits_for): New function.
13555 (gen_isa): New function.
13556 (gen_comm_data): Use print_isa_bits_for.
13557 (define feature): New keyword.
13558 (define fgroup): New keyword.
13559 * config/arm/t-arm (TM_H): Remove.
13560 (GTM_H): Add arm-isa.h.
13561 (arm-isa.h): Add rule to generate file.
13562 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13563 case for feature bit names.
13565 2017-09-22 Richard Biener <rguenther@suse.de>
13567 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13569 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13570 print of no dependency loops ...
13571 * graphite.c (graphite_transform_loops): ... here.
13572 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13574 (same_close_phi_node, remove_duplicate_close_phi,
13575 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13576 (canonicalize_loop_closed_ssa): ... here and simplify.
13577 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13578 (optimize_isl): Use dump_printf_loc to tell when we stopped
13579 optimizing because of an ISL timeout.
13581 2017-09-22 Richard Biener <rguenther@suse.de>
13583 PR tree-optimization/82291
13584 * tree-if-conv.c (predicate_mem_writes): Make sure to
13585 remove writes in blocks predicated with false.
13587 2017-09-22 Richard Biener <rguenther@suse.de>
13589 * sese.c: Include cfganal.h.
13590 (if_region_set_false_region): Remove.
13591 (create_if_region_on_edge): Likewise.
13592 (move_sese_in_condition): Re-implement without destroying
13595 2017-09-22 Richard Biener <rguenther@suse.de>
13597 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13598 Verify both BBs contain loop PHI nodes before dispatching to
13599 copy_loop_phi_args.
13600 (graphite_regenerate_ast_isl): Do not recompute dominators,
13601 do not verify three times. Restructure for clarity.
13602 * graphite-scop-detection.c (same_close_phi_node,
13603 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13604 defined_in_loop_p, canonicalize_loop_closed_ssa,
13605 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13606 checking and SSA rewrite, move to ...
13607 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13608 (graphite_initialize): Do not pass in ctx, do not reset the
13609 SCEV cache, compute only dominators.
13610 (graphite_transform_loops): Allocate ISL ctx after
13611 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13612 Maintain post-dominators only around build_scops.
13613 * sese.c (if_region_set_false_region): Make static. Free
13614 and recompute dominators.
13615 (move_sese_in_condition): Assert we don't get called with
13616 post-dominators computed.
13617 * sese.h (if_region_set_false_region): Remove.
13619 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13621 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13622 mode attribute for TARGET_AVX512VL.
13624 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13626 * config/i386/i386.opt (mprefer-avx256): New option.
13627 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13629 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13630 for TARGET_PREFER_AVX256.
13631 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13633 2017-09-21 Jeff Law <law@redhat.com>
13635 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13636 Fix dump output if the only stack space is for pushed registers.
13638 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13640 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13643 2017-09-21 Martin Sebor <msebor@redhat.com>
13646 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13647 code (in C++) or code that triggers warnings.
13649 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13651 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13653 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13655 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13656 * sched-rgn.c: Ditto.
13657 * sel-sched-ir.c: Ditto.
13659 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13661 * toplev.h (set_random_seed): Adjust return type.
13662 * toplev.c (init_local_tick): Move eager initialization of random_seed
13663 to get_random_seed. Adjust comment.
13664 (init_random_seed): Inline to get_random_seed, delete.
13665 (get_random_seed): Initialize random_seed lazily.
13666 (set_random_seed): Do not return previous value.
13667 (print_switch_value): Do not call get_random_seed.
13669 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13671 * cgraph.c (delete_function_version): New, broken out from...
13672 (cgraph_node::delete_function_version): ...here. Rename to
13673 cgraph_node::delete_function_version_by_decl. Update all uses.
13674 (cgraph_node::remove): Call delete_function_version.
13676 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13679 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13680 VAR_DECLs to which addressable non-volatile parameters are mapped
13681 and for id->retvar after the return value assignment. Clear
13682 id->retval and id->retbnd after inlining.
13684 2017-09-21 Richard Biener <rguenther@suse.de>
13686 PR tree-optimization/82276
13687 PR tree-optimization/82244
13688 * tree-vrp.c (build_assert_expr_for): Set
13689 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13691 (remove_range_assertions): Revert earlier change.
13693 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13696 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13698 2017-09-21 Richard Biener <rguenther@suse.de>
13700 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13701 Restore valid IL after code generation errors.
13702 * graphite.c (graphite_transform_loops): Diagnose code
13703 generation issues as MSG_MISSED_OPTIMIZATION and continue
13704 with processing SCOPs.
13706 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13707 Alan Hayward <alan.hayward@arm.com>
13708 David Sherwood <david.sherwood@arm.com>
13710 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13711 rather than choosing between plus_constant and gen_rtx_<CODE>.
13712 * expr.c (emit_push_insn): Likewise.
13713 (expand_expr_real_2): Likewise.
13715 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13716 Alan Hayward <alan.hayward@arm.com>
13717 David Sherwood <david.sherwood@arm.com>
13719 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13721 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13722 Alan Hayward <alan.hayward@arm.com>
13723 David Sherwood <david.sherwood@arm.com>
13725 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13726 calling tree_to_uhwi.
13728 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13729 Alan Hayward <alan.hayward@arm.com>
13730 David Sherwood <david.sherwood@arm.com>
13732 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13733 INTEGER_CST rather than a negative test for ADDR_EXPR.
13735 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13736 Alan Hayward <alan.hayward@arm.com>
13737 David Sherwood <david.sherwood@arm.com>
13739 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13740 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13742 2017-09-21 Richard Biener <rguenther@suse.de>
13744 PR tree-optimization/71351
13745 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13746 graphite_create_new_loop_guard): Remove, fold remaining parts
13748 (translate_isl_ast_node_for): ... here and simplify.
13750 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13753 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13754 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13755 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13756 alternative always use QI mode, for -Os imov (=R,R) alternative
13757 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13760 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13761 Jeff Law <law@redhat.com>
13763 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13764 (allocate_stack_space): New function, partially extracted from
13765 s390_emit_prologue.
13766 (s390_emit_prologue): Track offset to most recent stack probe.
13767 Code to allocate space moved into allocate_stack_space.
13768 Dump actions when no stack is allocated.
13769 (s390_prologue_plus_offset): New function.
13770 (s390_emit_stack_probe): Likewise.
13772 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13774 * common.opt (Wa, Wl, Wp, g, gz=): Add
13776 (gno-column-info): Remove.
13777 (gcolumn-info): Drop RejectNegative.
13778 (gno-): New prefix.
13779 (gno-record-gcc-switches): Remove.
13780 (grecord-gcc-switches): Drop RejectNegative.
13781 (gno-split-dwarf): Remove.
13782 (gsplit-dwarf): Drop RejectNegative.
13783 (gno-strict-dwarf): Remove.
13784 (gstrict-dwarf): Drop RejectNegative.
13785 * config/darwin.opt (gfull, gused): Add RejectNegative.
13786 * dwarf2out.c (gen_producer_string): Drop
13787 gno-record-gcc-switches handler.
13788 * optc-gen.awk: Add g to prefixes with negative forms.
13789 * opts-common.c (remapping_prefix_p): New.
13790 (find_opt): Check it.
13791 (generate_canonical_option): Test g prefix.
13792 (option_map): Add -gno- mapping.
13793 (add_misspelling_candidates): Check remapping_prefix_p.
13795 2017-09-20 Jeff Law <law@redhat.com>
13797 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13798 thinko in stack clash protection support.
13800 * explow.c (compute_stack_clash_protection_loop_data): Use
13801 CONST_INT_P instead of explicit test. Verify object is a
13802 CONST_INT_P before looking at INTVAL.
13803 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13804 instead of explicit test.
13806 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13809 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13810 address instead of to r1 and r11.
13812 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13814 * config.gcc: Support "knm".
13815 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13816 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13818 * config/i386/i386.c (m_KNM): Define.
13819 (processor_target_table): Add "knm".
13821 (ix86_option_override_internal): Add "knm".
13822 (ix86_issue_rate): Add PROCESSOR_KNM.
13823 (ix86_adjust_cost): Ditto.
13824 (ia32_multipass_dfa_lookahead): Ditto.
13825 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13826 (fold_builtin_cpu): Add M_INTEL_KNM.
13827 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13828 (processor_type): Add PROCESSOR_KNM.
13829 * config/i386/x86-tune.def: Add m_KNM.
13830 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13832 2017-09-20 Richard Biener <rguenther@suse.de>
13834 PR tree-optimization/80213
13835 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13836 are allowed in empty BBs as well.
13837 (canonicalize_loop_closed_ssa): Also look for other complex
13839 (scop_detection::get_sese): Include the loop-closed PHI block
13841 (scop_detection::merge_sese): Remove code adding extra blocks.
13842 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13843 (build_scops): Assert the final returned scop is invalid.
13845 2017-09-20 Richard Biener <rguenther@suse.de>
13847 PR tree-optimization/82264
13848 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13850 (vn_phi_lookup): Likewise.
13851 (vn_phi_insert): Likewise.
13853 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13855 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13856 that fits into uhwi or shwi, add DW_AT_const_value regardless
13857 of early_dwarf without going through RTL, using add_AT_unsigned
13860 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13861 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13862 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13863 (DEBUG_MACRO_SECTION): Likewise.
13864 (DEBUG_LTO_MACRO_SECTION): Likewise.
13865 (DEBUG_STR_DWO_SECTION): Likewise.
13866 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13867 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13868 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13869 (DEBUG_LTO_LINE_STR_SECTION): Define.
13870 (init_sections_and_labels): Initialize debug_line_str_section
13871 variable. Initialize debug_loc_section for -gdwarf-5 to
13872 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13874 2017-09-20 Richard Biener <rguenther@suse.de>
13876 * graphite-sese-to-poly.c (extract_affine): Properly handle
13877 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13879 2017-09-20 Richard Biener <rguenther@suse.de>
13881 PR tree-optimization/81373
13882 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13883 Force SESE live-out defs to be handled even if they are
13886 2017-09-19 Jeff Law <law@redhat.com>
13888 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13889 nothing for stack adjustments with REG_STACK_CHECK.
13890 * sched-deps.c (parse_add_or_inc): Reject insns with
13891 REG_STACK_CHECK from dependency breaking.
13892 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13893 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13894 * reg-notes.def (STACK_CHECK): New note.
13896 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13897 (ix86_expand_prologue): Dump stack clash info as needed.
13898 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13900 * function.c (dump_stack_clash_frame_info): New function.
13901 * function.h (dump_stack_clash_frame_info): Prototype.
13902 (enum stack_clash_probes): New enum.
13904 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13905 flag_stack_clash_protection.
13906 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13907 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13908 (arm_frame_pointer_required): Likewise.
13909 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13910 (ia64_expand_prologue): Likewise.
13911 * config/mips/mips.c (mips_expand_prologue): Likewise.
13912 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13913 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13914 (sparc_flat_expand_prologue): Likewise.
13915 * config/spu/spu.c (spu_expand_prologue): Likewise.
13917 * explow.c: Include "params.h".
13918 (anti_adjust_stack_and_probe_stack_clash): New function.
13919 (get_stack_check_protect): Likewise.
13920 (compute_stack_clash_protection_loop_data): Likewise.
13921 (emit_stack_clash_protection_loop_start): Likewise.
13922 (emit_stack_clash_protection_loop_end): Likewise.
13923 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13924 Use anti_adjust_stack_and_probe_stack_clash.
13925 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13926 (emit_stack_clash_protection_loop_start): Likewise.
13927 (emit_stack_clash_protection_loop_end): Likewise.
13928 * rtl.h (get_stack_check_protect): Prototype.
13929 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13930 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13931 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13933 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13935 * doc/tm.texi: Rebuilt.
13936 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13937 get_stack_check_protect.
13938 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13939 * config/arm/arm.c (arm_expand_prologue): Likewise.
13940 (arm_frame_pointer_required): Likewise.
13941 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13942 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13943 * config/mips/mips.c (mips_expand_prologue): Likewise.
13944 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13945 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13946 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13947 (sparc_flat_expand_prologue): Likewise.
13949 * common.opt (-fstack-clash-protection): New option.
13950 * flag-types.h (enum stack_check_type): Note difference between
13951 -fstack-check= and -fstack-clash-protection.
13952 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13953 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13954 * toplev.c (process_options): Issue warnings/errors for cases
13955 not handled with -fstack-clash-protection.
13956 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13957 (-fstack-check): Note additional problem with -fstack-check=generic.
13958 Note that -fstack-check is primarily for Ada and refer users
13959 to -fstack-clash-protection for stack-clash-protection.
13960 Document new params for stack clash protection.
13962 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13964 * config/i386/i386.c (ix86_split_long_move): Do not handle
13965 address used for LEA in a special way.
13967 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13969 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13971 2017-09-19 Martin Sebor <msebor@redhat.com>
13974 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13975 of incompatible types.
13977 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13979 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13980 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13981 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13982 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13984 2017-09-19 Richard Biener <rguenther@suse.de>
13986 PR tree-optimization/82244
13987 * tree-vrp.c (remove_range_assertions): Do not propagate
13988 a constant to abnormals but replace the assert with a copy.
13990 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13992 PR rtl-optimization/57878
13993 PR rtl-optimization/68988
13994 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13995 avoidance test involving non_reload_pseudos. Move frequency test
13996 below the general fragmentation avoidance test.
13998 2017-09-19 Richard Biener <rguenther@suse.de>
14000 PR tree-optimization/69728
14001 * graphite-sese-to-poly.c (schedule_error): New global.
14002 (add_loop_schedule): Handle empty domain by failing the
14004 (build_original_schedule): Handle schedule_error.
14006 2017-09-19 Richard Biener <rguenther@suse.de>
14008 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14009 Do not iterate to sibling loops but only to siblings of inner
14012 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14015 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14016 that the registers are different.
14018 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14020 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14021 to processor_model and "amdfam17h" to arch_names_table.
14022 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14024 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14027 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14029 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14035 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14037 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14038 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14039 vect_schedule_slp_instance.
14040 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14041 Update call to vect_slp_analyze_node_operations. Simplify return
14043 (vect_slp_analyze_bb_1): Update call accordingly.
14044 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14045 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14046 (vect_schedule_slp): Update call accordingly.
14048 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14049 Alan Hayward <alan.hayward@arm.com>
14050 David Sherwood <david.sherwood@arm.com>
14052 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14053 with types that aren't in fact scalar.
14055 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14057 * tree-vect-slp.c (vect_record_max_nunits): New function,
14059 (vect_build_slp_tree_1): ...here.
14060 (vect_build_slp_tree_2): Call it for phis too.
14062 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14064 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14065 to vect_get_vec_def_for_operand when getting the mask operand.
14067 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14068 Alan Hayward <alan.hayward@arm.com>
14069 David Sherwood <david.sherwood@arm.com>
14071 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14074 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14075 Alan Hayward <alan.hayward@arm.com>
14076 David Sherwood <david.sherwood@arm.com>
14078 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14079 calculation for vector booleans.
14081 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14082 Alan Hayward <alan.hayward@arm.com>
14083 David Sherwood <david.sherwood@arm.com>
14085 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14087 (vect_transform_stmt): ...here.
14088 (vect_analyze_stmt): Use it instead of calling
14089 vectorizable_live_operation directly.
14091 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14093 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14094 non-SIMT targets in acc vector loops.
14096 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14098 * configure.ac: Add arc and check if assembler supports gdwarf2.
14099 * configure: Regenerate.
14101 2017-09-18 Richard Biener <rguenther@suse.de>
14103 PR tree-optimization/82220
14104 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14105 epilogue niters from the min_profitable_iters compute.
14107 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14110 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14111 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14112 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14114 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14117 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14118 switching to a new text section.
14120 2017-09-18 Richard Biener <rguenther@suse.de>
14122 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14124 (build_alias_set): Reject aliases with no access function.
14126 2017-09-18 Richard Biener <rguenther@suse.de>
14128 PR tree-optimization/79622
14129 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14131 (build_cross_bb_scalars_use): Likewise.
14133 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14135 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14137 2017-09-18 Alan Modra <amodra@gmail.com>
14140 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14141 stack_pointer_rtx for count 0. Update comments. Break up
14142 large rtl expression.
14144 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14146 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14147 Increase to 20 bytes.
14148 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14149 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14150 or avx version of the stub.
14152 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14155 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14156 compute the minimum stack alignment. Also update preferred stack
14157 boundary for leaf functions.
14159 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14161 PR tree-optimization/82228
14162 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14165 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14167 * common/config/nds32/nds32-common.c
14168 (nds32_option_optimization_table): Refine formatting.
14169 (nds32_option_optimization_table): Use -fsched-pressure and
14170 -fomit-frame-pointer for specific optimization level.
14172 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14174 * config/nds32/nds32.c: Refine formatting and comments.
14175 * config/nds32/nds32.h: Likewise.
14176 * config/nds32/nds32.md: Likewise.
14177 * config/nds32/nds32-cost.c: Likewise.
14178 * config/nds32/nds32-isr.c: Likewise.
14179 * config/nds32/nds32-md-auxiliary.c: Likewise.
14180 * config/nds32/nds32-multiple.md: Likewise.
14181 * config/nds32/nds32-predicates.c: Likewise.
14183 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14184 Jakub Jelinek <jakub@redhat.com>
14186 Add support for -std=c++2a.
14187 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14189 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14191 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14194 * doc/extend.texi (Common Function Attributes): Add
14195 references to ARM, AArch64, and S/390 specific attributes.
14196 (Function Specific Option Pragmas): Add AArch64 and S/390
14197 to list of back ends that support the target pragma.
14199 2017-09-15 Nathan Sidwell <nathan@acm.org>
14201 * doc/standards.texi: Fix C++17 description. Update URLs for
14204 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14206 * common.opt (Wcast-align=strict): New warning option.
14207 * doc/invoke.texi: Document -Wcast-align=strict.
14209 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14211 * cgraph.h (cgraph_thunk_info): Add comments.
14212 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14213 assert for VIRTUAL_* arguments stricter.
14215 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14217 PR tree-optimization/71026
14218 * match.pd: Move RDIV patterns from fold-const.c
14219 * fold-const.c (distribute_real_division): Removed.
14220 (fold_binary_loc): Remove calls to distribute_real_divison.
14222 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14224 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14225 c++1z and gnu++1z as deprecated. Change other references to
14226 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14227 Change -Wc++1z-compat to -Wc++17-compat.
14228 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14229 * dwarf2out.c (highest_c_language): Handle C++17.
14230 (gen_compile_unit_die): Likewise.
14232 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14234 PR rtl-optimization/82192
14235 * combine.c (make_extraction): Don't look through non-paradoxical
14236 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14239 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14240 Alan Hayward <alan.hayward@arm.com>
14241 David Sherwood <david.sherwood@arm.com>
14243 * target.def (function_arg_offset): New hook.
14244 * targhooks.h (default_function_arg_offset): Declare.
14245 * targhooks.c (default_function_arg_offset): New function.
14246 * function.c (locate_and_pad_parm): Use
14247 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14248 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14249 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14250 * doc/tm.texi: Regenerate.
14251 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14252 * config/spu/spu.c (spu_function_arg_offset): New function.
14253 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14254 * system.h (FUNCTION_ARG_OFFSET): Poison.
14256 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14257 Alan Hayard <alan.hayward@arm.com>
14258 David Sherwood <david.sherwood@arm.com>
14260 * target.def (truly_noop_truncation): New hook.
14261 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14262 than TRULY_NOOP_TRUNCATION.
14263 * hooks.h (hook_bool_uint_uint_true): Declare.
14264 * hooks.c (hook_bool_uint_uint_true): New function.
14265 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14266 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14267 * doc/tm.texi: Regenerate.
14268 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14269 rather than TRULY_NOOP_TRUNCATION in comments.
14270 (simplify_comparison): Likewise.
14271 (record_truncated_value): Likewise.
14272 * expmed.c (extract_bit_field_1): Likewise.
14273 (extract_split_bit_field): Likewise.
14274 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14275 instead of TRULY_NOOP_TRUNCATION.
14276 * function.c (assign_parm_setup_block): Likewise.
14277 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14278 * rtlhooks.c: Include target.h.
14279 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14280 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14281 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14282 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14283 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14284 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14285 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14286 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14287 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14288 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14289 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14290 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14291 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14292 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14293 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14294 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14295 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14296 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14297 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14298 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14299 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14300 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14301 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14302 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14303 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14304 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14305 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14306 rather than TRULY_NOOP_TRUNCATION in comments.
14307 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14308 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14309 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14310 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14311 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14312 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14313 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14314 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14315 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14316 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14317 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14318 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14319 rather than TRULY_NOOP_TRUNCATION in comments.
14320 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14321 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14322 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14323 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14324 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14325 TRULY_NOOP_TRUNCATION condition.
14326 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14327 (TRULY_NOOP_TRUNCATION): Delete.
14328 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14329 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14330 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14331 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14332 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14333 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14334 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14335 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14336 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14337 rather than TRULY_NOOP_TRUNCATION in comments.
14338 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14340 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14341 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14342 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14343 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14345 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14348 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14349 (*cmp_ior): Likewise.
14350 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14351 (*ior_scc_scc_cmp): Likewise.
14352 (*and_scc_scc): Likewise.
14353 (*and_scc_scc_cmp): Likewise.
14355 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14356 Alan Hayard <alan.hayward@arm.com>
14357 David Sherwood <david.sherwood@arm.com>
14359 * target.def (can_change_mode_class): New hook.
14360 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14361 (hard_regno_nregs): Likewise.
14362 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14363 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14364 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14365 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14366 (LOAD_EXTEND_OP): Update accordingly.
14367 * doc/tm.texi: Regenerate.
14368 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14369 CANNOT_CHANGE_MODE_CLASS.
14370 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14371 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14372 * combine.c (simplify_set): Update accordingly.
14373 * emit-rtl.c (validate_subreg): Likewise.
14374 * recog.c (general_operand): Likewise.
14375 * regcprop.c (mode_change_ok): Likewise.
14376 * reload1.c (choose_reload_regs): Likewise.
14377 (inherit_piecemeal_p): Likewise.
14378 * rtlanal.c (simplify_subreg_regno): Likewise.
14379 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14380 instead of CANNOT_CHANGE_MODE_CLASS.
14381 (reload_cse_simplify_operands): Likewise.
14382 * reload.c (push_reload): Use targetm.can_change_mode_class
14383 instead of CANNOT_CHANGE_MODE_CLASS.
14384 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14385 REG_CANNOT_CHANGE_MODE_P.
14386 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14387 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14388 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14389 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14390 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14391 (arm_can_change_mode_class): New function.
14392 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14393 than CANNOT_CHANGE_MODE_CLASS in comments.
14394 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14395 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14396 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14397 (ix86_can_change_mode_class): ...this new function, inverting the
14398 sense of the return value.
14399 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14400 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14401 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14402 (ia64_can_change_mode_class): New function.
14403 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14404 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14405 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14406 (m32c_can_change_mode_class): ...this new function, inverting the
14407 sense of the return value.
14408 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14409 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14410 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14411 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14412 (mips_can_change_mode_class): ...this new function, inverting the
14413 sense of the return value.
14414 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14415 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14416 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14417 (msp430_can_change_mode_class): New function.
14418 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14419 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14420 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14421 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14422 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14423 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14424 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14425 (pa_cannot_change_mode_class): Replace with...
14426 (pa_can_change_mode_class): ...this new function, inverting the
14427 sense of the return value.
14428 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14429 than CANNOT_CHANGE_MODE_CLASS in comments.
14430 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14431 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14432 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14433 (pdp11_cannot_change_mode_class): Replace with...
14434 (pdp11_can_change_mode_class): ...this new function, inverting the
14435 sense of the return value.
14436 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14437 * config/powerpcspe/powerpcspe-protos.h
14438 (rs6000_cannot_change_mode_class_ptr): Delete.
14439 * config/powerpcspe/powerpcspe.c
14440 (rs6000_cannot_change_mode_class_ptr): Delete.
14441 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14442 (rs6000_option_override_internal): Assign to
14443 targetm.can_change_mode_class instead of
14444 rs6000_cannot_change_mode_class_ptr.
14445 (rs6000_cannot_change_mode_class): Replace with...
14446 (rs6000_can_change_mode_class): ...this new function, inverting the
14447 sense of the return value.
14448 (rs6000_debug_cannot_change_mode_class): Replace with...
14449 (rs6000_debug_can_change_mode_class): ...this new function.
14450 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14451 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14452 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14453 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14454 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14456 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14457 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14458 (rs6000_option_override_internal): Assign to
14459 targetm.can_change_mode_class instead of
14460 rs6000_cannot_change_mode_class_ptr.
14461 (rs6000_cannot_change_mode_class): Replace with...
14462 (rs6000_can_change_mode_class): ...this new function, inverting the
14463 sense of the return value.
14464 (rs6000_debug_cannot_change_mode_class): Replace with...
14465 (rs6000_debug_can_change_mode_class): ...this new function.
14466 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14467 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14468 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14469 (s390_can_change_mode_class): ...this new function, inverting the
14470 sense of the return value.
14471 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14472 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14473 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14474 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14475 (sh_cannot_change_mode_class): Replace with...
14476 (sh_can_change_mode_class): ...this new function, inverting the
14477 sense of the return value.
14478 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14479 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14480 (sparc_can_change_mode_class): New function.
14481 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14482 * config/spu/spu.c (spu_can_change_mode_class): New function.
14483 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14484 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14485 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14486 (visium_can_change_mode_class): New function.
14487 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14489 2017-09-15 Richard Biener <rguenther@suse.de>
14491 PR tree-optimization/82217
14492 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14493 but not undefined case.
14495 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14498 * postreload.c (reload_cse_simplify_operands): Skip
14499 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14501 2017-09-15 Richard Biener <rguenther@suse.de>
14503 PR tree-optimization/68823
14504 * graphite-scop-detection.c (build_alias_set): If we have a
14505 possible dependence check whether we can handle them by just
14506 looking at the DRs DR_ACCESS_FNs.
14507 (build_scops): If build_alias_set fails, fail the SCOP.
14509 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14511 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14512 to support float128 built-in functions that require the ISA 3.0
14514 (BU_FLOAT128_3_HW): Likewise.
14515 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14516 built-in functions.
14517 (FMAF128): Likewise.
14519 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14520 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14521 floating point instructions.
14522 (rs6000_invalid_builtin): Likewise.
14523 (rs6000_builtin_mask_names): Likewise.
14524 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14525 (RS6000_BTM_FLOAT128_HW): Likewise.
14526 (RS6000_BTM_COMMON): Likewise.
14527 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14529 * doc/extend.texi (RS/6000 built-in functions): Document the
14530 IEEE 128-bit floating point square root and fused multiply-add
14531 built-in functions.
14533 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14535 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14536 reg (r2) isn't in the set of registers defined in the prologue.
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 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14543 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14544 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14545 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14546 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14548 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14549 max_vectorization_factor.
14550 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14552 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14553 Alan Hayward <alan.hayward@arm.com>
14554 David Sherwood <david.sherwood@arm.com>
14556 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14557 (vect_worthwhile_without_simd_p): Declare.
14558 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14559 (vectorizable_reduction): Use it.
14560 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14561 (vectorizable_operation): Likewise.
14563 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14564 Alan Hayward <alan.hayward@arm.com>
14565 David Sherwood <david.sherwood@arm.com>
14567 * tree-vectorizer.h (vect_get_num_copies): New function.
14568 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14569 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14570 (vectorizable_induction): Likewise.
14571 (vectorizable_live_operation): Likewise.
14572 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14573 (vectorizable_bswap): Likewise.
14574 (vectorizable_call): Likewise.
14575 (vectorizable_conversion): Likewise.
14576 (vectorizable_assignment): Likewise.
14577 (vectorizable_shift): Likewise.
14578 (vectorizable_operation): Likewise.
14579 (vectorizable_store): Likewise.
14580 (vectorizable_load): Likewise.
14581 (vectorizable_condition): Likewise.
14582 (vectorizable_comparison): Likewise.
14583 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14585 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14586 Alan Hayward <alan.hayward@arm.com>
14587 David Sherwood <david.sherwood@arm.com>
14589 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14590 of vect_init_vector.
14592 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14593 Alan Hayward <alan.hayward@arm.com>
14594 David Sherwood <david.sherwood@arm.com>
14596 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14597 an inline wrapper that provides a location.
14598 (gimple_build_vector): Likewise.
14599 * gimple-fold.c (gimple_build_vector_from_val): New function.
14600 (gimple_build_vector): Likewise.
14601 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14602 functions to build the initial value. Always return a gimple value.
14603 (get_initial_defs_for_reduction): Likewise. Only compute
14605 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14606 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14607 (vectorizable_induction): Use gimple_build_vector rather than
14610 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14611 Alan Hayward <alan.hayward@arm.com>
14612 David Sherwood <david.sherwood@arm.com>
14614 * target.h (vec_perm_indices): New typedef.
14615 (auto_vec_perm_indices): Likewise.
14616 * optabs-query.h: Include target.h
14617 (can_vec_perm_p): Take a vec_perm_indices *.
14618 * optabs-query.c (can_vec_perm_p): Likewise.
14619 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14620 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14621 * tree-vect-generic.c (lower_vec_perm): Likewise.
14622 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14623 (vect_grouped_load_supported): Likewise.
14624 (vect_shift_permute_load_chain): Likewise.
14625 (vect_permute_store_chain): Use auto_vec_perm_indices.
14626 (vect_permute_load_chain): Likewise.
14627 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14628 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14629 Update uses of can_vec_perm_p.
14630 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14631 mode with a number of elements. Take a vec_perm_indices *.
14632 (vect_create_epilog_for_reduction): Update accordingly.
14633 Use auto_vec_perm_indices.
14634 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14635 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14636 (vect_transform_slp_perm_load): Likewise.
14637 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14638 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14639 (vect_gen_perm_mask_checked): Likewise.
14640 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14641 (vect_gen_perm_mask_checked): Likewise.
14642 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14643 (vectorizable_store): Likewise.
14644 (vectorizable_load): Likewise.
14645 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14646 (vectorizable_bswap): Likewise.
14648 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14649 Alan Hayward <alan.hayward@arm.com>
14650 David Sherwood <david.sherwood@arm.com>
14652 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14653 * tree.c (build_vector): Likewise.
14654 (build_vector_from_ctor): Update accordingly.
14655 (build_vector_from_val): Likewise.
14656 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14657 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14658 * tree-vect-generic.c (add_rshift): Likewise.
14659 (expand_vector_divmod): Likewise.
14660 (optimize_vector_constructor): Likewise.
14661 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14662 (vect_transform_slp_perm_load): Likewise.
14663 (vect_schedule_slp_instance): Likewise.
14664 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14665 (vectorizable_call): Likewise.
14666 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14667 * expmed.c (make_tree): Likewise.
14668 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14669 a vector passed to build_vector.
14670 (fold_convert_const): Likewise.
14671 (exact_inverse): Likewise.
14672 (fold_ternary_loc): Likewise.
14673 (fold_relational_const): Likewise.
14674 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14675 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14676 (const_unop): Likewise. Store the reduction accumulator in a
14677 variable rather than an array.
14678 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14679 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14680 the new vector, rather than constructing it after the input arrays.
14681 (native_interpret_vector): Use auto_vec<tree> when building
14682 a vector passed to build_vector. Add elements in order.
14683 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14684 auto_vec<tree> when building a vector passed to build_vector.
14685 (vect_create_epilog_for_reduction): Likewise.
14686 (vectorizable_induction): Likewise.
14687 (get_initial_def_for_reduction): Likewise. Fix indentation of
14689 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14691 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14692 passed to build_vector.
14694 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14695 Alan Hayward <alan.hayward@arm.com>
14696 David Sherwood <david.sherwood@arm.com>
14698 * tree-core.h (tree_base::u): Add an "nelts" field.
14699 (tree_vector): Use VECTOR_CST_NELTS as the length.
14700 * tree.c (tree_size): Likewise.
14701 (make_vector): Initialize VECTOR_CST_NELTS.
14702 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14703 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14704 TYPE_VECTOR_SUBPARTS.
14705 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14706 as the number of units, setting it from VECTOR_CST_NELTS.
14707 (const_vector_from_tree): Likewise.
14708 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14709 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14710 (fold_negate_expr_1): Likewise.
14711 (fold_convert_const): Likewise.
14712 (const_binop): Likewise. Differentiate the number of output and
14714 (const_unop): Likewise.
14715 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14716 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14717 in cases that did the opposite.
14719 2017-09-14 Richard Biener <rguenther@suse.de>
14721 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14724 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14726 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14728 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14731 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14732 if and where to split a bb, except for splitting before debug insn
14733 sequences followed by non-label real insn. Delete debug insns
14734 in between basic blocks.
14736 * combine.c (make_compound_operation_int): Formatting fixes.
14738 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14739 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14740 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14741 * config/sol2.h (LINK_EH_SPEC): Likewise.
14742 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14743 * config/s390/linux.h (LINK_SPEC): Likewise.
14744 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14745 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14746 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14747 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14748 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14749 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14750 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14751 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14753 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14755 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14757 (ENDFILE_LINUX_SPEC): Likewise.
14758 (LINK_EH_SPEC): Likewise.
14759 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14760 (LINK_OS_LINUX_SPEC32): Likewise.
14761 (LINK_OS_LINUX_SPEC64): Likewise.
14762 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14763 (LINK_OS_LINUX_SPEC): Likewise.
14765 2017-09-13 Martin Liska <mliska@suse.cz>
14767 PR middle-end/82154
14768 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14769 CASE_HIGH is NULL_TREE.
14771 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14772 Alan Hayward <alan.hayward@arm.com>
14773 David Sherwood <david.sherwood@arm.com>
14775 * target.def (secondary_memory_needed): New hook.
14776 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14777 instead of SECONDARY_MEMORY_NEEDED.
14778 (secondary_memory_needed_mode): Likewise.
14779 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14780 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14781 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14782 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14783 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14784 * doc/tm.texi: Regenerate.
14785 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14786 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14787 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14788 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14789 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14790 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14791 mode argument first and change the reg_class arguments to reg_class_t.
14792 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14793 Make static. Update the call to inline_secondary_memory_needed.
14794 (ix86_register_move_cost): Update the call to
14795 inline_secondary_memory_needed.
14796 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14797 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14799 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14800 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14802 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14803 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14804 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14805 and match hook interface. Add comment from mips.h.
14806 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14807 * config/mmix/mmix.md (truncdfsf2): Refer to
14808 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14810 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14811 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14812 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14813 (pa_secondary_memory_needed): New function.
14814 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14815 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14816 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14817 (pdp11_secondary_memory_needed): Make static and match hook interface.
14818 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14819 * config/powerpcspe/powerpcspe-protos.h
14820 (rs6000_secondary_memory_needed_ptr): Delete.
14821 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14823 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14824 (rs6000_option_override_internal): Assign to
14825 targetm.secondary_memory_needed rather than
14826 rs6000_secondary_memory_needed_ptr.
14827 (rs6000_secondary_memory_needed): Match hook interface.
14828 (rs6000_debug_secondary_memory_needed): Likewise.
14829 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14830 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14831 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14832 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14833 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14834 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14836 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14837 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14838 (rs6000_option_override_internal): Assign to
14839 targetm.secondary_memory_needed rather than
14840 rs6000_secondary_memory_needed_ptr.
14841 (rs6000_secondary_memory_needed): Match hook interface.
14842 (rs6000_debug_secondary_memory_needed): Likewise.
14843 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14844 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14845 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14846 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14847 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14848 (sparc_secondary_memory_needed): New function.
14849 * lra-constraints.c (check_and_process_move): Refer to
14850 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14852 (curr_insn_transform): Likewise.
14853 (process_alt_operands): Use targetm.secondary_memory_needed
14854 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14855 (check_secondary_memory_needed_p): Likewise.
14856 (choose_split_class): Likewise.
14857 * reload.c: Unconditionally include code that was previously
14858 conditional on SECONDARY_MEMORY_NEEDED.
14859 (push_secondary_reload): Use targetm.secondary_memory_needed
14860 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14861 (push_reload): Likewise.
14862 * reload1.c: Unconditionally include code that was previously
14863 conditional on SECONDARY_MEMORY_NEEDED.
14864 (choose_reload_regs): Use targetm.secondary_memory_needed
14865 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14866 (gen_reload): Likewise.
14867 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14869 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14870 Alan Hayward <alan.hayward@arm.com>
14871 David Sherwood <david.sherwood@arm.com>
14873 * target.def (secondary_memory_needed_mode): New hook:
14874 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14875 * targhooks.h (default_secondary_memory_needed_mode): New function.
14876 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14877 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14878 * doc/tm.texi: Regenerate.
14879 * lra-constraints.c (check_and_process_move): Use
14880 targetm.secondary_memory_needed_mode instead of
14881 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14882 (curr_insn_transform): Likewise.
14883 * reload.c (get_secondary_mem): Likewise.
14884 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14885 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14887 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14888 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14889 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14890 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14891 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14893 * config/powerpcspe/powerpcspe-protos.h
14894 (rs6000_secondary_memory_needed_mode): Delete.
14895 * config/powerpcspe/powerpcspe.c
14896 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14897 (rs6000_secondary_memory_needed_mode): Make static.
14898 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14899 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14901 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14903 (rs6000_secondary_memory_needed_mode): Make static.
14904 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14905 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14906 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14907 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14908 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14910 (sparc_secondary_memory_needed_mode): New function.
14911 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14913 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14915 * config/aarch64/constraints.md (Umq): New constraint.
14916 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14918 (mov<mode>): Update condition.
14920 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14922 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14923 when bitposition is the same.
14925 2017-09-13 Richard Biener <rguenther@suse.de>
14927 * dwarf2out.c (output_die_symbol): Remove.
14928 (output_die): Do not output a DIEs symbol.
14930 2017-09-13 Richard Biener <rguenther@suse.de>
14932 PR middle-end/82128
14933 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14934 default-def to avoid breaking iterator update with the weird
14935 interaction with cgraph_update_edges_for_call_stmt_node.
14937 2017-09-13 Richard Biener <rguenther@suse.de>
14939 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14940 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14941 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14942 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14944 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14946 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14947 Disable pc relative literal load irrespective of
14948 TARGET_FIX_ERR_A53_84341 for default.
14950 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14952 * config/sparc/sparc.c (output_return): Output the source location of
14953 the insn in the delay slot, if any.
14954 (output_sibcall): Likewise.
14956 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14959 * common.opt (-static-pie): New alias.
14960 (shared): Negate static-pie.
14961 (-no-pie): Update help text.
14963 (static-pie): New option.
14964 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14965 -static-pie support.
14966 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14967 (LINK_EH_SPEC): Likewise.
14968 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14969 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14970 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14971 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14972 (init_gcc_specs): Likewise.
14973 (init_spec): Likewise.
14974 (display_help): Update help message for -pie.
14975 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14978 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14980 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14981 (movdi_aarch64): Likewise.
14982 (movti_aarch64): Likewise.
14984 2017-09-12 Simon Wright <simon@pushface.org>
14987 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14988 calculation of the minor version, always output as 0.
14990 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14993 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14994 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14995 on it early, rather than manual conversion late. For
14996 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14997 instead of performing manual conversion.
14999 2017-09-12 Carl Love <cel@us.ibm.com>
15001 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15002 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15004 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15005 VMULOSW): Add definitions.
15006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15007 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15008 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15009 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15010 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15012 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15014 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15016 (movti_aarch64): Likewise.
15017 (movdf_aarch64): Likewise.
15018 (movtf_aarch64): Likewise.
15019 (load_pairdi): Likewise.
15020 (store_pairdi): Likewise.
15021 (load_pairdf): Likewise.
15022 (store_pairdf): Likewise.
15023 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15024 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15025 (ldr_got_small_<mode>): Likewise.
15026 (ldr_got_small_28k_<mode>): Likewise.
15027 (ldr_got_tiny): Likewise.
15028 * config/aarch64/iterators.md (ldst_sz): New.
15029 (ldpstp_sz): Likewise.
15030 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15032 (thunderx_load): Split load_8 to load_16.
15033 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15035 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15036 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15037 (xgene1_store_pair): Split store_8 to store_16.
15038 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15039 (falkor_st_0_st_sd): Split store_8 to store_16.
15041 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15043 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15044 and store1/2/3/4 to store_4/8/12/16.
15045 * config/aarch64/aarch64.md: Update for rename.
15046 * config/arm/arm.md: Likewise.: Likewise.
15047 * config/arm/arm.c: Likewise.
15048 * config/arm/thumb1.md: Likewise.
15049 * config/arm/thumb2.md: Likewise.
15050 * config/arm/vfp.md: Likewise.
15051 * config/arm/arm-generic.md: Likewise.
15052 * config/arm/arm1020e.md: Likewise.
15053 * config/arm/arm1026ejs.md: Likewise.
15054 * config/arm/arm1136jfs.md: Likewise.
15055 * config/arm/arm926ejs.md: Likewise.
15056 * config/arm/cortex-a15.md: Likewise.
15057 * config/arm/cortex-a17.md: Likewise.
15058 * config/arm/cortex-a5.md: Likewise.
15059 * config/arm/cortex-a53.md: Likewise.
15060 * config/arm/cortex-a57.md: Likewise.
15061 * config/arm/cortex-a7.md: Likewise.
15062 * config/arm/cortex-a8.md: Likewise.
15063 * config/arm/cortex-a9.md: Likewise.
15064 * config/arm/cortex-m4.md: Likewise.
15065 * config/arm/cortex-m7.md: Likewise.
15066 * config/arm/cortex-r4.md: Likewise.
15067 * config/arm/exynos-m1.md: Likewise.
15068 * config/arm/fa526.md: Likewise.
15069 * config/arm/fa606te.md: Likewise.
15070 * config/arm/fa626te.md: Likewise.
15071 * config/arm/fa726te.md: Likewise.
15072 * config/arm/fmp626.md: Likewise.
15073 * config/arm/iwmmxt.md: Likewise.
15074 * config/arm/ldmstm.md: Likewise.
15075 * config/arm/marvell-pj4.md: Likewise.
15076 * config/arm/xgene1.md: Likewise.
15077 * config/aarch64/thunderx.md: Likewise.
15078 * config/aarch64/thunderx2t99.md: Likewise.
15079 * config/aarch64/falkor.md: Likewise.
15081 2017-09-12 Martin Liska <mliska@suse.cz>
15083 * attribs.c (private_lookup_attribute): New function.
15084 * attribs.h (private_lookup_attribute): Declared here.
15085 (lookup_attribute): Called from this place.
15087 2017-09-12 Richard Biener <rguenther@suse.de>
15089 PR tree-optimization/82157
15090 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15091 stmts with side-effects.
15093 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15094 Alan Hayward <alan.hayward@arm.com>
15095 David Sherwood <david.sherwood@arm.com>
15097 * target.def (hard_regno_nregs): New hook.
15098 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15099 * targhooks.h (default_hard_regno_nregs): Declare.
15100 * targhooks.c (default_hard_regno_nregs): New function.
15101 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15102 (TARGET_HARD_REGNO_NREGS): ...this hook.
15103 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15104 (CLASS_MAX_NREGS): Likewise.
15105 * doc/tm.texi: Regenerate.
15106 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15107 instead of HARD_REGNO_NREGS.
15108 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15109 HARD_REGNO_NREGS in the comment.
15110 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15111 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15112 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15113 Return an unsigned int.
15114 (TARGET_HARD_REGNO_NREGS): Redefine.
15115 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15116 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15117 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15118 (arc_hard_regno_nregs): New function.
15119 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15120 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15121 (arm_hard_regno_nregs): New function.
15122 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15123 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15124 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15125 (TARGET_HARD_REGNO_NREGS): Redefine.
15126 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15127 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15128 (HARD_REGNO_NREGS): Delete.
15129 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15130 (cr16_hard_regno_nregs): New function.
15131 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15132 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15133 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15134 (cris_hard_regno_nregs): New function.
15135 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15136 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15137 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15138 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15139 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15140 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15141 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15142 (frv_hard_regno_nregs): Make static. Take and return an
15144 (frv_class_max_nregs): Remove outdated copy of documentation.
15145 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15146 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15147 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15148 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15149 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15150 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15151 (TARGET_HARD_REGNO_NREGS): Redefine.
15152 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15153 (CLASS_MAX_NREGS): Update comment.
15154 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15155 (ia64_hard_regno_nregs): New function.
15156 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15157 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15158 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15159 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15160 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15162 (m32c_hard_regno_nregs): Likewise. Make static.
15163 (TARGET_HARD_REGNO_NREGS): Redefine.
15164 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15165 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15166 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15167 (m68k_hard_regno_nregs): New function.
15168 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15169 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15170 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15171 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15172 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15173 Take and return an unsigned int.
15174 (TARGET_HARD_REGNO_NREGS): Redefine.
15175 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15176 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15177 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15178 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15179 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15180 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15181 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15182 (msp430_hard_regno_nregs): Make static. Take and return an
15184 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15185 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15186 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15187 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15188 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15189 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15190 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15191 (TARGET_HARD_REGNO_NREGS): Redefine.
15192 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15193 (PA_HARD_REGNO_NREGS): ...this.
15194 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15195 (PA_HARD_REGNO_NREGS): ...this.
15196 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15197 (pa_hard_regno_nregs): New function.
15198 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15199 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15200 (pdp11_hard_regno_nregs): New function.
15201 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15202 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15203 (rs6000_hard_regno_nregs_hook): New function.
15204 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15205 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15206 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15207 Take and return an unsigned int. Move earlier in file.
15208 (TARGET_HARD_REGNO_NREGS): Redefine.
15209 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15210 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15211 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15212 (rl78_hard_regno_nregs): Make static. Take and return an
15214 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15215 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15216 (rs6000_hard_regno_nregs_hook): New function.
15217 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15218 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15219 (TARGET_HARD_REGNO_NREGS): Redefine.
15220 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15221 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15222 instead of HARD_REGNO_NREGS.
15223 (s390_hard_regno_nregs): New function.
15224 (s390_hard_regno_mode_ok): Add comment from s390.h.
15225 (TARGET_HARD_REGNO_NREGS): Redefine.
15226 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15227 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15228 (sh_hard_regno_nregs): New function.
15229 (sh_pass_in_reg_p): Use it.
15230 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15231 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15232 (sparc_hard_regno_nregs): New function.
15233 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15234 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15235 (spu_function_arg_advance): Use it, supplying a valid register number.
15236 (TARGET_HARD_REGNO_NREGS): Redefine.
15237 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15238 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15239 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15240 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15241 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15242 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15243 (CLASS_MAX_NREGS): Remove copy of old documentation.
15244 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15245 (visium_hard_regno_nregs): New function.
15246 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15247 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15248 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15249 xtensa_hard_regno_nregs): New function.
15250 * system.h (HARD_REGNO_NREGS): Poison.
15252 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15254 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15255 hard_regno_nregs instead of HARD_REGNO_NREGS.
15256 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15257 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15258 (c6x_expand_epilogue): Likewise.
15259 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15260 (frv_read_iacc_argument): Likewise.
15261 * config/sh/sh.c: Include regs.h.
15262 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15263 (regs_used): Likewise.
15264 (output_stack_adjust): Likewise.
15265 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15266 * expmed.c: Include regs.h.
15267 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15268 * ree.c: Include regs.h.
15269 (combine_reaching_defs): Use hard_regno_nregs instead of
15271 (add_removable_extension): Likewise.
15273 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15275 * regs.h (hard_regno_nregs): Turn into a function.
15276 (end_hard_regno): Update accordingly.
15277 * caller-save.c (setup_save_areas): Likewise.
15278 (save_call_clobbered_regs): Likewise.
15279 (replace_reg_with_saved_mem): Likewise.
15280 (insert_restore): Likewise.
15281 (insert_save): Likewise.
15282 * combine.c (can_change_dest_mode): Likewise.
15283 (move_deaths): Likewise.
15284 (distribute_notes): Likewise.
15285 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15286 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15287 (rs6000_split_multireg_move): Likewise.
15288 (rs6000_register_move_cost): Likewise.
15289 (rs6000_memory_move_cost): Likewise.
15290 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15291 (rs6000_split_multireg_move): Likewise.
15292 (rs6000_register_move_cost): Likewise.
15293 (rs6000_memory_move_cost): Likewise.
15294 * cselib.c (cselib_reset_table): Likewise.
15295 (cselib_lookup_1): Likewise.
15296 * emit-rtl.c (set_mode_and_regno): Likewise.
15297 * function.c (aggregate_value_p): Likewise.
15298 * ira-color.c (setup_profitable_hard_regs): Likewise.
15299 (check_hard_reg_p): Likewise.
15300 (calculate_saved_nregs): Likewise.
15301 (assign_hard_reg): Likewise.
15302 (improve_allocation): Likewise.
15303 (calculate_spill_cost): Likewise.
15304 * ira-emit.c (modify_move_list): Likewise.
15305 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15306 (ira_hard_reg_in_set_p): Likewise.
15307 * ira.c (setup_reg_mode_hard_regset): Likewise.
15308 (clarify_prohibited_class_mode_regs): Likewise.
15309 (check_allocation): Likewise.
15310 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15311 (lra_setup_reg_renumber): Likewise.
15312 (setup_try_hard_regno_pseudos): Likewise.
15313 (spill_for): Likewise.
15314 (assign_hard_regno): Likewise.
15315 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15316 * lra-constraints.c (in_class_p): Likewise.
15317 (lra_constraint_offset): Likewise.
15318 (simplify_operand_subreg): Likewise.
15319 (lra_constraints): Likewise.
15320 (split_reg): Likewise.
15321 (split_if_necessary): Likewise.
15322 (invariant_p): Likewise.
15323 (inherit_in_ebb): Likewise.
15324 * lra-lives.c (process_bb_lives): Likewise.
15325 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15326 (get_hard_regs): Likewise.
15327 (do_remat): Likewise.
15328 * lra-spills.c (assign_spill_hard_regs): Likewise.
15329 * mode-switching.c (create_pre_exit): Likewise.
15330 * postreload.c (reload_combine_recognize_pattern): Likewise.
15331 * recog.c (peep2_find_free_register): Likewise.
15332 * regcprop.c (kill_value_regno): Likewise.
15333 (set_value_regno): Likewise.
15334 (copy_value): Likewise.
15335 (maybe_mode_change): Likewise.
15336 (find_oldest_value_reg): Likewise.
15337 (copyprop_hardreg_forward_1): Likewise.
15338 * regrename.c (check_new_reg_p): Likewise.
15339 (regrename_do_replace): Likewise.
15340 * reload.c (push_reload): Likewise.
15341 (combine_reloads): Likewise.
15342 (find_dummy_reload): Likewise.
15343 (operands_match_p): Likewise.
15344 (find_reloads): Likewise.
15345 (find_equiv_reg): Likewise.
15346 (reload_adjust_reg_for_mode): Likewise.
15347 * reload1.c (count_pseudo): Likewise.
15348 (count_spilled_pseudo): Likewise.
15349 (find_reg): Likewise.
15350 (clear_reload_reg_in_use): Likewise.
15351 (free_for_value_p): Likewise.
15352 (allocate_reload_reg): Likewise.
15353 (choose_reload_regs): Likewise.
15354 (reload_adjust_reg_for_temp): Likewise.
15355 (emit_reload_insns): Likewise.
15356 (delete_output_reload): Likewise.
15357 * rtlanal.c (subreg_get_info): Likewise.
15358 * sched-deps.c (sched_analyze_reg): Likewise.
15359 * sel-sched.c (init_regs_for_mode): Likewise.
15360 (mark_unavailable_hard_regs): Likewise.
15361 (choose_best_reg_1): Likewise.
15362 (verify_target_availability): Likewise.
15363 * valtrack.c (dead_debug_insert_temp): Likewise.
15364 * var-tracking.c (track_loc_p): Likewise.
15365 (emit_note_insn_var_location): Likewise.
15366 * varasm.c (make_decl_rtl): Likewise.
15367 * reginfo.c (choose_hard_reg_mode): Likewise.
15368 (init_reg_modes_target): Refer directly to
15369 this_target_regs->x_hard_regno_nregs.
15371 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15373 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15374 instead of hard_regno_nregs.
15376 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15378 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15379 end_hard_regno instead of hard_regno_nregs.
15380 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15381 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15382 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15383 * ira-color.c (improve_allocation): Likewise.
15384 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15385 * lra-lives.c (mark_regno_live): Likewise.
15386 (mark_regno_dead): Likewise.
15387 * lra-remat.c (operand_to_remat): Likewise.
15388 * lra.c (collect_non_operand_hard_regs): Likewise.
15389 * postreload.c (reload_combine_note_store): Likewise.
15390 (move2add_valid_value_p): Likewise.
15391 * reload.c (regno_clobbered_p): Likewise.
15393 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15395 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15397 * config/v850/v850.c (v850_reorg): Likewise.
15398 * reload.c (refers_to_regno_for_reload_p): Likewise.
15399 (find_equiv_reg): Likewise.
15400 * reload1.c (reload_reg_reaches_end_p): Likewise.
15402 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15404 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15406 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15407 * config/arm/arm.c (output_move_neon): Likewise.
15408 (arm_attr_length_move_neon): Likewise.
15409 (neon_split_vcombine): Likewise.
15410 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15411 (c6x_mark_reg_written): Likewise.
15412 (c6x_dwarf_register_span): Likewise.
15413 * config/i386/i386.c (ix86_save_reg): Likewise.
15414 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15415 (rws_access_reg): Likewise.
15416 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15417 * mode-switching.c (create_pre_exit): Likewise.
15418 * ree.c (combine_reaching_defs): Likewise.
15419 (add_removable_extension): Likewise.
15420 * regcprop.c (find_oldest_value_reg): Likewise.
15421 (copyprop_hardreg_forward_1): Likewise.
15422 * reload.c (reload_inner_reg_of_subreg): Likewise.
15423 (push_reload): Likewise.
15424 (combine_reloads): Likewise.
15425 (find_dummy_reload): Likewise.
15426 (reload_adjust_reg_for_mode): Likewise.
15427 * reload1.c (find_reload_regs): Likewise.
15428 (forget_old_reloads_1): Likewise.
15429 (reload_reg_free_for_value_p): Likewise.
15430 (reload_adjust_reg_for_temp): Likewise.
15431 (emit_reload_insns): Likewise.
15432 (delete_output_reload): Likewise.
15433 * sel-sched.c (choose_best_reg_1): Likewise.
15434 (choose_best_pseudo_reg): Likewise.
15436 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15437 Alan Hayward <alan.hayward@arm.com>
15438 David Sherwood <david.sherwood@arm.com>
15440 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15441 * target.def (slow_unaligned_access): New hook.
15442 * targhooks.h (default_slow_unaligned_access): Declare.
15443 * targhooks.c (default_slow_unaligned_access): New function.
15444 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15445 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15446 * doc/tm.texi: Regenerate.
15447 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15448 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15449 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15451 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15452 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15454 (rs6000_slow_unaligned_access): New function.
15455 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15456 (expand_block_compare): Likewise.
15457 (expand_strn_compare): Likewise.
15458 (rs6000_rtx_costs): Likewise.
15459 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15460 (riscv_slow_unaligned_access): Likewise.
15461 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15462 (riscv_slow_unaligned_access_p): ...this and make static.
15463 (riscv_option_override): Update accordingly.
15464 (riscv_slow_unaligned_access): New function.
15465 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15466 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15467 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15468 (rs6000_slow_unaligned_access): New function.
15469 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15470 (rs6000_rtx_costs): Likewise.
15471 * config/rs6000/rs6000-string.c (expand_block_compare)
15472 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15473 of SLOW_UNALIGNED_ACCESS.
15474 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15475 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15476 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15477 of SLOW_UNALIGNED_ACCESS.
15478 * expmed.c (simple_mem_bitfield_p): Likewise.
15479 * expr.c (alignment_for_piecewise_move): Likewise.
15480 (emit_group_load_1): Likewise.
15481 (emit_group_store): Likewise.
15482 (copy_blkmode_from_reg): Likewise.
15483 (emit_push_insn): Likewise.
15484 (expand_assignment): Likewise.
15485 (store_field): Likewise.
15486 (expand_expr_real_1): Likewise.
15487 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15488 * lra-constraints.c (simplify_operand_subreg): Likewise.
15489 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15490 * gimple-ssa-store-merging.c: Likewise in block comment at start
15492 * tree-ssa-strlen.c: Include target.h.
15493 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15494 of SLOW_UNALIGNED_ACCESS.
15495 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15497 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15499 PR rtl-optimization/82185
15500 * expmed.c (emit_store_flag_int): Only test tem if it has been
15503 2017-09-12 Richard Biener <rguenther@suse.de>
15505 PR middle-end/82149
15506 * match.pd ((FTYPE) N CMP CST): Fix typo.
15508 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15510 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15512 (mips_near_type_p): Add 'short_call' attribute as a synonym
15514 * doc/extend.texi (short_call): Document new function attribute.
15516 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15519 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15520 assertion check that in the condition.
15521 (get_atomic_generic_size): Likewise. Before testing if parameter
15522 has pointer type, if it has array type, call for C++
15523 default_conversion to perform array-to-pointer conversion.
15525 2017-09-12 Richard Biener <rguenther@suse.de>
15527 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15528 for operations we cannot scalarize.
15530 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15532 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15533 vectors heap vectors. Clean up comments.
15534 Make visited_bbs a reference.
15535 (profitable_jump_thread_path): Make GC
15536 vectors heap vectors. Clean up comments.
15538 (convert_and_register_jump_thread_path): Make GC vectors heap
15540 (check_subpath_and_update_thread_path): Same. Clean up comments.
15541 Make visited_bbs a reference.
15542 (handle_phi): Abstract common code to to
15543 register_jump_thread_path_if_profitable.
15544 Rename VAR_BB to DEF_BB.
15546 Make GC vectors heap vectors.
15547 Make visited_bbs a reference.
15548 (handle_assignment): Same.
15549 (register_jump_thread_path_if_profitable): New.
15550 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15552 Make GC vectors heap vectors. Clean up comments.
15553 Make visited_bbs a reference.
15554 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15555 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15558 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15561 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15562 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15564 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15566 Revert r251800 and r251799.
15568 2017-09-11 Martin Jambor <mjambor@suse.cz>
15571 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15572 arguments in advance.
15573 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15574 use it to find predecessor edges.
15575 (naive_outof_ssa): Collect vector of predecessors.
15577 2017-09-08 Jason Merrill <jason@redhat.com>
15579 PR c++/70029 - ICE with ref-qualifier and -flto
15580 * langhooks.h (struct lang_hooks_for_types): Add
15581 copy_lang_qualifiers.
15582 * attribs.c (build_type_attribute_qual_variant): Use it.
15583 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15585 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15586 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15588 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15591 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15592 (*mulsi3_sp64): New instruction.
15593 (mulsi3): New expander.
15595 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15597 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15599 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15601 * sancov.c: Include memmodel.h.
15603 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15606 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15609 2017-09-07 Carl Love <cel@us.ibm.com>
15611 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15612 the sldi instruction.
15614 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15616 * sancov.c: Include tm_p.h.
15618 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15621 * output.h (switch_to_other_text_partition): New declaration.
15622 * varasm.c (switch_to_other_text_partition): New function.
15623 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15624 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15625 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15626 to the other text partition before emitting LCL label and switch back
15627 after emitting the word after it.
15629 2017-09-07 Richard Biener <rguenther@suse.de>
15631 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15632 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15633 critical edges here, after loop init.
15634 (pass_data_pre): Remove PROP_no_crit_edges flags.
15635 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15636 for valueization of call args to avoid leaking VN_TOP.
15637 (visit_use): Assert we do not visit default defs.
15638 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15639 Use error_mark_node to more easily detect leaking VN_TOP.
15640 All default-defs are varying, not VN_TOP. Mark them visited.
15641 (run_scc_vn): Make code match comment.
15643 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15645 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15646 OPTION_MASK_FLOAT128_KEYWORD.
15647 (POWERPC_MASKS): Likewise.
15648 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15649 support for the -mfloat128-type option, and make -mfloat128
15650 default on PowerPC Linux systems. Define or undefine
15651 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15652 Define __float128 to be __ieee128 if IEEE 128-bit support is
15653 enabled, or undefine it.
15654 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15655 Delete defining __FLOAT128_TYPE__.
15656 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15657 -mfloat128-type option and make -mfloat128 default on PowerPC
15659 (TARGET_FLOAT128_TYPE): Likewise.
15660 (-mfloat128-type): Likewise.
15661 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15662 Delete the -mfloat128-type option and make -mfloat128 default on
15663 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15664 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15665 128-bit floating point is enabled. Change tests from using
15666 -mfloat128-type to -mfloat128.
15667 (rs6000_mangle_type): Use the correct mangling for the __float128
15668 type even if normal long double is restricted to 64-bits.
15669 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15671 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15672 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15673 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15675 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15676 documentation for -mfloat128.
15678 2017-09-06 Olivier Hainque <hainque@adacore.com>
15680 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15682 2017-09-06 Wish Wu <wishwu007@gmail.com>
15683 Jakub Jelinek <jakub@redhat.com>
15685 * asan.c (initialize_sanitizer_builtins): Add
15686 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15687 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15688 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15689 BT_FN_VOID_UINT64_PTR variables.
15690 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15691 (BT_FN_VOID_UINT16_UINT16): Likewise.
15692 (BT_FN_VOID_UINT32_UINT32): Likewise.
15693 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15694 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15695 (BT_FN_VOID_UINT64_PTR): Likewise.
15696 * common.opt (flag_sanitize_coverage): New variable.
15697 (fsanitize-coverage=trace-pc): Remove.
15698 (fsanitize-coverage=): Add.
15699 * flag-types.h (enum sanitize_coverage_code): New enum.
15700 * fold-const.c (fold_range_test): Disable non-short-circuit
15701 optimization if flag_sanitize_coverage.
15702 (fold_truth_andor): Likewise.
15703 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15704 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15705 (coverage_sanitizer_opts): New array.
15706 (get_closest_sanitizer_option): Add OPTS argument, handle also
15707 OPT_fsanitize_coverage_.
15708 (parse_sanitizer_options): Adjusted to also handle
15709 OPT_fsanitize_coverage_.
15710 (common_handle_option): Add OPT_fsanitize_coverage_.
15711 * sancov.c (instrument_comparison, instrument_switch): New function.
15712 (sancov_pass): Add trace-cmp support.
15713 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15714 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15715 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15716 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15717 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15718 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15719 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15720 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15721 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15722 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15724 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15726 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15727 error. Only quit immediately if parsing is complete.
15728 (BEGIN): Initialize fatal_err and parse_done.
15729 (begin fpu, end fpu): Check number of arguments.
15730 (begin arch, end arch): Likewise.
15731 (begin cpu, end cpu): Likewise.
15732 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15733 (optalias): Likewise.
15735 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15737 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15738 * config/arm/arm-isa.h: Delete. Move definitions to ...
15739 * arm-cpus.in: ... here. Use new feature and fgroup values.
15740 * config/arm/arm.c (arm_option_override): Use lower case for feature
15742 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15743 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15744 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15746 (print_isa_bits_for): New function.
15747 (gen_isa): New function.
15748 (gen_comm_data): Use print_isa_bits_for.
15749 (define feature): New keyword.
15750 (define fgroup): New keyword.
15751 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15752 (arm-isa.h): Add rule to generate file.
15753 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15754 case for feature bit names.
15756 2017-09-06 Richard Biener <rguenther@suse.de>
15758 * tree-ssa-pre.c (NECESSARY): Remove.
15759 (create_expression_by_pieces): Do not touch pass-local flags.
15760 (insert_into_preds_of_block): Likewise.
15761 (do_pre_regular_insertion): Likewise.
15762 (eliminate_insert): Likewise.
15763 (eliminate_dom_walker::before_dom_children): Likewise.
15764 (fini_eliminate): Do not look at inserted_exprs.
15765 (mark_operand_necessary): Remove.
15766 (remove_dead_inserted_code): Replace with simple work-list
15767 algorithm based on inserted_exprs and SSA uses.
15768 (pass_pre::execute): Re-order fini_eliminate and
15769 remove_dead_inserted_code.
15771 2017-09-06 Olivier Hainque <hainque@adacore.com>
15773 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15774 for VxWorks 7. Adjust surrounding comments.
15776 2017-09-06 Richard Biener <rguenther@suse.de>
15778 * gimple-ssa-strength-reduction.c
15779 (find_candidates_dom_walker::before_dom_children): Also allow
15782 2017-09-06 Richard Biener <rguenther@suse.de>
15784 PR tree-optimization/82108
15785 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15786 for gap in the non-permutation SLP case.
15788 2017-09-06 Martin Jambor <mjambor@suse.cz>
15790 PR tree-optimization/82078
15791 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15792 add_access_to_work_queue...
15793 (build_accesses_from_assign): ...here.
15794 (propagate_all_subaccesses): Make sure racc is the group
15795 representative, if there is one.
15797 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15799 PR middle-end/82095
15800 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15803 2017-09-06 Richard Biener <rguenther@suse.de>
15805 * gimple-ssa-strength-reduction.c
15806 (find_candidates_dom_walker::before_doom_children): Use a
15807 type and not a mode check.
15809 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15812 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15814 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15816 2017-09-05 Jeff Law <law@redhat.com>
15818 PR tree-optimization/64910
15819 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15820 cases where we have 3 or more operands.
15822 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15824 PR middle-end/81768
15825 * omp-low.c (lower_omp_for): Recompute tree invariant if
15826 gimple_omp_for_initial/final is ADDR_EXPR.
15828 PR middle-end/81768
15829 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15830 into gimple val before gimplification fo the COND_EXPR.
15832 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15834 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15835 REGION_COPY argument.
15836 (thread_through_all_blocks): Remove unused argument to
15837 duplicate_thread_path.
15839 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15840 Alan Hayward <alan.hayward@arm.com>
15841 David Sherwood <david.sherwood@arm.com>
15843 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15844 Take a scalar_mode rather than a machine_mode.
15845 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15846 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15847 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15848 (aarch64_gen_adjusted_ldpstp): Likewise.
15849 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15851 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15852 Alan Hayward <alan.hayward@arm.com>
15853 David Sherwood <david.sherwood@arm.com>
15855 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15856 Take a scalar_int_mode instead of a machine_mode.
15857 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15858 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15859 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15860 (aarch64_simd_attr_length_rglist): Delete.
15861 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15862 a scalar_int_mode instead of a machine_mode.
15863 (aarch64_add_offset): Likewise.
15864 (aarch64_internal_mov_immediate): Likewise
15865 (aarch64_add_constant_internal): Likewise.
15866 (aarch64_add_constant): Likewise.
15867 (aarch64_movw_imm): Likewise.
15868 (aarch64_rtx_arith_op_extract_p): Likewise.
15869 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15870 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15871 Remove assert that the mode isn't a vector.
15872 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15873 (aarch64_expand_mov_immediate): Update calls after above changes.
15874 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15875 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15876 (aarch64_move_imm): Likewise.
15877 (aarch64_can_const_movi_rtx_p): Likewise.
15878 (aarch64_strip_extend): Likewise.
15879 (aarch64_extr_rtx_p): Likewise.
15880 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15881 a CONST_INT when the mode parameter is VOIDmode.
15882 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15884 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15886 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15887 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15888 (bitwise_type_for_mode): Update accordingly.
15890 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15892 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15893 * stor-layout.c (mode_for_size_tree): Likewise.
15894 (mode_for_array): Update accordingly.
15895 (layout_decl): Likewise.
15896 (compute_record_mode): Likewise. Only set the mode once.
15898 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15900 * target.def (get_mask_mode): Change return type to opt_mode.
15902 * doc/tm.texi: Regenerate.
15903 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15904 * targhooks.c (default_get_mask_mode): Likewise.
15905 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15906 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15907 targetm.get_mask_mode.
15908 * tree.c (build_truth_vector_type): Likewise.
15910 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15912 * machmode.h (mode_for_vector): Return an opt_mode.
15913 * stor-layout.c (mode_for_vector): Likewise.
15914 (mode_for_int_vector): Update accordingly.
15915 (layout_type): Likewise.
15916 * config/i386/i386.c (emit_memmov): Likewise.
15917 (ix86_expand_set_or_movmem): Likewise.
15918 (ix86_expand_vector_init): Likewise.
15919 (ix86_get_mask_mode): Likewise.
15920 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15922 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15923 * expmed.c (extract_bit_field_1): Likewise.
15924 * expr.c (expand_expr_real_2): Likewise.
15925 * optabs-query.c (can_vec_perm_p): Likewise.
15926 (can_vec_mask_load_store_p): Likewise.
15927 * optabs.c (expand_vec_perm): Likewise.
15928 * targhooks.c (default_get_mask_mode): Likewise.
15929 * tree-vect-stmts.c (vectorizable_store): Likewise.
15930 (vectorizable_load): Likewise.
15931 (get_vectype_for_scalar_type_and_size): Likewise.
15933 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15935 * machmode.h (mode_for_int_vector): New function.
15936 * stor-layout.c (mode_for_int_vector): Likewise.
15937 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15938 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15939 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15940 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15941 (s390_expand_vcond): Likewise.
15943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15945 * machmode.h (opt_machine_mode): New type.
15946 (opt_mode<T>): Allow construction from anything that can be
15948 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15949 (mode_for_size): Return an opt_machine_mode.
15950 * stor-layout.c (mode_for_size): Likewise.
15951 (mode_for_size_tree): Update call accordingly.
15952 (bitwise_mode_for_mode): Likewise.
15953 (make_fract_type): Likewise.
15954 (make_accum_type): Likewise.
15955 * caller-save.c (replace_reg_with_saved_mem): Update call
15957 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15958 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15959 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15960 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15961 * expmed.c (extract_bit_field_1): Likewise.
15962 * reload.c (get_secondary_mem): Likewise.
15963 * varasm.c (assemble_integer): Likewise.
15964 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15967 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15969 * machmode.h (decimal_float_mode_for_size): New function.
15970 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15971 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15972 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15973 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15974 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15975 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15977 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15979 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15980 (get_builtin_sync_mode): Likewise.
15981 (expand_ifn_atomic_compare_exchange): Likewise.
15982 (expand_builtin_atomic_clear): Likewise.
15983 (expand_builtin_atomic_test_and_set): Likewise.
15984 (fold_builtin_atomic_always_lock_free): Likewise.
15985 * calls.c (compute_argument_addresses): Likewise.
15986 (emit_library_call_value_1): Likewise.
15987 (store_one_arg): Likewise.
15988 * combine.c (combine_instructions): Likewise.
15989 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15990 * config/arm/arm.c (arm_function_value): Likewise.
15991 (aapcs_allocate_return_reg): Likewise.
15992 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15993 * config/i386/i386.c (construct_container): Likewise.
15994 (ix86_gimplify_va_arg): Likewise.
15995 (ix86_expand_sse_cmp): Likewise.
15996 (emit_memmov): Likewise.
15997 (emit_memset): Likewise.
15998 (expand_small_movmem_or_setmem): Likewise.
15999 (ix86_expand_pextr): Likewise.
16000 (ix86_expand_pinsr): Likewise.
16001 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16002 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16004 * config/mips/mips.c (mips_function_value_1) Likewise.
16005 (mips_block_move_straight): Likewise.
16006 (mips_expand_ins_as_unaligned_store): Likewise.
16007 * config/powerpcspe/powerpcspe.c
16008 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16009 (rs6000_darwin64_record_arg_flush): Likewise.
16010 * config/rs6000/rs6000.c
16011 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16012 (rs6000_darwin64_record_arg_flush): Likewise.
16013 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16014 (sparc_function_value_1): Likewise.
16015 * config/spu/spu.c (adjust_operand): Likewise.
16016 (spu_emit_branch_or_set): Likewise.
16017 (arith_immediate_p): Likewise.
16018 * emit-rtl.c (gen_lowpart_common): Likewise.
16019 * expr.c (expand_expr_real_1): Likewise.
16020 * function.c (assign_parm_setup_block): Likewise.
16021 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16022 * reload1.c (alter_reg): Likewise.
16023 * stor-layout.c (mode_for_vector): Likewise.
16024 (layout_type): Likewise.
16026 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16028 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16029 (spu_convert_move): Likewise.
16030 * lower-subreg.c (resolve_simple_move): Likewise.
16032 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16035 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16036 define_insn to a define_expand.
16037 (altivec_vsum2sws_direct): New define_insn.
16038 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16040 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16042 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16045 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16050 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16051 realign stack if stack alignment needed is less than incoming
16054 2017-09-05 Marek Polacek <polacek@redhat.com>
16057 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16060 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16062 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16064 2017-09-05 Richard Biener <rguenther@suse.de>
16066 PR tree-optimization/82084
16067 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16069 2017-09-05 Richard Biener <rguenther@suse.de>
16071 PR tree-optimization/82102
16072 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16074 2017-09-05 Martin Liska <mliska@suse.cz>
16076 PR tree-optimization/82032
16077 * tree-cfg.c (generate_range_test): New function.
16078 * tree-cfg.h (generate_range_test): Declared here.
16079 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16080 (cleanup_control_expr_graph): Use it.
16081 * tree-switch-conversion.c (try_switch_expansion): Remove
16083 (emit_case_nodes): Use generate_range_test.
16085 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16088 * config/i386/i386.md (*<btsc><mode>_mask): Add
16089 TARGET_USE_BT to insn constraint.
16090 (*btr<mode>_mask): Ditto.
16092 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16094 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16095 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16097 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16100 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16101 TARGET_NEON and TARGET_IWMMXT.
16102 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16103 TARGET_NEON and TARGET_IWMMXT.
16104 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16106 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16108 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16109 (ix86_rewrite_tls_address): Ditto.
16110 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16111 (ix86_rewrite_tls_address_1): Ditto.
16112 (ix86_rewrite_tls_address): Ditto.
16113 * config/i386/predicates.md (tls_address_pattern): New predicate.
16114 * config/i386/i386.md (TLS address splitter): New splitter.
16116 2017-09-04 Richard Biener <rguenther@suse.de>
16118 PR tree-optimization/82084
16119 * fold-const.h (can_native_encode_string_p): Declare.
16120 * fold-const.c (can_native_encode_string_p): Factor out from ...
16121 (native_encode_string): ... here.
16122 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16123 vectorizing stores from constants we later cannot handle.
16125 2017-09-04 Marek Polacek <polacek@redhat.com>
16128 * doc/invoke.texi: Update -Wtautological-compare documentation.
16130 2017-09-04 Jeff Law <law@redhat.com>
16132 PR tree-optimization/64910
16133 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16134 swap the first and last operand if the last is a constant.
16136 2017-09-04 Marek Polacek <polacek@redhat.com>
16139 * convert.c (do_narrow): When sanitizing signed integer overflows,
16140 bail out for signed types.
16141 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16143 2017-09-04 Richard Biener <rguenther@suse.de>
16145 PR tree-optimization/82060
16146 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16147 Move devirtualization after stmt folding and before EH/AB/noreturn
16148 cleanup to get the stmt refs canonicalized. Use a bool instead
16149 of gimple_modified_p since that doesn't work for NOPs. Schedule
16150 NOPs generated by folding for removal.
16152 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16153 Alan Hayward <alan.hayward@arm.com>
16154 David Sherwood <david.sherwood@arm.com>
16156 * coretypes.h (pad_direction): New enum.
16157 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16158 (FUNCTION_ARG_PADDING): Likewise.
16159 * target.def (function_arg_padding): New hook.
16160 * targhooks.h (default_function_arg_padding): Declare.
16161 * targhooks.c (default_function_arg_padding): New function.
16162 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16163 (TARGET_FUNCTION_ARG_PADDING): ...this.
16164 * doc/tm.texi: Regenerate.
16165 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16166 instead of direction.
16167 (compute_argument_addresses): Likewise.
16168 (load_register_parameters): Likewise.
16169 (emit_library_call_value_1): Likewise.
16170 (store_one_arg): Use targetm.calls.function_arg_padding instead
16171 of FUNCTION_ARG_PADDING.
16172 (must_pass_in_stack_var_size_or_pad): Likewise.
16173 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16174 (emit_group_store): Likewise.
16175 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16176 instead of FUNCTION_ARG_PADDING.
16177 (emit_push_insn): Likewise, and propagate enum change throughout
16179 * function.h (direction): Delete.
16180 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16182 * function.c (assign_parm_find_stack_rtl): Likewise.
16183 (assign_parm_setup_block_p): Likewise.
16184 (assign_parm_setup_block): Likewise.
16185 (gimplify_parameters): Likewise.
16186 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16187 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16189 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16190 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16191 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16192 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16193 (aarch64_function_arg_padding): ...this new function.
16194 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16195 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16196 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16197 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16198 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16199 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16200 (arm_pad_arg_upward): Replace with...
16201 (arm_function_arg_padding): ...this new function.
16202 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16204 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16205 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16206 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16207 (ia64_hpux_function_arg_padding): Replace with...
16208 (ia64_function_arg_padding): ...this new function. Use pad_direction
16209 instead of direction. Check for TARGET_HPUX.
16210 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16211 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16212 (iq2000_function_arg_padding): New function.
16213 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16214 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16215 (mips_function_arg_padding): ...this new function.
16216 (mips_pad_reg_upward): Update accordingly.
16217 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16218 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16219 targetm.calls.function_arg_padding.
16220 (FUNCTION_ARG_PADDING): Delete.
16221 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16222 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16223 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16224 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16225 (nios2_block_reg_padding): Return pad_direction instead of direction.
16226 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16227 instead of direction.
16228 (nios2_function_arg_padding): Likewise. Make static.
16229 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16230 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16231 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16232 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16233 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16234 (pa_function_arg_padding): Make static. Return pad_direction instead
16236 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16237 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16238 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16239 instead of direction. Use targetm.calls.function_arg_padding.
16240 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16241 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16242 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16243 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16244 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16246 (function_arg_padding): Rename to...
16247 (rs6000_function_arg_padding): ...this. Make static. Return
16248 pad_direction instead of direction.
16249 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16250 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16251 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16252 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16253 instead of direction. Use targetm.calls.function_arg_padding.
16254 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16255 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16256 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16257 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16258 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16259 (function_arg_padding): Rename to...
16260 (rs6000_function_arg_padding): ...this. Make static. Return
16261 pad_direction instead of direction.
16262 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16263 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16264 * config/s390/s390.c (s390_function_arg_padding): New function.
16265 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16266 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16267 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16268 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16269 (function_arg_padding): Rename to...
16270 (sparc_function_arg_padding): ...this. Make static. Return
16271 pad_direction instead of direction.
16272 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16273 * config/spu/spu.c (spu_function_arg_padding): New function.
16274 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16275 * system.h (FUNCTION_ARG_PADDING): Poison.
16277 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16278 Alan Hayward <alan.hayward@arm.com>
16279 David Sherwood <david.sherwood@arm.com>
16281 * target.def (modes_tieable_p): New hook.
16282 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16283 (TARGET_MODES_TIEABLE_P): ...this.
16284 * doc/tm.texi.in: Regenerate.
16285 * hooks.h (hook_bool_mode_mode_true): Declare.
16286 * hooks.c (hook_bool_mode_mode_true): New function.
16287 * combine.c (subst): Use targetm.modes_tieable_p instead of
16289 * dse.c (find_shift_sequence): Likewise.
16290 * expmed.c (extract_low_bits): Likewise.
16291 * lower-subreg.c: Include target.h.
16292 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16294 * rtlanal.c (rtx_cost): Likewise.
16295 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16296 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16297 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16298 (TARGET_MODES_TIEABLE_P): Redefine.
16299 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16300 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16301 (TARGET_MODES_TIEABLE_P): Redefine.
16302 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16303 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16304 (arc_modes_tieable_p): New function.
16305 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16306 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16307 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16308 (arm_modes_tieable_p): Make static.
16309 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16310 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16311 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16312 (TARGET_MODES_TIEABLE_P): Redefine.
16313 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16314 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16315 (TARGET_MODES_TIEABLE_P): Redefine.
16316 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16317 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16318 (cr16_modes_tieable_p): New function.
16319 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16320 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16321 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16322 (TRULY_NOOP_TRUNCATION): Update comment.
16323 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16324 (TRULY_NOOP_TRUNCATION): Update comment.
16325 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16326 (frv_modes_tieable_p): New function.
16327 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16328 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16329 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16330 (TARGET_MODES_TIEABLE_P): Redefine.
16331 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16332 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16333 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16334 (TARGET_MODES_TIEABLE_P): Redefine.
16335 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16336 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16337 (ia64_modes_tieable_p): New function.
16338 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16339 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16340 (iq2000_modes_tieable_p): New function.
16341 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16342 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16343 (lm32_modes_tieable_p): New function.
16344 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16345 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16346 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16347 (TARGET_MODES_TIEABLE_P): Redefine.
16348 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16349 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16350 (m32r_modes_tieable_p): New function.
16351 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16352 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16353 (m68k_modes_tieable_p): New function.
16354 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16355 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16356 (mcore_modes_tieable_p): New function.
16357 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16358 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16360 (TARGET_MODES_TIEABLE_P): Redefine.
16361 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16362 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16363 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16364 (TARGET_MODES_TIEABLE_P): Redefine.
16365 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16366 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16367 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16368 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16369 (mn10300_modes_tieable_p): ...this and make static.
16370 (TARGET_MODES_TIEABLE_P): Redefine.
16371 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16372 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16373 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16374 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16375 (msp430_modes_tieable_p): Make static.
16376 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16377 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16378 (TARGET_MODES_TIEABLE_P): Redefine.
16379 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16380 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16381 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16382 (TARGET_MODES_TIEABLE_P): Redefine.
16383 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16384 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16385 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16386 (TARGET_MODES_TIEABLE_P): Redefine.
16387 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16388 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16389 (pdp11_modes_tieable_p): New function.
16390 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16391 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16392 (rs6000_modes_tieable_p): New function.
16393 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16394 * config/powerpcspe/powerpcspe.md: Update comment.
16395 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16396 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16397 (TARGET_MODES_TIEABLE_P): Redefine.
16398 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16399 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16400 (rl78_modes_tieable_p): New function.
16401 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16402 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16403 (rs6000_modes_tieable_p): New function.
16404 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16405 * config/rs6000/rs6000.md: Update comment.
16406 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16407 * config/rx/rx.c (rx_modes_tieable_p): New function.
16408 (TARGET_MODES_TIEABLE_P): Redefine.
16409 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16410 * config/s390/s390.c (s390_modes_tieable_p): New function.
16411 (TARGET_MODES_TIEABLE_P): Redefine.
16412 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16413 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16414 (sh_modes_tieable_p): New function.
16415 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16416 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16417 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16418 (sparc_modes_tieable_p): Make static.
16419 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16420 * config/spu/spu.c (spu_modes_tieable_p): New function.
16421 (TARGET_MODES_TIEABLE_P): Redefine.
16422 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16423 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16424 (TARGET_MODES_TIEABLE_P): Redefine.
16425 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16426 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16427 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16428 * config/v850/v850.c (v850_modes_tieable_p): New function.
16429 (TARGET_MODES_TIEABLE_P): Redefine.
16430 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16431 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16432 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16433 (visium_modes_tieable_p): New function.
16434 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16435 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16436 (xtensa_modes_tieable_p): New function.
16437 * system.h (MODES_TIEABLE_P): Poison.
16439 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16440 Alan Hayward <alan.hayward@arm.com>
16441 David Sherwood <david.sherwood@arm.com>
16443 * target.def (hard_regno_mode_ok): New hook.
16444 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16445 (TARGET_HARD_REGNO_MODE_OK): ...this.
16446 * doc/tm.texi.in: Regenerate.
16447 * hooks.h (hook_bool_uint_mode_true): Declare.
16448 * hooks.c (hook_bool_uint_mode_true): New function.
16449 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16450 HARD_REGNO_MODE_OK.
16451 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16452 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16453 instead of HARD_REGNO_MODE_OK.
16454 * caller-save.c: Include target.h.
16455 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16456 HARD_REGNO_MODE_OK.
16457 * combine.c (can_combine_p): Likewise.
16458 (combinable_i3pat): Likewise.
16459 (can_change_dest_mode): Likewise.
16460 * expr.c (init_expr_target): Likewise.
16461 (convert_move): Likewise.
16462 (convert_modes): Likewise.
16463 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16464 (setup_prohibited_mode_move_regs): Likewise.
16465 * ira.h (target_ira): Likewise.
16466 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16467 * lra-constraints.c (process_alt_operands): Likewise.
16468 (split_reg): Likewise.
16469 * recog.c (peep2_find_free_register): Likewise.
16470 * ree.c (combine_reaching_defs): Likewise.
16471 * regcprop.c (maybe_mode_change): Likewise.
16472 * reginfo.c (init_reg_sets_1): Likewise.
16473 (choose_hard_reg_mode): Likewise.
16474 (simplifiable_subregs): Likewise.
16475 * regrename.c (check_new_reg_p): Likewise.
16476 * reload.c (find_valid_class): Likewise.
16477 (find_valid_class_1): Likewise.
16478 (reload_inner_reg_of_subreg): Likewise.
16479 (push_reload): Likewise.
16480 (combine_reloads): Likewise.
16481 (find_dummy_reload): Likewise.
16482 (find_reloads): Likewise.
16483 * reload1.c (find_reg): Likewise.
16484 (set_reload_reg): Likewise.
16485 (allocate_reload_reg): Likewise.
16486 (choose_reload_regs): Likewise.
16487 (reload_adjust_reg_for_temp): Likewise.
16488 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16489 (simplify_subreg_regno): Likewise.
16490 * sel-sched.c (init_regs_for_mode): Likewise.
16491 * varasm.c (make_decl_rtl): Likewise.
16492 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16493 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16494 HARD_REGNO_MODE_OK.
16495 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16496 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16497 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16498 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16499 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16500 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16501 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16502 (arc_mode_class): Delete.
16503 (HARD_REGNO_MODE_OK): Delete.
16504 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16505 (arc_hard_regno_mode_ok): Rename old array to...
16506 (arc_hard_regno_mode_ok_modes): ...this.
16507 (arc_conditional_register_usage): Update accordingly.
16508 (arc_mode_class): Make static.
16509 (arc_hard_regno_mode_ok): New function.
16510 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16511 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16512 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16513 (arm_hard_regno_mode_ok): Make static.
16514 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16515 HARD_REGNO_MODE_OK.
16516 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16517 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16518 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16520 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16521 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16522 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16523 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16524 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16526 * config/bfin/predicates.md (valid_reg_operand): Use
16527 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16528 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16529 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16531 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16532 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16533 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16534 (cr16_hard_regno_mode_ok): Make static and return a bool.
16535 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16536 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16537 (cris_hard_regno_mode_ok): New function.
16538 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16539 (epiphany_mode_class): Delete.
16540 (HARD_REGNO_MODE_OK): Delete.
16541 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16542 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16543 (hard_regno_mode_ok): Rename to...
16544 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16545 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16546 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16547 HARD_REGNO_MODE_OK.
16548 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16549 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16550 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16551 (frv_hard_regno_mode_ok): Make static and return a bool.
16552 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16553 HARD_REGNO_MODE_OK.
16554 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16555 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16556 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16557 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16560 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16561 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16562 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16565 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16566 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16567 (ia64_hard_regno_mode_ok): New function.
16568 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16569 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16570 (iq2000_hard_regno_mode_ok): New function.
16571 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16572 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16573 (lm32_hard_regno_mode_ok): New function.
16574 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16575 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16576 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16577 instead of HARD_REGNO_MODE_OK.
16578 (m32c_hard_regno_ok): Rename to...
16579 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16580 (m32c_cannot_change_mode_class): Update accordingly.
16581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16582 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16583 (m32r_mode_class): Delete.
16584 (HARD_REGNO_MODE_OK): Delete.
16585 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16586 (m32r_hard_regno_mode_ok): Rename to...
16587 (m32r_hard_regno_modes): ...this.
16588 (m32r_mode_class): Make static.
16589 (m32r_hard_regno_mode_ok): New function.
16590 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16591 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16592 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16593 (m68k_hard_regno_mode_ok): Make static.
16594 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16595 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16596 (mcore_hard_regno_mode_ok): New function.
16597 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16598 (HARD_REGNO_MODE_OK): Delete.
16599 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16601 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16602 (microblaze_hard_regno_mode_ok): New function.
16603 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16604 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16605 (mips_hard_regno_mode_ok): Delete.
16606 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16607 (mips_hard_regno_mode_ok_p): ...this and make static.
16608 (mips_hard_regno_mode_ok_p): Rename to...
16609 (mips_hard_regno_mode_ok_uncached): ...this.
16610 (mips_hard_regno_mode_ok): New function.
16611 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16612 of HARD_REGNO_MODE_OK.
16613 (mips_option_override): Update after above name changes.
16614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16615 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16616 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16617 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16618 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16620 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16621 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16622 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16623 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16624 (msp430_hard_regno_mode_ok): Make static and return a bool.
16625 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16626 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16627 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16630 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16631 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16632 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16633 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16634 (PA_HARD_REGNO_MODE_OK): ...this
16635 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16636 (PA_HARD_REGNO_MODE_OK): ...this.
16637 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16638 (pa_hard_regno_mode_ok): New function.
16639 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16640 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16641 (pdp11_hard_regno_mode_ok): New function.
16642 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16643 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16645 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16647 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16648 (rs6000_hard_regno_mode_ok): Rename to...
16649 (rs6000_hard_regno_mode_ok_uncached): ...this.
16650 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16651 (rs6000_hard_regno_mode_ok): New function.
16652 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16653 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16654 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16655 (riscv_hard_regno_mode_ok): ...this and make static.
16656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16657 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16658 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16659 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16660 (rl78_hard_regno_mode_ok): Make static and return bool.
16661 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16662 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16664 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16665 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16666 (rs6000_hard_regno_mode_ok): Rename to...
16667 (rs6000_hard_regno_mode_ok_uncached): ...this.
16668 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16669 (rs6000_hard_regno_mode_ok): New function.
16670 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16671 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16672 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16673 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16674 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16675 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16677 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16678 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16679 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16680 (sh_hard_regno_mode_ok): Make static.
16681 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16682 instead of HARD_REGNO_MODE_OK.
16683 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16684 (sparc_mode_class): Delete.
16685 (HARD_REGNO_MODE_OK): Delete.
16686 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16687 (hard_regno_mode_classes): Make static.
16688 (sparc_mode_class): Likewise.
16689 (sparc_hard_regno_mode_ok): New function.
16690 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16691 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16692 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16695 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16696 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16697 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16698 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16700 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16701 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16702 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16703 (visium_hard_regno_mode_ok): New function.
16704 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16705 instead of HARD_REGNO_MODE_OK.
16706 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16707 (HARD_REGNO_MODE_OK): Delete.
16708 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16709 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16710 (xtensa_option_override): Update accordingly.
16711 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16712 (xtensa_hard_regno_mode_ok): New function.
16713 * system.h (HARD_REGNO_MODE_OK): Poison.
16715 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16716 Alan Hayward <alan.hayward@arm.com>
16717 David Sherwood <david.sherwood@arm.com>
16719 * target.def (hard_regno_call_part_clobbered): New hook.
16720 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16721 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16722 * doc/tm.texi: Regenerate.
16723 * hooks.h (hook_bool_uint_mode_false): Declare.
16724 * hooks.c (hook_bool_uint_mode_false): New function.
16725 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16726 * cselib.c (cselib_process_insn): Use
16727 targetm.hard_regno_call_part_clobbered instead of
16728 HARD_REGNO_CALL_PART_CLOBBERED.
16729 * ira-conflicts.c (ira_build_conflicts): Likewise.
16730 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16731 * lra-constraints.c (need_for_call_save_p): Likewise.
16732 * lra-lives.c: Include target.h.
16733 (check_pseudos_live_through_calls): Use
16734 targetm.hard_regno_call_part_clobbered instead of
16735 HARD_REGNO_CALL_PART_CLOBBERED.
16736 * regcprop.c: Include target.h.
16737 (copyprop_hardreg_forward_1): Use
16738 targetm.hard_regno_call_part_clobbered instead of
16739 HARD_REGNO_CALL_PART_CLOBBERED.
16740 * reginfo.c (choose_hard_reg_mode): Likewise.
16741 * regrename.c (check_new_reg_p): Likewise.
16742 * reload.c (find_equiv_reg): Likewise.
16743 * reload1.c (emit_reload_insns): Likewise.
16744 * sched-deps.c (deps_analyze_insn): Likewise.
16745 * sel-sched.c (init_regs_for_mode): Likewise.
16746 (mark_unavailable_hard_regs): Likewise.
16747 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16748 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16749 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16751 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16752 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16753 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16755 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16757 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16758 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16759 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16761 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16762 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16763 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16765 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16766 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16768 * config/powerpcspe/powerpcspe.c
16769 (rs6000_hard_regno_call_part_clobbered): New function.
16770 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16771 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16772 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16774 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16775 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16776 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16778 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16779 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16780 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16782 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16783 Alan Hayward <alan.hayward@arm.com>
16784 David Sherwood <david.sherwood@arm.com>
16786 * rtl.h (subreg_memory_offset): Declare.
16787 * emit-rtl.c (subreg_memory_offset): New function.
16788 * expmed.c (store_bit_field_1): Use it.
16789 * expr.c (undefined_operand_subword_p): Likewise.
16790 * simplify-rtx.c (simplify_subreg): Likewise.
16792 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16794 PR rtl-optimization/57448
16797 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16798 using atomic_load pattern.
16799 (expand_atomic_store): Likewise.
16801 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16804 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16805 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16806 handling. Use replace_call_with_value with NULL instead of
16807 gsi_replace, unlink_stmt_vdef and release_defs.
16809 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16812 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16814 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16817 * rtl.h (emit_library_call_value_1): Declare.
16818 (emit_library_call): Replace declaration with a series of overloads.
16819 Remove the parameter count argument.
16820 (emit_library_call_value): Likewise.
16821 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16822 with an "rtx_mode_t *".
16823 (emit_library_call_value): Delete.
16824 (emit_library_call): Likewise.
16825 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16826 (asan_emit_allocas_unpoison): Likewise.
16827 * builtins.c (expand_builtin_powi): Likewise.
16828 (expand_asan_emit_allocas_unpoison): Likewise.
16829 * cfgexpand.c (expand_main_function): Likewise.
16830 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16831 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16832 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16833 * config/arm/arm.c (arm_trampoline_init): Likewise.
16834 (arm_call_tls_get_addr): Likewise.
16835 (arm_expand_divmod_libfunc): Likewise.
16836 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16837 (smulsi3_highpart): Likewise.
16838 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16839 (c6x_expand_compare): Likewise.
16840 (c6x_expand_movmem): Likewise.
16841 * config/frv/frv.c (frv_trampoline_init): Likewise.
16842 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16843 (ix86_expand_divmod_libfunc): Likewise.
16844 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16845 (ia64_expand_compare): Likewise.
16846 (ia64_profile_hook): Likewise.
16847 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16848 (nonlocal_goto): Likewise.
16849 (restore_stack_nonlocal): Likewise.
16850 * config/m32r/m32r.c (block_move_call): Likewise.
16851 (m32r_trampoline_init): Likewise.
16852 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16853 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16854 (m68k_call_m68k_read_tp): Likewise.
16855 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16856 (microblaze_expand_divide): Likewise.
16857 * config/mips/mips.h (mips_args): Likewise.
16858 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16859 (MIPS_ICACHE_SYNC): Likewise.
16860 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16861 (nios2_trampoline_init): Likewise.
16862 * config/pa/pa.c (hppa_tls_call): Likewise.
16863 (pa_trampoline_init): Likewise.
16864 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16865 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16866 (expand_strn_compare): Likewise.
16867 (rs6000_generate_compare): Likewise.
16868 (rs6000_expand_float128_convert): Likewise.
16869 (output_profile_hook): Likewise.
16870 (rs6000_trampoline_init): Likewise.
16871 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16872 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16873 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16874 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16875 (rs6000_generate_compare): Likewise.
16876 (rs6000_expand_float128_convert): Likewise.
16877 (output_profile_hook): Likewise.
16878 (rs6000_trampoline_init): Likewise.
16879 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16880 * config/sh/sh.c (sh_trampoline_init): Likewise.
16881 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16882 (sparc_emit_float_lib_cmp): Likewise.
16883 (sparc32_initialize_trampoline): Likewise.
16884 (sparc64_initialize_trampoline): Likewise.
16885 (sparc_profile_hook): Likewise.
16886 * config/spu/spu.c (ea_load_store): Likewise.
16887 * config/spu/spu.md (floatunssidf2): Likewise.
16888 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16889 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16890 * config/visium/visium.c (expand_block_move_4): Likewise.
16891 (expand_block_move_2): Likewise.
16892 (expand_block_move_1): Likewise.
16893 (expand_block_set_4): Likewise.
16894 (expand_block_set_2): Likewise.
16895 (expand_block_set_1): Likewise.
16896 (visium_trampoline_init): Likewise.
16897 (visium_profile_hook): Likewise.
16898 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16899 (xtensa_setup_frame_addresses): Likewise.
16900 (xtensa_trampoline_init): Likewise.
16901 * except.c (sjlj_emit_function_enter): Likewise.
16902 (sjlj_emit_function_exit): Likewise.
16903 * explow.c (allocate_dynamic_stack_space): Likewise.
16904 (probe_stack_range): Likewise.
16905 * expr.c (convert_mode_scalar): Likewise.
16906 * optabs.c (expand_binop): Likewise.
16907 (expand_twoval_binop_libfunc): Likewise.
16908 (expand_unop): Likewise.
16909 (prepare_cmp_insn): Likewise.
16910 (prepare_float_lib_cmp): Likewise.
16911 (expand_float): Likewise.
16912 (expand_fix): Likewise.
16913 (expand_fixed_convert): Likewise.
16914 (maybe_emit_sync_lock_test_and_set): Likewise.
16915 (expand_atomic_compare_and_swap): Likewise.
16916 (expand_mem_thread_fence): Likewise.
16917 (expand_atomic_fetch_op): Likewise.
16919 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16921 * doc/generic.texi (OpenACC): Adjust URL.
16922 * doc/invoke.texi (C Dialect Options): Ditto.
16924 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16926 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16927 predicate for operand 1. Add (m,<S>) constraint.
16928 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16929 Prevent memory operand 1 with register operand 2.
16931 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16933 PR rtl-optimization/82024
16934 * combine.c (try_combine): If the combination result is a PARALLEL,
16935 and we only need to retain the SET in there that would be placed
16936 at I2, check that we can place that at I3 instead, before doing so.
16938 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16941 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16943 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16944 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16947 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16948 Jeff Law <law@redhat.com>
16950 * varasm.c (bss_initializer_p): Do not put constants into .bss
16951 (categorize_decl_for_section): Handle bss_initializer_p returning
16952 false when DECL_INITIAL is NULL.
16954 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16957 * config/s390/s390.c (s390_can_inline_p): New function.
16959 2017-09-01 Jeff Law <law@redhat.com>
16961 PR tree-optimization/82052
16962 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16963 Always initialize the returned slot after a hash table miss
16964 when INSERT is true.
16966 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16968 * config/s390/s390.md (mem_signal_fence): Remove.
16969 * doc/md.texi (mem_signal_fence): Remove.
16970 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16972 * target-insns.def (mem_signal_fence): Remove.
16974 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16977 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16980 * asan.c (create_odr_indicator): Strip name encoding from assembler
16981 name before appending it after __odr_asan_.
16983 2017-09-01 Martin Liska <mliska@suse.cz>
16985 PR tree-optimization/82059
16986 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16987 frequency only when an edge is redirected.
16989 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16991 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16992 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16993 (arc_conditional_register_usage): Remove ARC600 lp_count
16995 (arc_file_start): Emit Tag_ARC_CPU_variation.
16996 (arc_can_use_doloop_p): New conditions to use ZOLs.
16997 (hwloop_fail): New function.
16998 (hwloop_optimize): Likewise.
16999 (hwloop_pattern_reg): Likewise.
17000 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17001 (arc_reorg_loops): New function, calls reorg_loops.
17002 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17003 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17005 (arc_loop_hazard): Remove function, functionality moved into
17007 (arc_hazard): Remove arc_loop_hazard call.
17008 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17009 into hwloop_optimize.
17010 (arc_label_align): Remove ZOL handling.
17011 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17012 * config/arc/arc.md (doloop_begin): Remove pattern.
17013 (doloop_begin_i): Likewise.
17014 (doloop_end_i): Likewise.
17015 (doloop_fallback): Likewise.
17016 (doloop_fallback_m): Likewise.
17017 (doloop_end): Reimplement expand.
17018 (arc_lp): New pattern for LP instruction.
17019 (loop_end): New pattern.
17020 (loop_fail): Likewise.
17021 (decrement_and_branch_until_zero): Likewise.
17022 * config/arc/arc.opt (mlpc-width): New option.
17023 * doc/invoke.texi (mlpc-width): Document option.
17025 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17027 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17028 (arc_ccfsm_advance): Fix checking for delay slots.
17029 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17031 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17033 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17035 (movhi_insn): Update store instruction constraint which are saving
17036 6-bit short immediates.
17037 (movsi_insn): Consider also short scaled load operations.
17038 (zero_extendhisi2_i): Use Usd constraint instead of T.
17039 (extendhisi2_i): Add q constraint.
17040 (arc_clzsi2): Add type and length attributes.
17041 (arc_ctzsi2): Likewise.
17042 * config/arc/constraints.md (Usc): Update constraint, the
17043 assembler can parse two relocations for a single instruction.
17045 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17047 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17048 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17050 2017-08-31 Olivier Hainque <hainque@adacore.com>
17052 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17053 match as powerpc-wrs-vxworks*.
17055 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17057 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17058 register constraint for by-element operand.
17059 (aarch64_mls_elt_merge<mode>): Likewise.
17061 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17063 * config/arc/arc.c (arc_can_follow_jump): Check for short
17066 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17068 * config.gcc: Use g.opt for arc.
17069 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17070 functionality moved to ...
17071 (legitimate_scaled_address_p): New function, ...here.
17072 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17073 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17074 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17076 (arc_override_options): Handle G option.
17077 (arc_output_pic_addr_const): Correct function definition.
17078 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17079 (arc_decl_anon_ns_mem_p): Delete.
17080 (arc_in_small_data_p): Overhaul this function to take into
17081 consideration the value given via G option.
17082 (arc_rewrite_small_data_1): Renamed and corrected old
17083 arc_rewrite_small_data function.
17084 (arc_rewrite_small_data): New function.
17085 (small_data_pattern): Don't use pic_offset_table_rtx.
17086 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17087 * config/arc/simdext.md (movmisalignv2hi): Use
17088 prepare_move_operands function.
17090 (movmisalign*): Likewise.
17091 * doc/invoke.texi (ARC options): Document -G option.
17093 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17095 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17097 * config/arc/arc.c (arc_print_operand): Output scalled address for
17098 sdata whenever is possible.
17099 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17100 load/stores are available.
17101 (compact_sda_memory_operand): Check for the alignment required by
17102 code density instructions.
17103 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17105 * config/arc/constraints.md (Usd): Update constraint.
17106 (Us0): New constraint.
17107 (Usc): Update constraint.
17109 2017-08-31 Richard Biener <rguenther@suse.de>
17111 PR middle-end/82054
17112 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17113 function only once.
17115 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17117 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17118 Resize type_signature.
17120 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17121 Alan Hayward <alan.hayward@arm.com>
17122 David Sherwood <david.sherwood@arm.com>
17124 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17125 subregs whose inner modes can be stored in GPRs.
17126 (aarch64_classify_index): Likewise.
17128 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17129 Alan Hayward <alan.hayward@arm.com>
17130 David Sherwood <david.sherwood@arm.com>
17132 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17133 (V_INT_EQUIV): ...this.
17134 (v_cmp_result): Rename to...
17135 (v_int_equiv): ...this.
17136 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17137 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17138 (copysign<mode>3): Likewise.
17139 (aarch64_simd_bsl<mode>_internal): Likewise.
17140 (aarch64_simd_bsl<mode>): Likewise.
17141 (vec_cmp<mode><mode>): Likewise.
17142 (vcond<mode><mode>): Likewise.
17143 (vcond<v_cmp_mixed><mode>): Likewise.
17144 (vcondu<mode><v_cmp_mixed>): Likewise.
17145 (aarch64_cm<optab><mode>): Likewise.
17146 (aarch64_cmtst<mode>): Likewise.
17147 (aarch64_fac<optab><mode>): Likewise.
17148 (vec_perm_const<mode>): Likewise.
17149 (vcond_mask_<mode><v_cmp_result>): Rename to...
17150 (vcond_mask_<mode><v_int_equiv>): ...this.
17151 (vec_cmp<mode><v_cmp_result>): Rename to...
17152 (vec_cmp<mode><v_int_equiv>): ...this.
17154 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17155 Alan Hayward <alan.hayward@arm.com>
17156 David Sherwood <david.sherwood@arm.com>
17158 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17160 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17161 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17162 (UNSPEC_LD4_DREG): New unspecs.
17163 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17164 (aarch64_ld2<mode>_dreg_be): Replace with...
17165 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17167 (aarch64_ld3<mode>_dreg_le)
17168 (aarch64_ld3<mode>_dreg_be): Replace with...
17169 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17171 (aarch64_ld4<mode>_dreg_le)
17172 (aarch64_ld4<mode>_dreg_be): Replace with...
17173 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17176 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17178 PR tree-optimization/81987
17179 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17180 insert an initializer in a location not dominated by the stride
17183 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17185 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17186 on the entire header of the finally block in the fallthru case.
17188 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17190 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17192 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17194 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17195 rs6000_emit_move_from_cr and call renamed function.
17196 (rs6000_emit_prologue): Call renamed functions.
17197 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17198 movesi_from_cr, remove volatile CRs.
17200 2017-08-30 Jon Beniston <jon@beniston.com>
17201 Richard Biener <rguenther@suse.de>
17203 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17204 of VECTOR_MODE_P check.
17205 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17206 element vector types.
17208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17210 * df.h (df_read_modify_subreg_p): Remove in favor of...
17211 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17212 const_rtx instead of an rtx.
17213 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17214 * df-problems.c (df_word_lr_mark_ref): Likewise.
17215 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17216 (mark_pseudo_reg_dead): Likewise.
17217 (mark_ref_dead): Likewise.
17218 * reginfo.c (init_subregs_of_mode): Likewise.
17219 * sched-deps.c (sched_analyze_1): Likewise.
17220 * df-scan.c (df_def_record_1): Likewise.
17221 (df_uses_record): Likewise.
17222 (df_read_modify_subreg_p): Remove in favor of...
17223 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17224 const_rtx instead of an rtx.
17226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17227 Alan Hayward <alan.hayward@arm.com>
17228 David Sherwood <david.sherwood@arm.com>
17230 * rtl.h (partial_subreg_p): New function.
17231 * caller-save.c (save_call_clobbered_regs): Use it.
17232 * calls.c (expand_call): Likewise.
17233 * combine.c (combinable_i3pat): Likewise.
17234 (simplify_set): Likewise.
17235 (make_extraction): Likewise.
17236 (make_compound_operation_int): Likewise.
17237 (gen_lowpart_or_truncate): Likewise.
17238 (force_to_mode): Likewise.
17239 (make_field_assignment): Likewise.
17240 (reg_truncated_to_mode): Likewise.
17241 (record_truncated_value): Likewise.
17242 (move_deaths): Likewise.
17243 * cse.c (record_jump_cond): Likewise.
17244 (cse_insn): Likewise.
17245 * cselib.c (cselib_lookup_1): Likewise.
17246 * expmed.c (extract_bit_field_using_extv): Likewise.
17247 * function.c (assign_parm_setup_reg): Likewise.
17248 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17249 * ira-build.c (create_insn_allocnos): Likewise.
17250 * lra-coalesce.c (merge_pseudos): Likewise.
17251 * lra-constraints.c (match_reload): Likewise.
17252 (simplify_operand_subreg): Likewise.
17253 (curr_insn_transform): Likewise.
17254 * lra-lives.c (process_bb_lives): Likewise.
17255 * lra.c (new_insn_reg): Likewise.
17256 (lra_substitute_pseudo): Likewise.
17257 * regcprop.c (mode_change_ok): Likewise.
17258 (maybe_mode_change): Likewise.
17259 (copyprop_hardreg_forward_1): Likewise.
17260 * reload.c (push_reload): Likewise.
17261 (find_reloads): Likewise.
17262 (find_reloads_subreg_address): Likewise.
17263 * reload1.c (alter_reg): Likewise.
17264 (eliminate_regs_1): Likewise.
17265 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17267 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17269 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17270 back to if statements, including unpack.
17272 2017-08-30 Martin Liska <mliska@suse.cz>
17274 PR inline-asm/82001
17275 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17277 (func_checker::compare_asm_inputs_outputs): ... this function.
17278 (func_checker::compare_gimple_asm): Use the function to compare
17279 also ASM constrains.
17280 * ipa-icf-gimple.h: Rename the function.
17282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17283 Alan Hayward <alan.hayward@arm.com>
17284 David Sherwood <david.sherwood@arm.com>
17286 * coretypes.h (complex_mode): New type.
17287 * gdbhooks.py (build_pretty_printer): Handle it.
17288 * machmode.h (complex_mode): New class.
17289 (complex_mode::includes_p): New function.
17290 (is_complex_int_mode): Likewise.
17291 (is_complex_float_mode): Likewise.
17292 * genmodes.c (get_mode_class): Handle complex mode classes.
17293 * function.c (expand_function_end): Use is_complex_int_mode.
17295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17296 Alan Hayward <alan.hayward@arm.com>
17297 David Sherwood <david.sherwood@arm.com>
17299 * coretypes.h (scalar_mode_pod): New typedef.
17300 * gdbhooks.py (build_pretty_printer): Handle it.
17301 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17302 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17303 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17304 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17305 as_a <scalar_mode>.
17307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17308 Alan Hayward <alan.hayward@arm.com>
17309 David Sherwood <david.sherwood@arm.com>
17311 * machmode.h (mode_for_vector): Take a scalar_mode instead
17313 * stor-layout.c (mode_for_vector): Likewise.
17314 * explow.c (promote_mode): Use as_a <scalar_mode>.
17315 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17318 Alan Hayward <alan.hayward@arm.com>
17319 David Sherwood <david.sherwood@arm.com>
17321 * target.def (preferred_simd_mode): Take a scalar_mode
17322 instead of a machine_mode.
17323 * targhooks.h (default_preferred_simd_mode): Likewise.
17324 * targhooks.c (default_preferred_simd_mode): Likewise.
17325 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17326 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17327 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17328 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17329 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17330 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17331 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17332 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17334 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17335 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17336 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17337 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17338 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17339 * doc/tm.texi: Regenerate.
17340 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17344 Alan Hayward <alan.hayward@arm.com>
17345 David Sherwood <david.sherwood@arm.com>
17347 * target.def (scalar_mode_supported_p): Take a scalar_mode
17348 instead of a machine_mode.
17349 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17350 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17351 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17352 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17353 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17354 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17355 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17356 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17357 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17358 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17359 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17360 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17361 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17362 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17364 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17365 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17366 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17367 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17368 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17370 * doc/tm.texi: Regenerate.
17372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17373 Alan Hayward <alan.hayward@arm.com>
17374 David Sherwood <david.sherwood@arm.com>
17376 * coretypes.h (opt_scalar_mode): New typedef.
17377 * gdbhooks.py (build_pretty_printers): Handle it.
17378 * machmode.h (mode_iterator::get_2xwider): Add overload for
17380 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17382 * expr.c (convert_mode_scalar): Likewise.
17383 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17384 * optabs.c (expand_float): Likewise.
17385 (expand_fix): Likewise.
17386 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17389 Alan Hayward <alan.hayward@arm.com>
17390 David Sherwood <david.sherwood@arm.com>
17392 * optabs.c (expand_float): Explicitly check for scalars before
17393 using a branching expansion.
17394 (expand_fix): Likewise.
17396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17397 Alan Hayward <alan.hayward@arm.com>
17398 David Sherwood <david.sherwood@arm.com>
17400 * expr.c (convert_mode): Split scalar handling out into...
17401 (convert_mode_scalar): ...this new function. Treat the modes
17404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17405 Alan Hayward <alan.hayward@arm.com>
17406 David Sherwood <david.sherwood@arm.com>
17408 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17410 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17413 Alan Hayward <alan.hayward@arm.com>
17414 David Sherwood <david.sherwood@arm.com>
17416 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17417 rather than a machine_mode.
17418 (fixed_from_string): Likewise.
17419 (fixed_convert): Likewise.
17420 (fixed_convert_from_int): Likewise.
17421 (fixed_convert_from_real): Likewise.
17422 (real_convert_from_fixed): Likewise.
17423 * fixed-value.c (fixed_from_double_int): Likewise.
17424 (fixed_from_string): Likewise.
17425 (fixed_convert): Likewise.
17426 (fixed_convert_from_int): Likewise.
17427 (fixed_convert_from_real): Likewise.
17428 (real_convert_from_fixed): Likewise.
17429 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17432 Alan Hayward <alan.hayward@arm.com>
17433 David Sherwood <david.sherwood@arm.com>
17435 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17436 of separate mode class checks. Do not allow vector modes here.
17437 (immed_wide_int_const): Use as_a <scalar_mode>.
17438 * explow.c (trunc_int_for_mode): Likewise.
17439 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17440 (wi::shwi): Likewise.
17441 (wi::min_value): Likewise.
17442 (wi::max_value): Likewise.
17443 * dwarf2out.c (loc_descriptor): Likewise.
17444 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17445 for CONST_WIDE_INT.
17447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17448 Alan Hayward <alan.hayward@arm.com>
17449 David Sherwood <david.sherwood@arm.com>
17451 * tree.h (SCALAR_TYPE_MODE): New macro.
17452 * expr.c (expand_expr_addr_expr_1): Use it.
17453 (expand_expr_real_2): Likewise.
17454 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17455 (fold_convert_const_fixed_from_int): Likewise.
17456 (fold_convert_const_fixed_from_real): Likewise.
17457 (native_encode_fixed): Likewise
17458 (native_encode_complex): Likewise
17459 (native_encode_vector): Likewise.
17460 (native_interpret_fixed): Likewise.
17461 (native_interpret_real): Likewise.
17462 (native_interpret_complex): Likewise.
17463 (native_interpret_vector): Likewise.
17464 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17465 (simd_clone_adjust_argument_types): Likewise.
17466 (simd_clone_init_simd_arrays): Likewise.
17467 (simd_clone_adjust): Likewise.
17468 * stor-layout.c (layout_type): Likewise.
17469 * tree.c (build_minus_one_cst): Likewise.
17470 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17471 * tree-inline.c (estimate_move_cost): Likewise.
17472 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17473 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17474 (vectorizable_reduction): Likewise.
17475 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17476 (vect_recog_mixed_size_cond_pattern): Likewise.
17477 (check_bool_pattern): Likewise.
17478 (adjust_bool_pattern): Likewise.
17479 (search_type_for_mask_1): Likewise.
17480 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17481 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17482 (vectorizable_load): Likewise.
17483 (vectorizable_store): Likewise.
17484 * ubsan.c (ubsan_encode_value): Likewise.
17485 * varasm.c (output_constant): Likewise.
17487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17488 Alan Hayward <alan.hayward@arm.com>
17489 David Sherwood <david.sherwood@arm.com>
17491 * coretypes.h (scalar_mode): New class.
17492 * machmode.h (scalar_mode): Likewise.
17493 (scalar_mode::includes_p): New function.
17494 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17495 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17496 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17497 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17498 * expmed.c (store_bit_field_1): Likewise.
17499 (extract_bit_field_1): Likewise.
17500 * expr.c (write_complex_part): Likewise.
17501 (read_complex_part): Likewise.
17502 (emit_move_complex_push): Likewise.
17503 (expand_expr_real_2): Likewise.
17504 * function.c (assign_parm_setup_reg): Likewise.
17505 (assign_parms_unsplit_complex): Likewise.
17506 * optabs.c (expand_binop): Likewise.
17507 * rtlanal.c (subreg_get_info): Likewise.
17508 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17509 * varasm.c (output_constant_pool_2): Likewise.
17511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17512 Alan Hayward <alan.hayward@arm.com>
17513 David Sherwood <david.sherwood@arm.com>
17515 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17517 (expmed_mult_highpart_optab): Likewise.
17518 (expmed_mult_highpart): Likewise.
17520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17521 Alan Hayward <alan.hayward@arm.com>
17522 David Sherwood <david.sherwood@arm.com>
17524 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17525 instead of a machine_mode.
17526 (builtin_memset_read_str): Likewise.
17527 * builtins.c (c_readstr): Likewise.
17528 (builtin_memcpy_read_str): Likewise.
17529 (builtin_strncpy_read_str): Likewise.
17530 (builtin_memset_read_str): Likewise.
17531 (builtin_memset_gen_str): Likewise.
17532 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17533 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17534 instead of a machine_mode.
17535 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17537 (make_extraction): Likewise.
17538 (try_widen_shift_mode): Take and return scalar_int_modes instead
17540 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17541 a scalar_int_mode instead of a machine_mode.
17542 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17543 (avr_addr_space_pointer_mode): Likewise.
17544 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17545 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17546 (msp430_unwind_word_mode): Likewise.
17547 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17548 (spu_addr_space_pointer_mode): Likewise.
17549 (spu_addr_space_address_mode): Likewise.
17550 (spu_libgcc_cmp_return_mode): Likewise.
17551 (spu_libgcc_shift_count_mode): Likewise.
17552 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17553 (rl78_addr_space_pointer_mode): Likewise.
17554 (fl78_unwind_word_mode): Likewise.
17555 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17557 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17558 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17559 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17560 (mips_valid_pointer_mode): Likewise.
17561 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17562 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17563 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17565 (ft32_addr_space_address_mode): Likewise.
17566 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17567 scalar_int_mode instead of a machine_mode.
17568 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17570 (m32c_addr_space_address_mode): Likewise.
17571 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17572 (rs6000_eh_return_filter_mode): Likewise.
17573 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17574 (rs6000_eh_return_filter_mode): Likewise.
17575 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17576 (s390_libgcc_shift_count_mode): Likewise.
17577 (s390_unwind_word_mode): Likewise.
17578 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17580 * target.def (mode_rep_extended): Likewise.
17581 (valid_pointer_mode): Likewise.
17582 (addr_space.valid_pointer_mode): Likewise.
17583 (eh_return_filter_mode): Return a scalar_int_mode rather than
17585 (libgcc_cmp_return_mode): Likewise.
17586 (libgcc_shift_count_mode): Likewise.
17587 (unwind_word_mode): Likewise.
17588 (addr_space.pointer_mode): Likewise.
17589 (addr_space.address_mode): Likewise.
17590 * doc/tm.texi: Regenerate.
17591 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17593 (do_jump): Use scalar_int_mode for local variables.
17594 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17595 rather than a machine_mode.
17596 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17597 (scompare_loc_descriptor_wide): Likewise.
17598 (scompare_loc_descriptor_narrow): Likewise.
17599 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17601 * except.c (sjlj_emit_dispatch_table): Likewise.
17602 (expand_builtin_eh_copy_values): Likewise.
17603 * explow.c (convert_memory_address_addr_space_1): Likewise.
17604 Take a scalar_int_mode rather than a machine_mode.
17605 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17606 than a machine_mode.
17607 (memory_address_addr_space): Use scalar_int_mode for local variables.
17608 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17609 rather than a machine_mode.
17610 * expmed.c (mask_rtx): Likewise.
17611 (init_expmed_one_conv): Likewise.
17612 (expand_mult_highpart_adjust): Likewise.
17613 (extract_high_half): Likewise.
17614 (expmed_mult_highpart_optab): Likewise.
17615 (expmed_mult_highpart): Likewise.
17616 (expand_smod_pow2): Likewise.
17617 (expand_sdiv_pow2): Likewise.
17618 (emit_store_flag_int): Likewise.
17619 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17621 (extract_low_bits): Likewise.
17622 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17624 * expr.c (pieces_addr::adjust): Likewise.
17625 (can_store_by_pieces): Likewise.
17626 (store_by_pieces): Likewise.
17627 (clear_by_pieces_1): Likewise.
17628 (expand_expr_addr_expr_1): Likewise.
17629 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17630 (expand_expr_real_1): Likewise.
17631 (try_casesi): Likewise.
17632 * final.c (shorten_branches): Likewise.
17633 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17634 type of "mode" to machine_mode.
17635 * internal-fn.c (expand_arith_overflow_result_store): Take a
17636 scalar_int_mode rather than a machine_mode.
17637 (expand_mul_overflow): Use scalar_int_mode for local variables.
17638 * loop-doloop.c (doloop_modify): Likewise.
17639 (doloop_optimize): Likewise.
17640 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17641 than a machine_mode.
17642 (expand_doubleword_shift_condmove): Likewise.
17643 (expand_doubleword_shift): Likewise.
17644 (expand_doubleword_clz): Likewise.
17645 (expand_doubleword_popcount): Likewise.
17646 (expand_doubleword_parity): Likewise.
17647 (expand_absneg_bit): Use scalar_int_mode for local variables.
17648 (prepare_float_lib_cmp): Likewise.
17649 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17650 rather than a machine_mode.
17651 (convert_memory_address_addr_space): Likewise.
17652 (get_mode_bounds): Likewise.
17653 (get_address_mode): Return a scalar_int_mode rather than a
17655 * rtlanal.c (get_address_mode): Likewise.
17656 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17657 than a machine_mode.
17658 * targhooks.c (default_mode_rep_extended): Likewise.
17659 (default_valid_pointer_mode): Likewise.
17660 (default_addr_space_valid_pointer_mode): Likewise.
17661 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17662 than a machine_mode.
17663 (default_libgcc_cmp_return_mode): Likewise.
17664 (default_libgcc_shift_count_mode): Likewise.
17665 (default_unwind_word_mode): Likewise.
17666 (default_addr_space_pointer_mode): Likewise.
17667 (default_addr_space_address_mode): Likewise.
17668 * targhooks.h (default_eh_return_filter_mode): Likewise.
17669 (default_libgcc_cmp_return_mode): Likewise.
17670 (default_libgcc_shift_count_mode): Likewise.
17671 (default_unwind_word_mode): Likewise.
17672 (default_addr_space_pointer_mode): Likewise.
17673 (default_addr_space_address_mode): Likewise.
17674 (default_mode_rep_extended): Take a scalar_int_mode rather than
17676 (default_valid_pointer_mode): Likewise.
17677 (default_addr_space_valid_pointer_mode): Likewise.
17678 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17680 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17681 rather than a machine_mode.
17682 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17683 for local variables.
17684 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17685 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17686 than a machine_mode.
17688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17689 Alan Hayward <alan.hayward@arm.com>
17690 David Sherwood <david.sherwood@arm.com>
17692 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17693 the mode argument to scalar_int_mode.
17694 (do_jump_by_parts_zero_rtx): Likewise.
17695 (do_jump_by_parts_equality_rtx): Likewise.
17696 (do_jump_by_parts_greater): Take a mode argument.
17697 (do_jump_by_parts_equality): Likewise.
17698 (do_jump_1): Update calls accordingly.
17700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17701 Alan Hayward <alan.hayward@arm.com>
17702 David Sherwood <david.sherwood@arm.com>
17704 * is-a.h (safe_dyn_cast): New function.
17705 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17706 (jump_table_for_label): Likewise.
17707 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17708 instead of an rtx_insn *.
17709 (shorten_branches): Use dyn_cast instead of LABEL_P and
17710 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17711 rtx_jump_table_data::get_data_mode.
17712 (final_scan_insn): Likewise.
17714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17715 Alan Hayward <alan.hayward@arm.com>
17716 David Sherwood <david.sherwood@arm.com>
17718 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17719 trying to combine a full-register integer set with a subreg
17722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17723 Alan Hayward <alan.hayward@arm.com>
17724 David Sherwood <david.sherwood@arm.com>
17726 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17727 that is always either address_mode or pointer_mode.
17729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17730 Alan Hayward <alan.hayward@arm.com>
17731 David Sherwood <david.sherwood@arm.com>
17733 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17734 when the two are known to be equal.
17736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17737 Alan Hayward <alan.hayward@arm.com>
17738 David Sherwood <david.sherwood@arm.com>
17740 * simplify-rtx.c (simplify_const_unary_operation): Use
17741 is_a <scalar_int_mode> instead of checking for a nonzero
17742 precision. Forcibly convert op_mode to a scalar_int_mode
17743 in that case. More clearly differentiate the operand and
17744 result modes and use the former when deciding what the value
17745 of a count-bits operation should be. Use is_int_mode instead
17746 of checking for a MODE_INT. Remove redundant check for whether
17747 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
17754 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17755 (widen_bswap): Likewise.
17756 (expand_parity): Likewise.
17757 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17758 (expand_ffs): Likewise.
17759 (epand_unop): Check for scalar integer modes before calling the
17762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17763 Alan Hayward <alan.hayward@arm.com>
17764 David Sherwood <david.sherwood@arm.com>
17766 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17768 (expand_expr_real_1): Update call accordingly.
17770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17771 Alan Hayward <alan.hayward@arm.com>
17772 David Sherwood <david.sherwood@arm.com>
17774 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17775 value_mode arguments. Use scalar_int_mode internally.
17776 (store_bit_field_1): Rename the new integer mode from imode
17777 to op0_mode and use it instead of GET_MODE (op0). Update calls
17778 to store_split_bit_field, store_bit_field_using_insv and
17779 store_fixed_bit_field.
17780 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17781 Use scalar_int_mode internally. Use a bit count rather than a mode
17782 when calculating the largest bit size for get_best_mode.
17783 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17784 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17785 Remove assertion that OP0 has a scalar integer mode.
17786 (store_split_bit_field): Add op0_mode and value_mode arguments.
17787 Update calls to extract_fixed_bit_field.
17788 (extract_bit_field_using_extv): Add an op0_mode argument.
17789 Use scalar_int_mode internally.
17790 (extract_bit_field_1): Rename the new integer mode from imode to
17791 op0_mode and use it instead of GET_MODE (op0). Update calls to
17792 extract_split_bit_field, extract_bit_field_using_extv and
17793 extract_fixed_bit_field.
17794 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17795 to extract_split_bit_field and extract_fixed_bit_field_1.
17796 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17797 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17798 on the target mode.
17799 (extract_split_bit_field): Add an op0_mode argument. Update call
17800 to extract_fixed_bit_field.
17802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17803 Alan Hayward <alan.hayward@arm.com>
17804 David Sherwood <david.sherwood@arm.com>
17806 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17807 * explow.c (hard_function_value): Likewise.
17808 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17809 convert_to_mode call outside the loop.
17810 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17811 for the mode iterator. Require the mode specified by max_pieces
17813 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17815 (copy_blkmode_to_reg): Likewise.
17816 (set_storage_via_setmem): Likewise.
17817 * optabs.c (prepare_cmp_insn): Likewise.
17818 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17819 * stor-layout.c (finish_bitfield_representative): Likewise.
17821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17822 Alan Hayward <alan.hayward@arm.com>
17823 David Sherwood <david.sherwood@arm.com>
17825 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17826 * expr.c (convert_move): Use them.
17827 (convert_modes): Likewise.
17828 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17835 parameter for the mode of "x". Remove the "known_x", "known_mode"
17836 and "known_ret" arguments. Change the type of the mode argument
17837 to scalar_int_mode.
17838 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17839 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17840 (reg_num_sign_bit_copies_for_combine): Likewise.
17841 * rtlanal.c (nonzero_bits1): Likewise.
17842 (num_sign_bit_copies1): Likewise.
17843 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17844 (reg_num_sign_bit_copies_general): Likewise.
17845 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17846 (reg_nonzero_bits_general): Likewise.
17848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17849 Alan Hayward <alan.hayward@arm.com>
17850 David Sherwood <david.sherwood@arm.com>
17852 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17853 than in subroutines. Return 1 for non-integer modes.
17854 (cached_num_sign_bit_copies): Change the type of the mode parameter
17855 to scalar_int_mode.
17856 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17857 classes. Handle CONST_INT_P first and then check whether X also
17858 has a scalar integer mode. Check the same thing for inner registers
17859 of a SUBREG and for values that are being extended or truncated.
17861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17862 Alan Hayward <alan.hayward@arm.com>
17863 David Sherwood <david.sherwood@arm.com>
17865 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17866 in subroutines. Return the mode mask for non-integer modes.
17867 (cached_nonzero_bits): Change the type of the mode parameter
17868 to scalar_int_mode.
17869 (nonzero_bits1): Likewise. Remove early exit for other mode
17870 classes. Handle CONST_INT_P first and then check whether X
17871 also has a scalar integer mode.
17873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17874 Alan Hayward <alan.hayward@arm.com>
17875 David Sherwood <david.sherwood@arm.com>
17877 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17878 Return a scalar_int_mode and assert that the size is greater than
17880 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17881 (op_by_pieces_d::op_by_pieces_d): Likewise.
17882 (op_by_pieces_d::run): Likewise.
17883 (can_store_by_pieces): Likewise.
17885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17886 Alan Hayward <alan.hayward@arm.com>
17887 David Sherwood <david.sherwood@arm.com>
17889 * combine.c (extract_left_shift): Add a mode argument and update
17891 (make_compound_operation_int): Change the type of the mode parameter
17892 to scalar_int_mode and update the call to extract_left_shift.
17894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17895 Alan Hayward <alan.hayward@arm.com>
17896 David Sherwood <david.sherwood@arm.com>
17898 * combine.c (simplify_and_const_int): Change the type of the mode
17899 parameter to scalar_int_mode.
17900 (simplify_and_const_int_1): Likewise. Update recursive call.
17902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17903 Alan Hayward <alan.hayward@arm.com>
17904 David Sherwood <david.sherwood@arm.com>
17906 * combine.c (simplify_compare_const): Check that the mode is a
17907 scalar_int_mode (rather than VOIDmode) before testing its
17909 (simplify_comparison): Move COMPARISON_P handling out of the
17910 loop and restrict the latter part of the loop to scalar_int_modes.
17911 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17912 and when considering SUBREG_REGs. Use is_int_mode instead of
17913 checking GET_MODE_CLASS against MODE_INT.
17915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17916 Alan Hayward <alan.hayward@arm.com>
17917 David Sherwood <david.sherwood@arm.com>
17919 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17920 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17921 shift_unit_mode and for modes involved in scalar shifts.
17923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17924 Alan Hayward <alan.hayward@arm.com>
17925 David Sherwood <david.sherwood@arm.com>
17927 * combine.c (force_int_to_mode): New function, split out from...
17928 (force_to_mode): ...here. Keep xmode up-to-date and use it
17929 instead of GET_MODE (x).
17931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17932 Alan Hayward <alan.hayward@arm.com>
17933 David Sherwood <david.sherwood@arm.com>
17935 * optabs-query.h (extraction_insn::struct_mode): Change type to
17936 opt_scalar_int_mode and update comment.
17937 (extraction_insn::field_mode): Change type to scalar_int_mode.
17938 (extraction_insn::pos_mode): Likewise.
17939 * combine.c (make_extraction): Update accordingly.
17940 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17941 (get_optab_extraction_insn): Likewise.
17942 * recog.c (simplify_while_replacing): Likewise.
17943 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17944 parameter to opt_scalar_int_mode.
17946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17947 Alan Hayward <alan.hayward@arm.com>
17948 David Sherwood <david.sherwood@arm.com>
17950 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17951 to a scalar_int_mode instead of a machine_mode.
17952 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17953 (get_best_mode): Return a boolean and use a pointer argument to store
17954 the selected mode. Replace the limit mode parameter with a bit limit.
17955 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17956 for the values returned by bit_field_mode_iterator::next_mode.
17957 (store_bit_field): Update call to get_best_mode.
17958 (store_fixed_bit_field): Likewise.
17959 (extract_fixed_bit_field): Likewise.
17960 * expr.c (optimize_bitfield_assignment_op): Likewise.
17961 * fold-const.c (optimize_bit_field_compare): Likewise.
17962 (fold_truth_andor_1): Likewise.
17963 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17964 Update for new type of m_mode.
17965 (get_best_mode): As above.
17967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17968 Alan Hayward <alan.hayward@arm.com>
17969 David Sherwood <david.sherwood@arm.com>
17971 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17972 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17973 (store_bit_field): Check is_a <scalar_int_mode> before calling
17974 strict_volatile_bitfield_p.
17975 (extract_bit_field): Likewise.
17977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17978 Alan Hayward <alan.hayward@arm.com>
17979 David Sherwood <david.sherwood@arm.com>
17981 * target.def (cstore_mode): Return a scalar_int_mode.
17982 * doc/tm.texi: Regenerate.
17983 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17984 * targhooks.h (default_cstore_mode): Likewise.
17985 * targhooks.c (default_cstore_mode): Likewise, using a forced
17987 * expmed.c (emit_cstore): Expect the target of the cstore to be
17990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17991 Alan Hayward <alan.hayward@arm.com>
17992 David Sherwood <david.sherwood@arm.com>
17994 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17996 (niter_desc): Likewise mode.
17997 (iv_analyze): Add a mode parameter.
17999 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18000 and change its type to scalar_int_mode.
18001 * loop-iv.c: Update commentary at head of file.
18002 (iv_constant): Pass the mode paraeter before the rtx it describes
18003 and change its type to scalar_int_mode. Remove VOIDmode handling.
18004 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18005 (iv_extend): Likewise.
18006 (shorten_into_mode): Likewise.
18007 (iv_add): Use scalar_int_mode.
18008 (iv_mult): Likewise.
18009 (iv_shift): Likewise.
18010 (canonicalize_iv_subregs): Likewise.
18011 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18012 it describes and change its mode to scalar_int_mode. Also change
18013 the type of the returned inner_mode to scalar_int_mode.
18014 (get_biv_step): Likewise, turning outer_mode from a pointer
18015 into a direct parameter. Update call to get_biv_step_1.
18016 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18017 iv_constant and get_biv_step.
18018 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18019 and change its type to scalar_int_mode. Don't initialise iv->mode
18020 to VOIDmode and remove later checks for its still being VOIDmode.
18021 Update calls to iv_analyze_op and iv_analyze_expr. Check
18022 is_a <scalar_int_mode> when changing the mode under consideration.
18023 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18024 Update call to iv_analyze_expr.
18025 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18026 inner register is not also a scalar_int_mode. Update call to
18028 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18029 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18030 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18031 separate mode class checks. Update calls to iv_analyze. Remove
18032 fix-up of VOIDmodes after iv_analyze_biv.
18033 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18034 don't have a scalar_int_mode. Update call to biv_p.
18036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18037 Alan Hayward <alan.hayward@arm.com>
18038 David Sherwood <david.sherwood@arm.com>
18040 * cfgexpand.c (convert_debug_memory_address): Use
18041 as_a <scalar_int_mode>.
18042 * combine.c (expand_compound_operation): Likewise.
18043 (make_extraction): Likewise.
18044 (change_zero_ext): Likewise.
18045 (simplify_comparison): Likewise.
18046 * cse.c (cse_insn): Likewise.
18047 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18048 (mem_loc_descriptor): Likewise.
18049 (loc_descriptor): Likewise.
18050 * expmed.c (init_expmed_one_mode): Likewise.
18051 (synth_mult): Likewise.
18052 (emit_store_flag_1): Likewise.
18053 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18054 of a comparison with size.
18055 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18056 (reduce_to_bit_field_precision): Likewise.
18057 * function.c (expand_function_end): Likewise.
18058 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18059 * loop-doloop.c (doloop_modify): Likewise.
18060 * optabs.c (expand_binop): Likewise.
18061 (expand_unop): Likewise.
18062 (expand_copysign_absneg): Likewise.
18063 (prepare_cmp_insn): Likewise.
18064 (maybe_legitimize_operand): Likewise.
18065 * recog.c (const_scalar_int_operand): Likewise.
18066 * rtlanal.c (get_address_mode): Likewise.
18067 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18068 (simplify_cond_clz_ctz): Likewise.
18069 * tree-nested.c (get_nl_goto_field): Likewise.
18070 * tree.c (build_vector_type_for_mode): Likewise.
18071 * var-tracking.c (use_narrower_mode): Likewise.
18073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18074 Alan Hayward <alan.hayward@arm.com>
18075 David Sherwood <david.sherwood@arm.com>
18077 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18078 * builtins.c (expand_builtin_signbit): Use it.
18079 * cfgexpand.c (expand_debug_expr): Likewise.
18080 * dojump.c (do_jump): Likewise.
18081 (do_compare_and_jump): Likewise.
18082 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18083 * expmed.c (make_tree): Likewise.
18084 * expr.c (expand_expr_real_2): Likewise.
18085 (expand_expr_real_1): Likewise.
18086 (try_casesi): Likewise.
18087 * fold-const-call.c (fold_const_call_ss): Likewise.
18088 * fold-const.c (unextend): Likewise.
18089 (extract_muldiv_1): Likewise.
18090 (fold_single_bit_test): Likewise.
18091 (native_encode_int): Likewise.
18092 (native_encode_string): Likewise.
18093 (native_interpret_int): Likewise.
18094 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18095 * internal-fn.c (expand_addsub_overflow): Likewise.
18096 (expand_neg_overflow): Likewise.
18097 (expand_mul_overflow): Likewise.
18098 (expand_arith_overflow): Likewise.
18099 * match.pd: Likewise.
18100 * stor-layout.c (layout_type): Likewise.
18101 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18102 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18103 * tree-ssanames.c (get_range_info): Likewise.
18104 * tree-switch-conversion.c (array_value_type) Likewise.
18105 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18106 (vect_recog_divmod_pattern): Likewise.
18107 (vect_recog_mixed_size_cond_pattern): Likewise.
18108 * tree-vrp.c (extract_range_basic): Likewise.
18109 (simplify_float_conversion_using_ranges): Likewise.
18110 * tree.c (int_fits_type_p): Likewise.
18111 * ubsan.c (instrument_bool_enum_load): Likewise.
18112 * varasm.c (mergeable_string_section): Likewise.
18113 (narrowing_initializer_constant_valid_p): Likewise.
18114 (output_constant): Likewise.
18116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18117 Alan Hayward <alan.hayward@arm.com>
18118 David Sherwood <david.sherwood@arm.com>
18120 * machmode.h (NARROWEST_INT_MODE): New macro.
18121 * expr.c (alignment_for_piecewise_move): Use it instead of
18122 GET_CLASS_NARROWEST_MODE (MODE_INT).
18123 (push_block): Likewise.
18124 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18126 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18129 Alan Hayward <alan.hayward@arm.com>
18130 David Sherwood <david.sherwood@arm.com>
18132 * postreload.c (move2add_valid_value_p): Change the type of the
18133 mode parameter to scalar_int_mode.
18134 (move2add_use_add2_insn): Add a mode parameter and use it instead
18136 (move2add_use_add3_insn): Likewise.
18137 (reload_cse_move2add): Update accordingly.
18139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18140 Alan Hayward <alan.hayward@arm.com>
18141 David Sherwood <david.sherwood@arm.com>
18143 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18145 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18146 * optabs.c (expand_unop): Likewise.
18148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18149 Alan Hayward <alan.hayward@arm.com>
18150 David Sherwood <david.sherwood@arm.com>
18152 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18153 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18154 (popcount_loc_descriptor): Likewise.
18155 (bswap_loc_descriptor): Likewise.
18156 (rotate_loc_descriptor): Likewise.
18157 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18158 calling the functions above.
18160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18161 Alan Hayward <alan.hayward@arm.com>
18162 David Sherwood <david.sherwood@arm.com>
18164 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18166 (try_combine): Likewise.
18167 (simplify_if_then_else): Likewise.
18168 * cse.c (cse_insn): Likewise.
18169 * dwarf2out.c (mem_loc_descriptor): Likewise.
18170 * emit-rtl.c (gen_lowpart_common): Likewise.
18171 * simplify-rtx.c (simplify_truncation): Likewise.
18172 (simplify_binary_operation_1): Likewise.
18173 (simplify_const_relational_operation): Likewise.
18174 (simplify_ternary_operation): Likewise.
18175 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18178 Alan Hayward <alan.hayward@arm.com>
18179 David Sherwood <david.sherwood@arm.com>
18181 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18182 * reload.c (push_reload): Likewise.
18183 (find_reloads): Likewise.
18185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18186 Alan Hayward <alan.hayward@arm.com>
18187 David Sherwood <david.sherwood@arm.com>
18189 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18190 (make_compound_operation_int): Likewise.
18191 (change_zero_ext): Likewise.
18192 * expr.c (convert_move): Likewise.
18193 (convert_modes): Likewise.
18194 * fwprop.c (forward_propagate_subreg): Likewise.
18195 * loop-iv.c (get_biv_step_1): Likewise.
18196 * optabs.c (widen_operand): Likewise.
18197 * postreload.c (move2add_valid_value_p): Likewise.
18198 * recog.c (simplify_while_replacing): Likewise.
18199 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18200 (simplify_binary_operation_1): Likewise. Remove redundant
18201 mode equality check.
18203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18204 Alan Hayward <alan.hayward@arm.com>
18205 David Sherwood <david.sherwood@arm.com>
18207 * combine.c (combine_simplify_rtx): Add checks for
18208 is_a <scalar_int_mode>.
18209 (simplify_if_then_else): Likewise.
18210 (make_field_assignment): Likewise.
18211 (simplify_comparison): Likewise.
18212 * ifcvt.c (noce_try_bitop): Likewise.
18213 * loop-invariant.c (canonicalize_address_mult): Likewise.
18214 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18217 Alan Hayward <alan.hayward@arm.com>
18218 David Sherwood <david.sherwood@arm.com>
18220 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18221 is_a <scalar_int_mode> instead of != BLKmode.
18223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18224 Alan Hayward <alan.hayward@arm.com>
18225 David Sherwood <david.sherwood@arm.com>
18227 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18228 instead of != VOIDmode.
18229 * combine.c (if_then_else_cond): Likewise.
18230 (change_zero_ext): Likewise.
18231 * dwarf2out.c (mem_loc_descriptor): Likewise.
18232 (loc_descriptor): Likewise.
18233 * rtlanal.c (canonicalize_condition): Likewise.
18234 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18237 Alan Hayward <alan.hayward@arm.com>
18238 David Sherwood <david.sherwood@arm.com>
18240 * simplify-rtx.c (simplify_binary_operation_1): Use
18241 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18244 Alan Hayward <alan.hayward@arm.com>
18245 David Sherwood <david.sherwood@arm.com>
18247 * wide-int.h (int_traits<unsigned char>) New class.
18248 (int_traits<unsigned short>) Likewise.
18249 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18250 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18252 * combine.c (set_nonzero_bits_and_sign_copies): Use
18253 is_a <scalar_int_mode>.
18254 (find_split_point): Likewise.
18255 (combine_simplify_rtx): Likewise.
18256 (simplify_logical): Likewise.
18257 (expand_compound_operation): Likewise.
18258 (expand_field_assignment): Likewise.
18259 (make_compound_operation): Likewise.
18260 (extended_count): Likewise.
18261 (change_zero_ext): Likewise.
18262 (simplify_comparison): Likewise.
18263 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18264 (ucompare_loc_descriptor): Likewise.
18265 (minmax_loc_descriptor): Likewise.
18266 (mem_loc_descriptor): Likewise.
18267 (loc_descriptor): Likewise.
18268 * expmed.c (init_expmed_one_mode): Likewise.
18269 * lra-constraints.c (lra_constraint_offset): Likewise.
18270 * optabs.c (prepare_libcall_arg): Likewise.
18271 * postreload.c (move2add_note_store): Likewise.
18272 * reload.c (operands_match_p): Likewise.
18273 * rtl.h (load_extend_op): Likewise.
18274 * rtlhooks.c (gen_lowpart_general): Likewise.
18275 * simplify-rtx.c (simplify_truncation): Likewise.
18276 (simplify_unary_operation_1): Likewise.
18277 (simplify_binary_operation_1): Likewise.
18278 (simplify_const_binary_operation): Likewise.
18279 (simplify_const_relational_operation): Likewise.
18280 (simplify_subreg): Likewise.
18281 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18282 * var-tracking.c (adjust_mems): Likewise.
18283 (prepare_call_arguments): Likewise.
18285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18286 Alan Hayward <alan.hayward@arm.com>
18287 David Sherwood <david.sherwood@arm.com>
18289 * machmode.h (is_int_mode): New fuction.
18290 * combine.c (find_split_point): Use it.
18291 (combine_simplify_rtx): Likewise.
18292 (simplify_if_then_else): Likewise.
18293 (simplify_set): Likewise.
18294 (simplify_shift_const_1): Likewise.
18295 (simplify_comparison): Likewise.
18296 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18297 * cse.c (notreg_cost): Likewise.
18298 (cse_insn): Likewise.
18299 * cselib.c (cselib_lookup_1): Likewise.
18300 * dojump.c (do_jump_1): Likewise.
18301 (do_compare_rtx_and_jump): Likewise.
18302 * dse.c (get_call_args): Likewise.
18303 * dwarf2out.c (rtl_for_decl_init): Likewise.
18304 (native_encode_initializer): Likewise.
18305 * expmed.c (emit_store_flag_1): Likewise.
18306 (emit_store_flag): Likewise.
18307 * expr.c (convert_modes): Likewise.
18308 (store_field): Likewise.
18309 (expand_expr_real_1): Likewise.
18310 * fold-const.c (fold_read_from_constant_string): Likewise.
18311 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18312 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18313 * optabs.c (expand_binop): Likewise.
18314 (expand_unop): Likewise.
18315 (expand_abs_nojump): Likewise.
18316 (expand_one_cmpl_abs_nojump): Likewise.
18317 * simplify-rtx.c (mode_signbit_p): Likewise.
18318 (val_signbit_p): Likewise.
18319 (val_signbit_known_set_p): Likewise.
18320 (val_signbit_known_clear_p): Likewise.
18321 (simplify_relational_operation_1): Likewise.
18322 * tree.c (vector_type_mode): Likewise.
18324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18325 Alan Hayward <alan.hayward@arm.com>
18326 David Sherwood <david.sherwood@arm.com>
18328 * machmode.h (smallest_mode_for_size): Fix formatting.
18329 (smallest_int_mode_for_size): New function.
18330 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18331 instead of smallest_mode_for_size.
18332 * combine.c (make_extraction): Likewise.
18333 * config/arc/arc.c (arc_expand_movmem): Likewise.
18334 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18335 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18336 * config/s390/s390.c (s390_expand_insv): Likewise.
18337 * config/sparc/sparc.c (assign_int_registers): Likewise.
18338 * config/spu/spu.c (spu_function_value): Likewise.
18339 (spu_function_arg): Likewise.
18340 * coverage.c (get_gcov_type): Likewise.
18341 (get_gcov_unsigned_t): Likewise.
18342 * dse.c (find_shift_sequence): Likewise.
18343 * expmed.c (store_bit_field_1): Likewise.
18344 * expr.c (convert_move): Likewise.
18345 (store_field): Likewise.
18346 * internal-fn.c (expand_arith_overflow): Likewise.
18347 * optabs-query.c (get_best_extraction_insn): Likewise.
18348 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18349 * stor-layout.c (layout_type): Likewise.
18350 (initialize_sizetypes): Likewise.
18351 * targhooks.c (default_get_mask_mode): Likewise.
18352 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18355 Alan Hayward <alan.hayward@arm.com>
18356 David Sherwood <david.sherwood@arm.com>
18358 * machmode.h (opt_mode::else_blk): New function.
18359 (int_mode_for_mode): Declare.
18360 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18361 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18363 * cfgexpand.c (expand_debug_expr): Likewise.
18364 * combine.c (gen_lowpart_or_truncate): Likewise.
18365 (gen_lowpart_for_combine): Likewise.
18366 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18367 * config/avr/avr.c (avr_to_int_mode): Likewise.
18368 (avr_out_plus_1): Likewise.
18369 (avr_out_plus): Likewise.
18370 (avr_out_round): Likewise.
18371 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18372 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18373 (s390_expand_vcond): Likewise.
18374 * config/spu/spu.c (spu_split_immediate): Likewise.
18375 (spu_expand_mov): Likewise.
18376 * dse.c (get_stored_val): Likewise.
18377 * expmed.c (store_bit_field_1): Likewise.
18378 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18380 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18381 (extract_low_bits): Likewise.
18382 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18383 handling rather than repeating the check.
18384 (emit_group_store): Likewise.
18385 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18386 * optabs.c (expand_absneg_bit): Likewise.
18387 (expand_copysign_absneg): Likewise.
18388 (expand_copysign_bit): Likewise.
18389 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18390 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18391 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18392 * var-tracking.c (prepare_call_arguments): Likewise.
18393 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18394 int_mode_for_mode instead of mode_for_size.
18395 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18398 Alan Hayward <alan.hayward@arm.com>
18399 David Sherwood <david.sherwood@arm.com>
18401 * machmode.h (int_mode_for_size): New function.
18402 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18403 instead of mode_for_size.
18404 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18406 * combine.c (expand_field_assignment): Use int_mode_for_size
18407 instead of mode_for_size.
18408 (make_extraction): Likewise.
18409 (simplify_shift_const_1): Likewise.
18410 (simplify_comparison): Likewise.
18411 * dojump.c (do_jump): Likewise.
18412 * dwarf2out.c (mem_loc_descriptor): Likewise.
18413 * emit-rtl.c (init_derived_machine_modes): Likewise.
18414 * expmed.c (flip_storage_order): Likewise.
18415 (convert_extracted_bit_field): Likewise.
18416 * expr.c (copy_blkmode_from_reg): Likewise.
18417 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18418 * internal-fn.c (expand_mul_overflow): Likewise.
18419 * lower-subreg.c (simple_move): Likewise.
18420 * optabs-libfuncs.c (init_optabs): Likewise.
18421 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18422 * tree.c (vector_type_mode): Likewise.
18423 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18424 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18425 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18426 * tree-vect-stmts.c (vectorizable_load): Likewise.
18427 (vectorizable_store): Likewise.
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 * coretypes.h (pod_mode): New type.
18434 (scalar_int_mode_pod): New typedef.
18435 * machmode.h (pod_mode): New class.
18436 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18437 * genmodes.c (emit_mode_int_n): Update accordingly.
18438 * lower-subreg.h (target_lower_subreg): Change type to
18439 scalar_int_mode_pod.
18440 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18441 scalar_int_mode_pod.
18443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18444 Alan Hayward <alan.hayward@arm.com>
18445 David Sherwood <david.sherwood@arm.com>
18447 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18448 machine_mode to scalar_int_mode.
18449 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18450 (rs6000_option_override_internal): Remove cast to int.
18451 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18452 machine_mode to scalar_int_mode.
18453 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18454 (rs6000_option_override_internal): Remove cast to int.
18455 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18456 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18458 * config/s390/s390.c (s390_expand_builtin): Likewise.
18459 * coretypes.h (scalar_int_mode): New type.
18460 (opt_scalar_int_mode): New typedef.
18461 * machmode.h (scalar_int_mode): New class.
18462 (scalar_int_mode::includes_p): New function.
18463 (byte_mode): Change type to scalar_int_mode.
18464 (word_mode): Likewise.
18465 (ptr_mode): Likewise.
18466 * emit-rtl.c (byte_mode): Likewise.
18467 (word_mode): Likewise.
18468 (ptr_mode): Likewise.
18469 (init_derived_machine_modes): Update accordingly.
18470 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18471 and MODE_PARTIAL_INT.
18472 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18473 opt_scalar_int_mode.
18475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18476 Alan Hayward <alan.hayward@arm.com>
18477 David Sherwood <david.sherwood@arm.com>
18479 * target.def (libgcc_floating_mode_supported_p): Take a
18481 * doc/tm.texi: Regenerate.
18482 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18484 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18485 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18489 Alan Hayward <alan.hayward@arm.com>
18490 David Sherwood <david.sherwood@arm.com>
18492 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18493 * doc/tm.texi: Regenerate.
18494 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18495 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18496 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18497 * targhooks.h (default_floatn_mode): Likewise.
18498 * targhooks.c (default_floatn_mode): Likewise.
18499 * tree.c (build_common_tree_nodes): Update accordingly.
18501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18502 Alan Hayward <alan.hayward@arm.com>
18503 David Sherwood <david.sherwood@arm.com>
18505 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18506 (mode_iterator::iterate_p): Likewise.
18507 (mode_iterator::get_wider): Likewise.
18508 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18511 Alan Hayward <alan.hayward@arm.com>
18512 David Sherwood <david.sherwood@arm.com>
18514 * coretypes.h (opt_scalar_float_mode): New typedef.
18515 * machmode.h (float_mode_for_size): New function.
18516 * emit-rtl.c (double_mode): Delete.
18517 (init_emit_once): Use float_mode_for_size.
18518 * stor-layout.c (layout_type): Likewise.
18519 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18522 Alan Hayward <alan.hayward@arm.com>
18523 David Sherwood <david.sherwood@arm.com>
18525 * output.h (assemble_real): Take a scalar_float_mode.
18526 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18527 * config/arm/arm.md (consttable_4): Likewise.
18528 (consttable_8): Likewise.
18529 (consttable_16): Likewise.
18530 * config/mips/mips.md (consttable_float): Likewise.
18531 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18532 * varasm.c (assemble_real): Take a scalar_float_mode.
18533 (output_constant_pool_2): Update accordingly.
18534 (output_constant): Likewise.
18536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18537 Alan Hayward <alan.hayward@arm.com>
18538 David Sherwood <david.sherwood@arm.com>
18540 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18541 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18542 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18543 (native_encode_real): Likewise.
18544 (native_interpret_real): Likewise.
18545 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18546 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18549 Alan Hayward <alan.hayward@arm.com>
18550 David Sherwood <david.sherwood@arm.com>
18552 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18553 <scalar_float_mode>. Simplify.
18554 (gen_extend_conv_libfunc): Likewise.
18556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18557 Alan Hayward <alan.hayward@arm.com>
18558 David Sherwood <david.sherwood@arm.com>
18560 * coretypes.h (scalar_float_mode): New type.
18561 * machmode.h (mode_traits::from_int): Use machine_mode if
18562 USE_ENUM_MODES is defined.
18563 (is_a): New function.
18565 (dyn_cast): Likewise.
18566 (scalar_float_mode): New class.
18567 (scalar_float_mode::includes_p): New function.
18568 (is_float_mode): Likewise.
18569 * gdbhooks.py (MachineModePrinter): New class.
18570 (build_pretty_printer): Use it for scalar_float_mode.
18571 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18572 (format_helper::format_helper): Turn into a template.
18573 * genmodes.c (get_mode_class): New function.
18574 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18575 or machine_mode if none.
18576 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18577 as_a <scalar_float_mode>.
18578 * dwarf2out.c (mem_loc_descriptor): Likewise.
18579 (insert_float): Likewise.
18580 (add_const_value_attribute): Likewise.
18581 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18582 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18583 (expand_unop): Update accordingly.
18584 (expand_abs_nojump): Likewise.
18585 (expand_copysign_absneg): Take a scalar_float_mode.
18586 (expand_copysign_bit): Likewise.
18587 (expand_copysign): Update accordingly.
18589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18590 Alan Hayward <alan.hayward@arm.com>
18591 David Sherwood <david.sherwood@arm.com>
18593 * coretypes.h (opt_mode): New class.
18594 * machmode.h (opt_mode): Likewise.
18595 (opt_mode::else_void): New function.
18596 (opt_mode::require): Likewise.
18597 (opt_mode::exists): Likewise.
18598 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18599 (GET_MODE_2XWIDER_MODE): Likewise.
18600 (mode_iterator::get_wider): Update accordingly.
18601 (mode_iterator::get_2xwider): Likewise.
18602 (mode_iterator::get_known_wider): Likewise, turning into a template.
18603 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18604 forcing a wider mode to exist.
18605 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18606 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18607 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18608 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18609 * lower-subreg.c (init_lower_subreg): Likewise.
18610 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18611 on the final iteration.
18612 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18613 a wider mode exists before asking for a move pattern.
18614 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18615 forcing a wider mode to exist.
18616 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18617 returning false if no such mode exists.
18618 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18619 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18620 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18621 Avoid checking for a MODE_INT if we already know the mode is not a
18623 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18624 forcing a wider mode to exist.
18625 (expmed_mult_highpart_optab): Likewise.
18626 (expmed_mult_highpart): Likewise.
18627 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18629 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18630 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18631 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18632 * internal-fn.c (expand_mul_overflow): Update use of
18633 GET_MODE_2XWIDER_MODE.
18634 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18635 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18636 GET_MODE_WIDER_MODE.
18637 (convert_plusminus_to_widen): Likewise.
18638 * tree-switch-conversion.c (array_value_type): Likewise.
18639 * var-tracking.c (emit_note_insn_var_location): Likewise.
18640 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18641 Return false inside rather than outside the loop if no wider mode
18643 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18644 and GET_MODE_2XWIDER_MODE
18645 (can_compare_p): Use else_void.
18646 * gdbhooks.py (OptMachineModePrinter): New class.
18647 (build_pretty_printer): Use it for opt_mode.
18649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18650 Alan Hayward <alan.hayward@arm.com>
18651 David Sherwood <david.sherwood@arm.com>
18653 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18654 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18657 Alan Hayward <alan.hayward@arm.com>
18658 David Sherwood <david.sherwood@arm.com>
18660 * machmode.h (mode_traits): New structure.
18661 (get_narrowest_mode): New function.
18662 (mode_iterator::start): Likewise.
18663 (mode_iterator::iterate_p): Likewise.
18664 (mode_iterator::get_wider): Likewise.
18665 (mode_iterator::get_known_wider): Likewise.
18666 (mode_iterator::get_2xwider): Likewise.
18667 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18668 (FOR_EACH_MODE): Likewise.
18669 (FOR_EACH_MODE_FROM): Likewise.
18670 (FOR_EACH_MODE_UNTIL): Likewise.
18671 (FOR_EACH_WIDER_MODE): Likewise.
18672 (FOR_EACH_2XWIDER_MODE): Likewise.
18673 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18674 * combine.c (simplify_comparison): Likewise
18675 * config/i386/i386.c (type_natural_mode): Likewise.
18676 * cse.c (cse_insn): Likewise.
18677 * dse.c (find_shift_sequence): Likewise.
18678 * emit-rtl.c (init_derived_machine_modes): Likewise.
18679 (init_emit_once): Likewise.
18680 * explow.c (hard_function_value): Likewise.
18681 * expmed.c (extract_fixed_bit_field_1): Likewise.
18682 (extract_bit_field_1): Likewise.
18683 (expand_divmod): Likewise.
18684 (emit_store_flag_1): Likewise.
18685 * expr.c (init_expr_target): Likewise.
18686 (convert_move): Likewise.
18687 (alignment_for_piecewise_move): Likewise.
18688 (widest_int_mode_for_size): Likewise.
18689 (emit_block_move_via_movmem): Likewise.
18690 (copy_blkmode_to_reg): Likewise.
18691 (set_storage_via_setmem): Likewise.
18692 (compress_float_constant): Likewise.
18693 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18694 * optabs-query.c (get_best_extraction_insn): Likewise.
18695 * optabs.c (expand_binop): Likewise.
18696 (expand_twoval_unop): Likewise.
18697 (expand_twoval_binop): Likewise.
18698 (widen_leading): Likewise.
18699 (widen_bswap): Likewise.
18700 (expand_parity): Likewise.
18701 (expand_unop): Likewise.
18702 (prepare_cmp_insn): Likewise.
18703 (prepare_float_lib_cmp): Likewise.
18704 (expand_float): Likewise.
18705 (expand_fix): Likewise.
18706 (expand_sfix_optab): Likewise.
18707 * postreload.c (move2add_use_add2_insn): Likewise.
18708 * reg-stack.c (reg_to_stack): Likewise.
18709 * reginfo.c (choose_hard_reg_mode): Likewise.
18710 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18711 * stor-layout.c (mode_for_size): Likewise.
18712 (smallest_mode_for_size): Likewise.
18713 (mode_for_vector): Likewise.
18714 (finish_bitfield_representative): Likewise.
18715 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18716 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18717 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18718 * var-tracking.c (prepare_call_arguments): Likewise.
18720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18721 Alan Hayward <alan.hayward@arm.com>
18722 David Sherwood <david.sherwood@arm.com>
18724 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18725 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18726 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18727 * machmode.h (mode_size): Move earlier in file.
18728 (mode_precision): Likewise.
18729 (mode_inner): Likewise.
18730 (mode_nunits): Likewise.
18731 (mode_unit_size): Likewise.
18732 (unit_unit_precision): Likewise.
18733 (mode_wider): Likewise.
18734 (mode_2xwider): Likewise.
18735 (machine_mode): New class.
18736 (mode_to_bytes): New function.
18737 (mode_to_bits): Likewise.
18738 (mode_to_precision): Likewise.
18739 (mode_to_inner): Likewise.
18740 (mode_to_unit_size): Likewise.
18741 (mode_to_unit_precision): Likewise.
18742 (mode_to_nunits): Likewise.
18743 (GET_MODE_SIZE): Use mode_to_bytes.
18744 (GET_MODE_BITSIZE): Use mode_to_bits.
18745 (GET_MODE_PRECISION): Use mode_to_precision.
18746 (GET_MODE_INNER): Use mode_to_inner.
18747 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18748 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18749 (GET_MODE_NUNITS): Use mode_to_nunits.
18750 * system.h (ALWAYS_INLINE): New macro.
18751 * config/powerpcspe/powerpcspe-c.c
18752 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18753 int for arg1_mode and arg2_mode.
18755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18756 Alan Hayward <alan.hayward@arm.com>
18757 David Sherwood <david.sherwood@arm.com>
18759 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18760 Prefix mode names with E_ in case statements.
18761 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18762 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18763 (aarch64_split_simd_move): Likewise.
18764 (aarch64_gen_storewb_pair): Likewise.
18765 (aarch64_gen_loadwb_pair): Likewise.
18766 (aarch64_gen_store_pair): Likewise.
18767 (aarch64_gen_load_pair): Likewise.
18768 (aarch64_get_condition_code_1): Likewise.
18769 (aarch64_constant_pool_reload_icode): Likewise.
18770 (get_rsqrte_type): Likewise.
18771 (get_rsqrts_type): Likewise.
18772 (get_recpe_type): Likewise.
18773 (get_recps_type): Likewise.
18774 (aarch64_gimplify_va_arg_expr): Likewise.
18775 (aarch64_simd_container_mode): Likewise.
18776 (aarch64_emit_load_exclusive): Likewise.
18777 (aarch64_emit_store_exclusive): Likewise.
18778 (aarch64_expand_compare_and_swap): Likewise.
18779 (aarch64_gen_atomic_cas): Likewise.
18780 (aarch64_emit_bic): Likewise.
18781 (aarch64_emit_atomic_swap): Likewise.
18782 (aarch64_emit_atomic_load_op): Likewise.
18783 (aarch64_evpc_trn): Likewise.
18784 (aarch64_evpc_uzp): Likewise.
18785 (aarch64_evpc_zip): Likewise.
18786 (aarch64_evpc_ext): Likewise.
18787 (aarch64_evpc_rev): Likewise.
18788 (aarch64_evpc_dup): Likewise.
18789 (aarch64_gen_ccmp_first): Likewise.
18790 (aarch64_gen_ccmp_next): Likewise.
18791 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18792 (alpha_emit_xfloating_libcall): Likewise.
18793 (emit_insxl): Likewise.
18794 (alpha_arg_type): Likewise.
18795 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18796 (arc_preferred_simd_mode): Likewise.
18797 (arc_secondary_reload): Likewise.
18798 (get_arc_condition_code): Likewise.
18799 (arc_print_operand): Likewise.
18800 (arc_legitimate_constant_p): Likewise.
18801 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18802 * config/arc/arc.md (casesi_load): Likewise.
18803 (casesi_compact_jump): Likewise.
18804 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18805 (cc_use_register): Likewise.
18806 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18807 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18808 (arm_init_iwmmxt_builtins): Likewise.
18809 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18810 (neon_expand_vector_init): Likewise.
18811 (arm_attr_length_move_neon): Likewise.
18812 (maybe_get_arm_condition_code): Likewise.
18813 (arm_emit_vector_const): Likewise.
18814 (arm_preferred_simd_mode): Likewise.
18815 (arm_output_iwmmxt_tinsr): Likewise.
18816 (thumb1_output_casesi): Likewise.
18817 (thumb2_output_casesi): Likewise.
18818 (arm_emit_load_exclusive): Likewise.
18819 (arm_emit_store_exclusive): Likewise.
18820 (arm_expand_compare_and_swap): Likewise.
18821 (arm_evpc_neon_vuzp): Likewise.
18822 (arm_evpc_neon_vzip): Likewise.
18823 (arm_evpc_neon_vrev): Likewise.
18824 (arm_evpc_neon_vtrn): Likewise.
18825 (arm_evpc_neon_vext): Likewise.
18826 (arm_validize_comparison): Likewise.
18827 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18828 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18829 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18830 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18831 (c6x_preferred_simd_mode): Likewise.
18832 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18833 (epiphany_rtx_costs): Likewise.
18834 * config/epiphany/predicates.md (proper_comparison_operator):
18836 * config/frv/frv.c (condexec_memory_operand): Likewise.
18837 (frv_emit_move): Likewise.
18838 (output_move_single): Likewise.
18839 (output_condmove_single): Likewise.
18840 (frv_hard_regno_mode_ok): Likewise.
18841 (frv_matching_accg_mode): Likewise.
18842 * config/h8300/h8300.c (split_adds_subs): Likewise.
18843 (h8300_rtx_costs): Likewise.
18844 (h8300_print_operand): Likewise.
18845 (compute_mov_length): Likewise.
18846 (output_logical_op): Likewise.
18847 (compute_logical_op_length): Likewise.
18848 (compute_logical_op_cc): Likewise.
18849 (h8300_shift_needs_scratch_p): Likewise.
18850 (output_a_shift): Likewise.
18851 (compute_a_shift_length): Likewise.
18852 (compute_a_shift_cc): Likewise.
18853 (expand_a_rotate): Likewise.
18854 (output_a_rotate): Likewise.
18855 * config/i386/i386.c (classify_argument): Likewise.
18856 (function_arg_advance_32): Likewise.
18857 (function_arg_32): Likewise.
18858 (function_arg_64): Likewise.
18859 (function_value_64): Likewise.
18860 (ix86_gimplify_va_arg): Likewise.
18861 (ix86_legitimate_constant_p): Likewise.
18862 (put_condition_code): Likewise.
18863 (split_double_mode): Likewise.
18864 (ix86_avx256_split_vector_move_misalign): Likewise.
18865 (ix86_expand_vector_logical_operator): Likewise.
18866 (ix86_split_idivmod): Likewise.
18867 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18868 (ix86_build_const_vector): Likewise.
18869 (ix86_build_signbit_mask): Likewise.
18870 (ix86_match_ccmode): Likewise.
18871 (ix86_cc_modes_compatible): Likewise.
18872 (ix86_expand_branch): Likewise.
18873 (ix86_expand_sse_cmp): Likewise.
18874 (ix86_expand_sse_movcc): Likewise.
18875 (ix86_expand_int_sse_cmp): Likewise.
18876 (ix86_expand_vec_perm_vpermi2): Likewise.
18877 (ix86_expand_vec_perm): Likewise.
18878 (ix86_expand_sse_unpack): Likewise.
18879 (ix86_expand_int_addcc): Likewise.
18880 (ix86_split_to_parts): Likewise.
18881 (ix86_vectorize_builtin_gather): Likewise.
18882 (ix86_vectorize_builtin_scatter): Likewise.
18883 (avx_vpermilp_parallel): Likewise.
18884 (inline_memory_move_cost): Likewise.
18885 (ix86_tieable_integer_mode_p): Likewise.
18886 (x86_maybe_negate_const_int): Likewise.
18887 (ix86_expand_vector_init_duplicate): Likewise.
18888 (ix86_expand_vector_init_one_nonzero): Likewise.
18889 (ix86_expand_vector_init_one_var): Likewise.
18890 (ix86_expand_vector_init_concat): Likewise.
18891 (ix86_expand_vector_init_interleave): Likewise.
18892 (ix86_expand_vector_init_general): Likewise.
18893 (ix86_expand_vector_set): Likewise.
18894 (ix86_expand_vector_extract): Likewise.
18895 (emit_reduc_half): Likewise.
18896 (ix86_emit_i387_round): Likewise.
18897 (ix86_mangle_type): Likewise.
18898 (ix86_expand_round_sse4): Likewise.
18899 (expand_vec_perm_blend): Likewise.
18900 (canonicalize_vector_int_perm): Likewise.
18901 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18902 (expand_vec_perm_1): Likewise.
18903 (expand_vec_perm_interleave3): Likewise.
18904 (expand_vec_perm_even_odd_pack): Likewise.
18905 (expand_vec_perm_even_odd_1): Likewise.
18906 (expand_vec_perm_broadcast_1): Likewise.
18907 (ix86_vectorize_vec_perm_const_ok): Likewise.
18908 (ix86_expand_vecop_qihi): Likewise.
18909 (ix86_expand_mul_widen_hilo): Likewise.
18910 (ix86_expand_sse2_abs): Likewise.
18911 (ix86_expand_pextr): Likewise.
18912 (ix86_expand_pinsr): Likewise.
18913 (ix86_preferred_simd_mode): Likewise.
18914 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18915 * config/i386/sse.md (*andnot<mode>3): Likewise.
18916 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18917 (*<code><mode>3): Likewise.
18918 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18919 (ia64_expand_atomic_op): Likewise.
18920 (ia64_arg_type): Likewise.
18921 (ia64_mode_to_int): Likewise.
18922 (ia64_scalar_mode_supported_p): Likewise.
18923 (ia64_vector_mode_supported_p): Likewise.
18924 (expand_vec_perm_broadcast): Likewise.
18925 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18926 (iq2000_function_arg_advance): Likewise.
18927 (iq2000_function_arg): Likewise.
18928 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18929 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18930 (m68k_libcall_value): Likewise.
18931 (m68k_function_value): Likewise.
18932 (sched_attr_op_type): Likewise.
18933 * config/mcore/mcore.c (mcore_output_move): Likewise.
18934 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18936 (microblaze_function_arg): Likewise.
18937 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18938 (mips_print_operand): Likewise.
18939 (mips_mode_ok_for_mov_fmt_p): Likewise.
18940 (mips_vector_mode_supported_p): Likewise.
18941 (mips_preferred_simd_mode): Likewise.
18942 (mips_expand_vpc_loongson_even_odd): Likewise.
18943 (mips_expand_vec_unpack): Likewise.
18944 (mips_expand_vi_broadcast): Likewise.
18945 (mips_expand_vector_init): Likewise.
18946 (mips_expand_vec_reduc): Likewise.
18947 (mips_expand_msa_cmp): Likewise.
18948 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18949 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18950 (cc_flags_for_mode): Likewise.
18951 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18952 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18953 (nds32_output_casesi_pc_relative): Likewise.
18954 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18955 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18956 (nvptx_gen_unpack): Likewise.
18957 (nvptx_gen_pack): Likewise.
18958 (nvptx_gen_shuffle): Likewise.
18959 (nvptx_gen_wcast): Likewise.
18960 (nvptx_preferred_simd_mode): Likewise.
18961 * config/pa/pa.c (pa_secondary_reload): Likewise.
18962 * config/pa/predicates.md (base14_operand): Likewise.
18963 * config/powerpcspe/powerpcspe-c.c
18964 (altivec_resolve_overloaded_builtin): Likewise.
18965 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18967 (rs6000_preferred_simd_mode): Likewise.
18968 (output_vec_const_move): Likewise.
18969 (rs6000_expand_vector_extract): Likewise.
18970 (rs6000_split_vec_extract_var): Likewise.
18971 (reg_offset_addressing_ok_p): Likewise.
18972 (rs6000_legitimate_offset_address_p): Likewise.
18973 (rs6000_legitimize_address): Likewise.
18974 (rs6000_emit_set_const): Likewise.
18975 (rs6000_const_vec): Likewise.
18976 (rs6000_emit_move): Likewise.
18977 (spe_build_register_parallel): Likewise.
18978 (rs6000_darwin64_record_arg_recurse): Likewise.
18979 (swap_selector_for_mode): Likewise.
18980 (spe_init_builtins): Likewise.
18981 (paired_init_builtins): Likewise.
18982 (altivec_init_builtins): Likewise.
18983 (do_load_for_compare): Likewise.
18984 (rs6000_generate_compare): Likewise.
18985 (rs6000_expand_float128_convert): Likewise.
18986 (emit_load_locked): Likewise.
18987 (emit_store_conditional): Likewise.
18988 (rs6000_output_function_epilogue): Likewise.
18989 (rs6000_handle_altivec_attribute): Likewise.
18990 (rs6000_function_value): Likewise.
18991 (emit_fusion_gpr_load): Likewise.
18992 (emit_fusion_p9_load): Likewise.
18993 (emit_fusion_p9_store): Likewise.
18994 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18995 (fusion_gpr_mem_load): Likewise.
18996 (fusion_addis_mem_combo_load): Likewise.
18997 (fusion_addis_mem_combo_store): Likewise.
18998 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18999 (fusion_gpr_mem_load): Likewise.
19000 (fusion_addis_mem_combo_load): Likewise.
19001 (fusion_addis_mem_combo_store): Likewise.
19002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19004 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19005 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19006 (rs6000_preferred_simd_mode): Likewise.
19007 (output_vec_const_move): Likewise.
19008 (rs6000_expand_vector_extract): Likewise.
19009 (rs6000_split_vec_extract_var): Likewise.
19010 (reg_offset_addressing_ok_p): Likewise.
19011 (rs6000_legitimate_offset_address_p): Likewise.
19012 (rs6000_legitimize_address): Likewise.
19013 (rs6000_emit_set_const): Likewise.
19014 (rs6000_const_vec): Likewise.
19015 (rs6000_emit_move): Likewise.
19016 (rs6000_darwin64_record_arg_recurse): Likewise.
19017 (swap_selector_for_mode): Likewise.
19018 (paired_init_builtins): Likewise.
19019 (altivec_init_builtins): Likewise.
19020 (rs6000_expand_float128_convert): Likewise.
19021 (emit_load_locked): Likewise.
19022 (emit_store_conditional): Likewise.
19023 (rs6000_output_function_epilogue): Likewise.
19024 (rs6000_handle_altivec_attribute): Likewise.
19025 (rs6000_function_value): Likewise.
19026 (emit_fusion_gpr_load): Likewise.
19027 (emit_fusion_p9_load): Likewise.
19028 (emit_fusion_p9_store): Likewise.
19029 * config/rx/rx.c (rx_gen_move_template): Likewise.
19030 (flags_from_mode): Likewise.
19031 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19032 (s390_slb_comparison): Likewise.
19033 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19034 (s390_vector_mode_supported_p): Likewise.
19035 (s390_cc_modes_compatible): Likewise.
19036 (s390_match_ccmode_set): Likewise.
19037 (s390_canonicalize_comparison): Likewise.
19038 (s390_emit_compare_and_swap): Likewise.
19039 (s390_branch_condition_mask): Likewise.
19040 (s390_rtx_costs): Likewise.
19041 (s390_secondary_reload): Likewise.
19042 (__SECONDARY_RELOAD_CASE): Likewise.
19043 (s390_expand_cs): Likewise.
19044 (s390_preferred_simd_mode): Likewise.
19045 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19046 * config/sh/sh.c (sh_print_operand): Likewise.
19047 (dump_table): Likewise.
19048 (sh_secondary_reload): Likewise.
19049 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19050 * config/sh/sh.md (casesi_worker_1): Likewise.
19051 (casesi_worker_2): Likewise.
19052 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19053 (fcc_comparison_operator): Likewise.
19054 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19055 (emit_soft_tfmode_cvt): Likewise.
19056 (sparc_preferred_simd_mode): Likewise.
19057 (output_cbranch): Likewise.
19058 (sparc_print_operand): Likewise.
19059 (sparc_expand_vec_perm_bmask): Likewise.
19060 (vector_init_bshuffle): Likewise.
19061 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19062 (spu_vector_mode_supported_p): Likewise.
19063 (spu_expand_insv): Likewise.
19064 (spu_emit_branch_or_set): Likewise.
19065 (spu_handle_vector_attribute): Likewise.
19066 (spu_builtin_splats): Likewise.
19067 (spu_builtin_extract): Likewise.
19068 (spu_builtin_promote): Likewise.
19069 (spu_expand_sign_extend): Likewise.
19070 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19071 (tilegx_simd_int): Likewise.
19072 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19073 (tilepro_simd_int): Likewise.
19074 * config/v850/v850.c (const_double_split): Likewise.
19075 (v850_print_operand): Likewise.
19076 (ep_memory_offset): Likewise.
19077 * config/vax/vax.c (vax_rtx_costs): Likewise.
19078 (vax_output_int_move): Likewise.
19079 (vax_output_int_add): Likewise.
19080 (vax_output_int_subtract): Likewise.
19081 * config/visium/predicates.md (visium_branch_operator): Likewise.
19082 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19083 (visium_print_operand_address): Likewise.
19084 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19085 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19086 (xtensa_expand_conditional_branch): Likewise.
19087 (xtensa_copy_incoming_a7): Likewise.
19088 (xtensa_output_literal): Likewise.
19089 * dfp.c (decimal_real_maxval): Likewise.
19090 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19093 Alan Hayward <alan.hayward@arm.com>
19094 David Sherwood <david.sherwood@arm.com>
19096 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19097 (mode_nunits_inline): Likewise.
19098 (mode_inner_inline): Likewise.
19099 (mode_unit_size_inline): Likewise.
19100 (mode_unit_precision_inline): Likewise.
19101 (emit_insn_modes_h): Likewise. Also emit a #define of the
19103 (emit_mode_wider): Add an E_ prefix to mode names.
19104 (emit_mode_complex): Likewise.
19105 (emit_mode_inner): Likewise.
19106 (emit_mode_adjustments): Likewise.
19107 (emit_mode_int_n): Likewise.
19108 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19109 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19110 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19111 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19112 (CRC32_BUILTIN, ENTRY): Likewise.
19113 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19114 (aarch64_pop_regs): Likewise.
19115 (aarch64_process_components): Likewise.
19116 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19117 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19118 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19119 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19120 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19121 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19123 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19124 (pcmpistr): Likewise.
19125 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19126 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19127 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19129 * config/rl78/rl78.c (mduc_regs): Likewise.
19130 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19131 (htm_expand_builtin): Likewise.
19132 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19133 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19134 * config/xtensa/xtensa.c (print_operand): Likewise.
19135 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19136 (NUM_MODE_VECTOR_INT): Likewise.
19137 * genoutput.c (null_operand): Likewise.
19138 (output_operand_data): Likewise.
19139 * genrecog.c (print_parameter_value): Likewise.
19140 * lra.c (debug_operand_data): Likewise.
19142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19144 * dwarf2out.c (scompare_loc_descriptor_wide)
19145 (scompare_loc_descriptor_narrow): New functions, split out from...
19146 (scompare_loc_descriptor): ...here.
19147 * expmed.c (emit_store_flag_int): New function, split out from...
19148 (emit_store_flag): ...here.
19150 2017-08-30 Richard Biener <rguenther@suse.de>
19152 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19153 (dwarf2out_early_finish): Move setting of AT_pubnames from
19154 early debug output to early finish.
19156 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19158 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19159 and -mdata-region to the assembler.
19161 2017-08-30 Richard Biener <rguenther@suse.de>
19163 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19165 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19166 (dwarf2out_early_global_decl): Only generate a DIE for the
19167 abstract origin if it doesn't already exist or is a declaration DIE.
19168 (resolve_addr): Do not add the linkage name twice when
19169 generating a stub DIE for the DW_TAG_GNU_call_site target.
19171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19173 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19174 Use machine_mode rather than int for arg1_mode.
19176 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19179 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19180 that the second argument of the built-in functions to unpack
19181 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19182 switch statement instead a lot of if statements.
19183 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19184 Allow 64-bit values to be in Altivec registers as well as
19185 traditional floating point registers.
19186 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19188 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19190 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19191 MAX_REGS_PER_ADDRESS == 1.
19193 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19195 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19196 * config/i386/i386.c (ix86_option_override_internal): Simplify
19197 setting of opts->x_flag_entry.
19199 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19200 Jakub Jelinek <jakub@redhat.com>
19201 Richard Biener <rguenther@suse.de>
19203 PR tree-optimization/81503
19204 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19205 folded constant fits in the target type; reorder tests for clarity.
19207 2017-08-29 Martin Liska <mliska@suse.cz>
19209 * passes.def: Include pass_lower_switch.
19210 * stmt.c (dump_case_nodes): Remove and move to
19211 tree-switch-conversion.
19212 (case_values_threshold): Likewise.
19213 (expand_switch_as_decision_tree_p): Likewise.
19214 (emit_case_decision_tree): Likewise.
19215 (expand_case): Likewise.
19216 (balance_case_nodes): Likewise.
19217 (node_has_low_bound): Likewise.
19218 (node_has_high_bound): Likewise.
19219 (node_is_bounded): Likewise.
19220 (emit_case_nodes): Likewise.
19221 (struct simple_case_node): New struct.
19222 (add_case_node): Remove.
19223 (emit_case_dispatch_table): Use vector instead of case_list.
19224 (reset_out_edges_aux): Remove.
19225 (compute_cases_per_edge): Likewise.
19226 (expand_case): Build list of simple_case_node.
19227 (expand_sjlj_dispatch_table): Use it.
19228 * tree-switch-conversion.c (struct case_node): Moved from
19229 stmt.c and adjusted.
19230 (emit_case_nodes): Likewise.
19231 (node_has_low_bound): Likewise.
19232 (node_has_high_bound): Likewise.
19233 (node_is_bounded): Likewise.
19234 (case_values_threshold): Likewise.
19235 (reset_out_edges_aux): Likewise.
19236 (compute_cases_per_edge): Likewise.
19237 (add_case_node): Likewise.
19238 (dump_case_nodes): Likewise.
19239 (balance_case_nodes): Likewise.
19240 (expand_switch_as_decision_tree_p): Likewise.
19241 (emit_jump): Likewise.
19242 (emit_case_decision_tree): Likewise.
19243 (try_switch_expansion): Likewise.
19244 (do_jump_if_equal): Likewise.
19245 (emit_cmp_and_jump_insns): Likewise.
19246 (fix_phi_operands_for_edge): New function.
19247 (record_phi_operand_mapping): Likewise.
19248 (class pass_lower_switch): New pass.
19249 (pass_lower_switch::execute): New function.
19250 (make_pass_lower_switch): Likewise.
19251 (conditional_probability):
19252 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19253 * tree-pass.h: Add make_pass_lower_switch.
19255 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19258 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19261 2017-08-29 Richard Biener <rguenther@suse.de>
19263 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19264 we do not add a DW_AT_inline attribute twice.
19265 (gen_subprogram_die): Remove code setting DW_AT_inline on
19266 DECL_ABSTRACT_P nodes.
19268 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19270 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19271 calls to internal functions.
19272 (gimplify_modify_expr): Likewise.
19273 * tree-call-cdce.c (use_internal_fn): Likewise.
19274 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19275 (convert_to_divmod): Set the nothrow flag.
19276 * tree-if-conv.c (predicate_mem_writes): Likewise.
19277 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19278 (vectorizable_call): Likewise.
19279 (vectorizable_store): Likewise.
19280 (vectorizable_load): Likewise.
19281 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19282 (vect_recog_mask_conversion_pattern): Likewise.
19284 2017-08-29 Martin Liska <mliska@suse.cz>
19287 * gcc.c (driver_handle_option): Add new argument.
19288 * opts-common.c (handle_option): Pass
19289 target_option_override_hook.
19290 * opts-global.c (lang_handle_option): Add new option.
19291 (set_default_handlers): Add new argument.
19292 (decode_options): Likewise.
19293 * opts.c (target_handle_option): Likewise.
19294 (common_handle_option): Call target_option_override_hook.
19295 * opts.h (struct cl_option_handler_func): Add hook for
19296 target option override.
19297 (struct cl_option_handlers): Likewise.
19298 (set_default_handlers): Add new argument.
19299 (decode_options): Likewise.
19300 (common_handle_option): Likewise.
19301 (target_handle_option): Likewise.
19302 * toplev.c (toplev::main): Pass targetm.target_option.override
19305 2017-08-29 Richard Biener <rguenther@suse.de>
19306 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19308 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19309 life to the active subtree.
19311 2017-08-28 Jeff Law <law@redhat.com>
19313 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19314 derive_equivalences.
19315 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19316 Code moved into....
19317 (edge_info::derive_equivalences): New private member function
19319 * tree-ssa-dom.c (class edge_info): Changed from a struct
19320 to a class. Add ctor/dtor, methods and data members.
19321 (edge_info::edge_info): Renamed from allocate_edge_info.
19322 Initialize additional members.
19323 (edge_info::~edge_info): New.
19324 (free_dom_edge_info): Delete the edge info.
19325 (record_edge_info): Use new class & associated member functions.
19326 Tighten forms for testing for edge equivalences.
19327 (record_temporary_equivalences): Iterate over the simple
19328 equivalences rather than assuming there's only one per edge.
19329 (cprop_into_successor_phis): Iterate over the simple
19330 equivalences rather than assuming there's only one per edge.
19331 (optimize_stmt): Use operand_equal_p rather than pointer
19332 equality for mini-DSE code.
19334 2017-08-28 Nathan Sidwell <nathan@acm.org>
19336 * gcc.c (execute): Fold SIGPIPE handling into switch
19337 statement. Adjust internal error message.
19339 2017-08-28 Richard Biener <rguenther@suse.de>
19342 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19343 Do nothing for removed DIEs.
19345 2017-08-28 Richard Biener <rguenther@suse.de>
19347 PR tree-optimization/81977
19348 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19351 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19354 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19355 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19356 using targetm.gen_mem_thread_fence.
19358 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19361 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19362 predicate to register_operand. Reorder operands.
19363 (*btr<mode>): Ditto.
19364 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19365 (*btr<mode>_mask): Ditto.
19367 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19369 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19370 * config/rs6000/xmmintrin.h: New file.
19371 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19373 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19376 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19377 parameter and_insn and return it.
19378 (recombine_lvx_pattern): Insert a copy to ensure availability of
19379 the base register of the copied masking operation at the point of
19380 the instruction replacement.
19381 (recombine_stvx_pattern): Likewise.
19383 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19385 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19386 undocumented switches.
19387 (-mpower9-dform-vector): Likewise.
19388 (-mpower9-dform): Likewise.
19389 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19390 comments to delete references to -mpower9-dform* switches.
19391 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19392 Delete reference to -mpower9-dform* switches, test for
19393 -mpower9-vector instead.
19394 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19395 (OTHER_P9_VECTOR_MASKS): Likewise.
19396 (POWERPC_MASKS): Likewise.
19397 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19398 tests against -mpower9-dform* to -mpower9-vector. Delete code
19399 that checked for -mpower9-dform* consistancy with other options.
19400 Add test for -mpower9-misc to enable other power9 switches.
19401 (rs6000_init_hard_regno_mode_ok): Likewise.
19402 (rs6000_option_override_internal): Likewise.
19403 (rs6000_emit_prologue): Likewise.
19404 (rs6000_emit_epilogue): Likewise.
19405 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19406 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19407 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19409 (emit_fusion_p9_store): Likewise.
19410 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19411 resetting these macros if the assembler does not support ISA 3.0
19413 (TARGET_P9_DFORM_VECTOR): Likewise.
19414 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19415 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19417 2017-08-25 Alan Modra <amodra@gmail.com>
19419 PR rtl-optimization/81747
19420 * cse.c (cse_extended_basic_block): Don't attempt to record
19421 equivalences for degenerate conditional branches that branch
19422 to their fall-through.
19424 2017-08-24 Martin Sebor <msebor@redhat.com>
19426 PR middle-end/81908
19427 * gimple-fold.c (size_must_be_zero_p): New function.
19428 (gimple_fold_builtin_memory_op): Call it.
19430 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19432 * config/rs6000/mm_malloc.h: New file.
19434 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19436 PR tree-optimization/81913
19437 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19438 analysis when either IVs in condition can wrap.
19440 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19442 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19443 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19445 2017-08-24 Richard Biener <rguenther@suse.de>
19448 * targhooks.c (default_target_can_inline_p): Properly
19449 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19450 is present and always compare.
19451 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19452 infer -mfpmath=sse from TARGET_SSE_P.
19453 (ix86_can_inline_p): Properly use target_option_default_node when
19454 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19456 2017-08-24 Richard Biener <rguenther@suse.de>
19459 * dwarf2out.c (output_die): Handle flag_generate_offload like
19461 (output_comp_unit): Likewise.
19462 (gen_array_type_die): Likewise.
19463 (dwarf2out_early_finish): Likewise.
19464 (note_variable_value_in_expr): Likewise.
19465 (dwarf2out_finish): Likewise. Adjust assert.
19466 * cgraphunit.c (symbol_table::compile): Move setting of
19467 flag_generate_offload earlier ...
19468 (symbol_table::finalize_compilation_unit): ... here, before
19469 early debug finalization.
19471 2017-08-24 Richard Biener <rguenther@suse.de>
19473 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19474 and ipa-fnsummary.h.
19475 (ix86_can_inline_p): When ix86_fpmath flags do not match
19476 check whether the callee uses FP math at all.
19478 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19480 PR middle-end/81931
19481 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19482 instead of TYPE_PRECISION.
19484 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19485 Alan Hayward <alan.hayward@arm.com>
19486 David Sherwood <david.sherwood@arm.com>
19488 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19490 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19492 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19494 (ix86_frame::stack_realign_allocate): New field.
19495 (struct machine_frame_state): Modify comments.
19496 (machine_frame_state::sp_realigned_fp_end): New field.
19497 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19498 layout calculation.
19499 (sp_valid_at): Add assertion to assure no attempt to access invalid
19500 offset of a realigned stack.
19501 (fp_valid_at): Likewise.
19502 (choose_baseaddr): Modify comments.
19503 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19504 ix86_expand_prologue.
19505 (ix86_expand_prologue): Modify stack realignment and allocation.
19506 (ix86_expand_epilogue): Modify comments.
19507 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19508 avx2_runtime, avx512f, and avx512f_runtime.
19510 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19512 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19513 (mstackrealign): Do not init to -1.
19514 * config/i386/i386.c (ix86_option_override_internal):
19515 Check opts_set, not opts when setting default value of
19516 opts->x_ix86_force_align_arg_pointer.
19518 2017-08-23 Richard Biener <rguenther@suse.de>
19520 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19521 lang_hooks.decl_printable_name.
19522 * print-rtl-function.c (print_rtx_function): Likewise.
19523 * tree-pretty-print.c (dump_function_header): Likewise.
19525 2017-08-23 Richard Biener <rguenther@suse.de>
19528 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19529 -g0 at compile-time.
19531 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19533 PR middle-end/19706
19534 * doc/sourcebuild.texi (Other hardware attributes):
19537 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19539 PR middle-end/19706
19540 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19541 Fix single-use check.
19543 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19545 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19547 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19549 * doc/install.texi: Modify to add more details on running selected
19552 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19554 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19555 is combined with -mabi=ms.
19556 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19559 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19561 PR tree-optimization/81488
19562 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19563 and cached_basis fields.
19564 (MAX_SPREAD): New constant.
19565 (alloc_cand_and_find_basis): Initialize new fields.
19566 (clear_visited): New function.
19567 (create_phi_basis_1): Rename from create_phi_basis, set visited
19568 and cached_basis fields.
19569 (create_phi_basis): New wrapper function.
19570 (phi_add_costs_1): Rename from phi_add_costs, add spread
19571 parameter, set visited field, short-circuit when limits reached.
19572 (phi_add_costs): New wrapper function.
19573 (record_phi_increments_1): Rename from record_phi_increments, set
19575 (record_phi_increments): New wrapper function.
19576 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19577 (phi_incr_cost): New wrapper function.
19578 (all_phi_incrs_profitable_1): Rename from
19579 all_phi_incrs_profitable, set visited field.
19580 (all_phi_incrs_profitable): New wrapper function.
19582 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19587 that takes the outer and inner modes.
19588 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19589 comparison as the canonical test for a paradoxical subreg.
19590 * combine.c (simplify_set): Use paradoxical_subreg_p.
19591 (make_extraction): Likewise.
19592 (force_to_mode): Likewise.
19593 (rtx_equal_for_field_assignment_p): Likewise.
19594 (gen_lowpart_for_combine): Likewise.
19595 (simplify_comparison): Likewise.
19596 * cse.c (equiv_constant): Likewise.
19597 * expmed.c (store_bit_field_1): Likewise.
19598 * final.c (alter_subreg): Likewise.
19599 * fwprop.c (propagate_rtx): Likewise.
19600 (forward_propagate_subreg): Likewise.
19601 * ira-conflicts.c (ira_build_conflicts): Likewise.
19602 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19603 * lra-constraints.c (curr_insn_transform): Likewise.
19604 (split_reg): Likewise.
19605 * lra-eliminations.c (move_plus_up): Likewise.
19606 (lra_eliminate_regs_1): Likewise.
19607 * recog.c (general_operand): Likewise.
19608 * ree.c (combine_reaching_defs): Likewise.
19609 * reload.c (push_reload): Likewise.
19610 (find_reloads): Likewise.
19611 * reload1.c (elimination_effects): Likewise.
19612 (compute_reload_subreg_offset): Likewise.
19613 (choose_reload_regs): Likewise.
19614 * rtlanal.c (subreg_lsb_1): Likewise.
19615 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19616 (simplify_subreg): Likewise.
19617 * var-tracking.c (track_loc_p): Likewise.
19618 * emit-rtl.c (byte_lowpart_offset): Likewise.
19619 (paradoxical_subreg_p): Delete out-of-line definition.
19621 2017-08-22 Jeff Law <law@redhat.com>
19623 PR tree-optimization/81741
19624 PR tree-optimization/71947
19625 * tree-ssa-dom.c: Include tree-inline.h.
19626 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19627 equivalences if one is more expensive to compute than the other.
19628 * tree-ssa-scopedtables.h (class const_or_copies): Make
19629 record_const_or_copy_raw method private.
19630 (class avail_exprs_stack): New method simplify_binary_operation.
19631 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19632 avail_exprs_stack::simplify_binary_operation as needed.
19633 (avail_exprs_stack::simplify_binary_operation): New function.
19635 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19637 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19638 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19639 (DOT_SYMBOLS): Likewise.
19640 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19641 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19642 (RS6000_ABI_NAME): Likewise.
19643 (TARGET_CMODEL): Likewise.
19644 (TOC_SECTION_ASM_OP): Likewise.
19645 (SET_CMODEL): New macro.
19646 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19648 2017-08-22 Richard Biener <rguenther@suse.de>
19650 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19651 to allow for free-lang-data replacements similar to verify_type_variant.
19653 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19654 Alan Hayward <alan.hayward@arm.com>
19655 David Sherwood <david.sherwood@arm.com>
19657 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19658 VOIDmode for the LABEL_REF.
19660 2017-08-22 Richard Biener <rguenther@suse.de>
19662 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19664 2017-08-22 Richard Biener <rguenther@suse.de>
19666 * common.opt (feliminate-dwarf2-dups): Ignore.
19667 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19668 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19669 same_die_p_wrap, compute_section_prefix,
19670 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19671 (comdat_symbol_id, comdat_symbol_number): Likewise.
19672 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19674 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19675 (output_die): Mark unreachable path unreachable.
19676 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19677 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19678 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19679 (dwarf2out_early_finish): Likewise.
19681 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19683 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19685 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19688 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19689 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19690 (avr_attribute_table) <io, io_low, address>: Initialize
19691 .decl_required with true.
19693 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19695 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19696 undocumented debugging options.
19697 (-mvsx-scalar-double): Likewise.
19698 (-mallow-df-permute): Likewise.
19699 (-mvectorize-builtins): Likewise.
19700 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19701 (rs6000_builtin_vectorized_function): Likewise.
19702 (rs6000_builtin_md_vectorized_function): Likewise.
19703 (rs6000_opt_vars): Likewise.
19705 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19708 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19709 (*btrq_imm): Rename from *btrq.
19710 (*btcq_imm): Rename from *btcq.
19711 (btsc): New code attribute.
19712 (*<btsc><mode>): New insn pattern.
19713 (*btr<mode>): Ditto.
19714 (*<btsc><mode>_mask): New insn_and_split pattern.
19715 (*btr<mode>_mask): Ditto.
19717 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19718 Alan Hayward <alan.hayward@arm.com>
19719 David Sherwood <david.sherwood@arm.com>
19721 * function.c (pad_below): Simplify padding calculation.
19723 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19724 Alan Hayward <alan.hayward@arm.com>
19725 David Sherwood <david.sherwood@arm.com>
19727 * target.def (function_prologue): Remove frame size argument.
19728 (function_epilogue): Likewise.
19729 * doc/tm.texi: Regenerate.
19730 * final.c (final_start_function): Update call to function_prologue.
19731 (final_end_function): Update call to function_epilogue.
19732 (default_function_pro_epilogue): Remove frame size argument.
19733 * output.h (default_function_pro_epilogue): Likewise.
19734 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19735 (arm_output_function_prologue): Likewise.
19736 * config/frv/frv.c (frv_function_prologue): Likewise.
19737 (frv_function_epilogue): Likewise.
19738 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19739 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19740 (ia64_output_function_epilogue): Likewise.
19741 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19742 (m32r_output_function_epilogue): Likewise.
19743 * config/microblaze/microblaze.c (microblaze_function_prologue)
19744 (microblaze_function_epilogue): Likewise.
19745 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19746 (mips_output_function_epilogue): Likewise.
19747 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19748 (mmix_target_asm_function_epilogue): Likewise.
19749 * config/msp430/msp430.c (msp430_start_function): Likewise.
19750 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19751 (nds32_asm_function_epilogue): Likewise.
19752 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19753 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19754 (pa_output_function_epilogue): Likewise.
19755 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19756 (rs6000_output_function_epilogue): Likewise.
19757 * config/rl78/rl78.c (rl78_start_function): Likewise.
19758 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19759 (rs6000_output_function_epilogue): Likewise.
19760 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19761 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19762 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19763 (sparc_asm_function_epilogue): Likewise.
19765 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19767 * tree.h (type_has_mode_precision_p): New function.
19768 * convert.c (convert_to_integer_1): Use it.
19769 * expr.c (expand_expr_real_2): Likewise.
19770 (expand_expr_real_1): Likewise.
19771 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19772 * match.pd: Likewise.
19773 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19774 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19775 * tree-tailcall.c (process_assignment): Likewise.
19776 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19777 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19778 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19779 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19780 (vectorizable_assignment): Likewise.
19781 (vectorizable_shift): Likewise.
19782 (vectorizable_operation): Likewise.
19783 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19785 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19787 * match.pd: Add pow (C, x) simplification.
19789 2017-08-21 Richard Biener <rguenther@suse.de>
19791 PR tree-optimization/81900
19792 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19793 for blocks with abnormal predecessors.
19794 (compute_antic): Do not set visited flag prematurely.
19796 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19799 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19801 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19803 * stor-layout.h (vector_type_mode): Move to...
19804 * tree.h (vector_type_mode): ...here.
19805 * stor-layout.c (vector_type_mode): Move to...
19806 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19808 2017-08-21 Richard Biener <rguenther@suse.de>
19810 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19811 register_external_die hooks.
19812 (debug_false_tree_charstarstar_uhwistar): Declare.
19813 (debug_nothing_tree_charstar_uhwi): Likewise.
19814 * debug.c (do_nothing_debug_hooks): Adjust.
19815 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19816 (debug_nothing_tree_charstar_uhwi): Likewise.
19817 * dbxout.c (dbx_debug_hooks): Adjust.
19818 (xcoff_debug_hooks): Likewise.
19819 * sdbout.c (sdb_debug_hooks): Likewise.
19820 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19821 * dwarf2out.c (macinfo_label_base): New global.
19822 (dwarf2out_register_external_die): New function for the
19823 register_external_die hook.
19824 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19825 (dwarf2_debug_hooks): Use them.
19826 (dwarf2_lineno_debug_hooks): Adjust.
19827 (struct die_struct): Add with_offset flag.
19828 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19829 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19830 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19831 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19832 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19833 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19834 defining section names for the early LTO debug variants.
19835 (reset_indirect_string): New helper.
19836 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19837 (print_dw_val): Add support for offsetted symbol references.
19838 (get_ultimate_context): Split out from is_cxx.
19839 (is_cxx): Use get_ultimate_context.
19840 (is_fortran): Add decl overload.
19841 (compute_comp_unit_symbol): Split out worker from
19842 compute_section_prefix.
19843 (compute_section_prefix): Call compute_comp_unit_symbol and
19844 set comdat_type_p here.
19845 (output_die): Skip DIE symbol output for the LTO added one.
19846 Handle DIE symbol references with offset.
19847 (output_comp_unit): Guard section name mangling properly.
19848 For LTO debug sections emit a symbol at the section beginning
19849 which we use to refer to its DIEs.
19850 (add_abstract_origin_attribute): For DIEs registered via
19851 dwarf2out_register_external_die directly refer to the early
19852 DIE rather than indirectly through the shadow one we created.
19853 Remove obsolete call to dwarf2out_abstract_function for
19854 non-function/block origins.
19855 (gen_array_type_die): When generating early LTO debug do
19856 not emit DW_AT_string_length.
19857 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19858 late when in LTO. As suggested place a gcc_unreachable for
19859 the DECL_ABSTRACT_P case.
19860 (gen_subprogram_die): Avoid another specification DIE
19861 for early built declarations/definitions for the late LTO case.
19862 (gen_variable_die): Add type references for late duplicated VLA dies
19864 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19865 we have the abstract instance already.
19866 (process_scope_var): Adjust decl DIE contexts in LTO which
19867 first puts them in limbo.
19868 (gen_decl_die): Do not generate type DIEs late apart from
19869 types for VLAs or for decls we do not yet have a DIE. Do not
19870 call dwarf2out_abstract_function late.
19871 (dwarf2out_early_global_decl): Make sure to create DIEs
19872 for abstract instances of a decl first.
19873 (dwarf2out_late_global_decl): Adjust comment.
19874 (output_macinfo_op): With multiple macro sections use
19875 macinfo_label_base to distinguish labels.
19876 (output_macinfo): Likewise. Update macinfo_label_base.
19877 Pass in the line info label.
19878 (note_variable_value_in_expr): When generating LTO resolve
19879 all variable values here by generating DIEs as needed.
19880 (init_sections_and_labels): Add early LTO debug flag parameter
19881 and generate different sections and names if set. Add generation
19882 counter for the labels so we can have multiple of them.
19883 (reset_dies): Helper to allow DIEs to be output multiple times.
19884 (dwarf2out_finish): When outputting DIEs to the fat part of an
19885 LTO object first reset DIEs.
19886 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19887 (modified_type_die): Check for decl_ultimate_origin being self
19889 (gen_type_die_with_usage): Likewise.
19890 (gen_typedef_die): Allow decl_ultimate_origin being self.
19891 (set_decl_abstract_flags): Remove.
19892 (set_block_abstract_flags): Likewise.
19893 (dwarf2out_abstract_function): Treat the early generated DIEs
19894 as the abstract copy and only add DW_AT_inline and
19895 DW_AT_artificial here and call set_decl_origin_self.
19896 If the DIE has an abstract origin don't do anything.
19897 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19898 if we have none yet (Go fails to build one, PR78628).
19899 (variably_modified_type_p): Prevent endless recursion for Ada
19900 cyclic pointer types.
19901 * lto-streamer-in.c: Include debug.h.
19902 (dref_queue): New global.
19903 (lto_read_tree_1): Stream in DIE references.
19904 (lto_input_tree): Register DIE references.
19905 (input_function): Stream DECL_DEBUG_ARGS.
19906 * lto-streamer-out.c: Include debug.h.
19907 (lto_write_tree_1): Output DIE references.
19908 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19909 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19910 (output_function): Stream DECL_DEBUG_ARGS.
19911 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19912 Stream DECL_ABSTRACT_ORIGIN.
19913 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19914 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19915 DECL_CONTEXT for file-scope decls.
19916 * lto-streamer.h (struct dref_entry): Declare.
19917 (dref_queue): Likewise.
19918 * cfgexpand.c (pass_expand::execute): Do not call the
19919 outlining_inline_function hook here.
19920 * lto-wrapper.c (debug_obj): New global.
19921 (tool_cleanup): Unlink it if required.
19922 (debug_objcopy): New function.
19923 (run_gcc): Handle early debug sections in the IL files by
19924 extracting them to separate files, partially linkin them and
19925 feeding the result back as result to the linker.
19926 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19927 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19928 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19929 sections into a separate segment.
19930 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19932 (darwin_asm_dwarf_section): Likewise.
19933 (darwin_asm_output_dwarf_offset): Likewise.
19934 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19937 Alan Hayward <alan.hayward@arm.com>
19938 David Sherwood <david.sherwood@arm.com>
19940 * read-md.h (md_reader::record_potential_iterator_use): Replace
19941 pointer argument with an rtx and an index.
19942 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19943 (apply_mode_iterator): Likewise.
19944 (apply_code_iterator): Likewise.
19945 (apply_int_iterator): Likewise.
19946 (apply_subst_iterator): Likewise.
19947 (record_iterator_use): Likewise.
19948 (record_attribute_use): Likewise.
19949 (md_reader::record_potential_iterator_use): Likewise. Update calls
19950 to record_iterator_use and apply_iterator.
19951 (iterator_use): Replace ptr with x and index.
19952 (attribute_use): Likewise.
19953 (apply_attribute_uses): Update calls to apply_iterator.
19954 (apply_iterators): Likewise. Update initialization of iterator_use.
19955 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19956 and record_potential_iterator_use.
19957 (rtx_reader::read_rtx_operand): Likewise.
19959 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19960 Alan Hayward <alan.hayward@arm.com>
19961 David Sherwood <david.sherwood@arm.com>
19963 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19966 2017-08-21 Richard Biener <rguenther@suse.de>
19968 PR middle-end/81884
19969 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19970 at struct end conservatively when comparing common bases.
19972 2017-08-21 Richard Biener <rguenther@suse.de>
19974 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19975 (mem_ref_in_stmt): Remove.
19976 (determine_max_movement): Use ref index to get at the reference.
19977 (invariantness_dom_walker::before_dom_children): Deal with
19978 lim data already initialized.
19979 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19981 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19983 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19984 (TARGET_ISA_ROUND): Ditto.
19985 (TARGET_ROUND): Ditto.
19986 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19987 * config/i386/i386.md: Ditto.
19988 * config/i386/sse.md: Ditto.
19989 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19990 with OPTION_MASK_ISA_SSE4_1.
19992 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19995 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19996 __builtin_ia32_lzcnt_u16.
19998 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20001 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20002 (rs6000_set_current_function): Rewrite function to use it.
20004 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20007 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20008 and TYPE_WARN_IF_NOT_ALIGN.
20009 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20010 (handle_warn_if_not_align): New.
20011 (place_union_field): Call handle_warn_if_not_align.
20012 (place_field): Call handle_warn_if_not_align.
20013 Copy TYPE_WARN_IF_NOT_ALIGN.
20014 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20015 (layout_type): Likewise.
20016 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20018 (tree_decl_common): Add warn_if_not_align.
20019 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20020 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20021 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20022 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20023 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20024 * doc/extend.texi: Document warn_if_not_aligned attribute.
20025 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20027 2017-08-17 Martin Liska <mliska@suse.cz>
20030 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20031 (get_data_dependence): Use it as pointer type.
20032 (distribute_loop): Likewise.
20034 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20036 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20037 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20038 (p8_vmrgow_<mode>_direct): New define_insn.
20039 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20040 handle endianness for vmrgew and vmrgow permute patterns.
20042 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20044 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20045 * config/rs6000/rs6000-cpus.def: Remove comment.
20046 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20047 (POWERPC_MASKS): Likewise.
20048 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20049 use of TARGET_VSX_TIMODE.
20050 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20051 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20052 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20053 (rs6000_option_override_internal): Remove dead code.
20054 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20055 (rs6000_legitimize_reload_address): Likewise.
20056 (rs6000_legitimate_address_p): Likewise.
20057 (rs6000_opt_masks): Delete "vsx-timode".
20058 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20059 from function comment.
20060 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20061 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20062 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20064 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20065 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20066 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20067 (VSX_TI): Likewise.
20069 (define_peephole2): Likewise.
20071 2017-08-17 Martin Sebor <msebor@redhat.com>
20074 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20075 past the end of an array.
20076 (test_pp_format): Add test cases.
20078 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20080 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20081 missing ECF_NOTHROW flags.
20083 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20086 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20087 operands residing in integer registers.
20088 (*vsx_le_perm_load_<mode>): Likewise.
20089 (*vsx_le_perm_store_<mode>): Likewise.
20090 (define_peephole2): Add peepholes to optimize the above.
20092 2017-08-17 Marek Polacek <polacek@redhat.com>
20094 PR middle-end/81814
20095 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20096 to mimic what shorten_compare did. Change the return type to bool.
20097 (fold_cond_expr_with_comparison): Update call to
20098 operand_equal_for_comparison_p.
20099 (fold_ternary_loc): Likewise.
20101 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20103 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20105 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20107 2017-08-17 Richard Biener <rguenther@suse.de>
20109 * tree-ssa-structalias.c (solve_graph): When propagating
20110 to successors update the graphs succ edges and avoid duplicate work.
20112 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20115 * config/i386/i386.c (ix86_option_override_internal): Save target
20116 specific options after ix86_stack_protector_guard_reg was changed.
20118 2017-08-17 Richard Biener <rguenther@suse.de>
20120 PR tree-optimization/81827
20121 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20122 (new_var_info): Initialize it conservatively.
20123 (get_call_vi): Mark register vars.
20124 (new_scalar_tmp_constraint_exp): Likewise.
20125 (handle_rhs_call): Likewise.
20126 (handle_const_call): Likewise.
20127 (create_function_info_for): Likewise.
20128 (solve_constraints): Sort varinfos to separate register from
20129 non-register vars to pack points-to solution bitmaps during
20132 2017-08-17 Marek Polacek <polacek@redhat.com>
20134 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20136 2017-08-17 Richard Biener <rguenther@suse.de>
20138 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20139 to true when overflow is undefined and we saturated the result.
20141 2017-08-17 Alan Modra <amodra@gmail.com>
20144 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20145 Don't use store multiple if only one reg needs saving.
20146 (interesting_frame_related_regno): New function.
20147 (rs6000_frame_related): Don't emit frame info for regs that
20149 (rs6000_emit_epilogue): Likewise.
20151 2017-08-16 Nathan Sidwell <nathan@acm.org>
20153 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20154 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20155 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20156 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20157 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20158 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20159 (verify_type): Adjust for TYPE_BINFO move.
20160 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20161 process TYPE_BINFO directly.
20162 (hash_tree): Likewise.
20163 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20165 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20168 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20170 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20172 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20175 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20176 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20177 Add ix86_binary_operator_ok to insn constraint.
20178 (*iordi_1_bts): Ditto.
20179 (*xordi_1_btc): Ditto.
20180 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20181 Update corresponding peephole2 pattern.
20185 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20187 PR tree-optimization/81832
20188 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20189 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20191 2017-08-16 Marek Polacek <polacek@redhat.com>
20194 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20195 regarding size_zero_node.
20197 2017-08-16 Martin Liska <mliska@suse.cz>
20200 * config.gcc: Respect previously set extra_objs in case
20203 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20205 PR tree-optimization/81835
20206 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20207 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20208 not depend on the phi.
20210 2017-08-16 Alan Modra <amodra@gmail.com>
20212 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20215 2017-08-16 Alan Modra <amodra@gmail.com>
20217 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20218 (save_reg_p): ..into this. Update all callers.
20219 (first_reg_to_save): Simplify.
20221 2017-08-16 Alan Modra <amodra@gmail.com>
20223 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20226 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20230 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20231 constant count if that count is less than 32.
20233 2017-08-15 Nathan Sidwell <nathan@acm.org>
20235 * gcc.c (execute): Emit friendlier message if inferior is killed
20236 by an external cause.
20238 2017-08-15 Richard Biener <rguenther@suse.de>
20240 PR tree-optimization/81790
20241 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20242 CONSTRUCTORs from simplifying and VN.
20244 2017-08-14 Martin Sebor <msebor@redhat.com>
20246 * builtin-attrs.def: Add comments.
20248 2017-08-14 Martin Sebor <msebor@redhat.com>
20251 * doc/extend.texi (attribute nonstring): Document new attribute.
20253 2017-08-14 Martin Sebor <msebor@redhat.com>
20256 * tree-diagnostic.c (default_tree_printer): Handle %G.
20257 * gimple-pretty-print.h (percent_G_format): Declare new function.
20258 * gimple-pretty-print.c (percent_G_format): Define.
20259 * tree-pretty-print.c (percent_K_format): Add argument.
20261 2017-08-14 Martin Sebor <msebor@redhat.com>
20263 PR translation/79998
20264 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20265 Remove a stray space.
20267 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20270 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20271 (*iordi_1_bts): Ditto.
20272 (*xordi_1_btc): Ditto.
20274 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20277 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20278 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20280 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20281 quoted strings, and make more translator-friendly.
20282 (darwin_rs6000_override_options): Likewise.
20283 (rs6000_option_override_internal): Likewise.
20284 (rs6000_return_in_memory): Fix overlong line.
20285 (init_cmulative_args): Use quoted strings, and make more
20286 translator-friendly.
20287 (rs6000_pass_by_reference): Fix overlong line.
20288 (def_builtin): Use quoted strings.
20289 (altivec_expand_predicate_builtin): Use quoted strings, and make
20290 more translator-friendly.
20291 (htm_expand_builtin): Use quoted strings.
20292 (cpu_expand_builtin): Use quoted strings, and make more
20293 translator-friendly.
20294 (altivec_expand_builtin): Likewise.
20295 (paired_expand_predicate_builtin): Likewise.
20296 (rs6000_invalid_builtin): Likewise.
20297 (builtin_function_type): Use quoted strings.
20298 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20299 more translator-friendly.
20300 (rs6000_trampoline_init): Likewise.
20301 (rs6000_handle_altivec_attribute): Likewise.
20302 (rs6000_inner_target_options): Use quoted strings.
20303 (rs6000_disable_incompatible_switches): Likewise.
20304 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20305 strings, and make more translator-friendly.
20306 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20308 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20310 PR tree-optimization/81799
20311 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20312 cond_expr to simple gimple operand.
20314 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20316 PR middle-end/46932
20317 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20319 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20323 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20324 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20325 TARGET_GASISR_PROLOGUES.
20326 * config/avr/avr.c (avr_option_override): Same.
20327 (avr_pass_pre_proep::execute): Same.
20329 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20332 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20333 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20335 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20337 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20338 define_insn_and_split. Split to a memory load from 0 in
20339 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20340 using PTR mode iterator.
20341 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20342 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20343 (*add_tp_<mode>): Redefine as define_insn_and_split.
20344 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20345 address space. Merge with *add_tp_x32 using PTR mode iterator.
20346 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20347 Split to an add with a memory load from 0 in
20348 DEFAULT_TLS_SEG_REG address space.
20350 2017-08-12 Andrew Pinski <apinski@cavium.com>
20352 * config/aarch64/aarch64-option-extensions.def (rdma):
20353 Fix feature string to what Linux prints out in /proc/cpuinfo.
20355 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20358 * dwarf2out.c (modified_type_die): For C typedef types that have
20359 an ultimate origin, process the ultimate origin instead of the
20361 (gen_typedef_die): Assert that input DECLs have no ultimate
20363 (gen_type_die_with_usage): For typedef variants that have an
20364 ultimate origin, just call gen_decl_die on the original DECL.
20365 (process_scope_var): Avoid creating DIEs for local typedefs and
20366 concrete static variables.
20368 2017-08-12 Alan Modra <amodra@gmail.com>
20372 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20373 match gnu-user.h startfile.
20374 (ENDFILE_LINUX_SPEC): Similarly.
20376 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20379 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20381 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20383 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20384 * config/aarch64/aarch64.md (mov<mode>): Change.
20385 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20386 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20387 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20389 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20391 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20392 for storage order barriers.
20394 2017-08-11 Martin Liska <mliska@suse.cz>
20397 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20398 variables of void type.
20400 2017-08-11 Martin Liska <mliska@suse.cz>
20402 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20403 TARGET_SUPPORTS_ALIASES.
20404 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20405 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20406 (optimize_weakref): Likewise.
20407 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20408 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20409 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20411 2017-08-11 Martin Liska <mliska@suse.cz>
20414 * config/i386/i386.c (make_resolver_func): Do complete
20415 refactoring of the function.
20417 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20420 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20421 * config/i386/i386.c (ix86_stack_protect_guard): Use
20422 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20423 * doc/invoke.texi (x86 Options): Document
20424 -mstack-protector-guard-symbol= option.
20426 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20428 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20429 * config/i386/i386.c (ix86_split_stack_guard): New function.
20430 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20431 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20432 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20433 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20434 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20435 (split_stack_space_check): Call ix86_split_stack_guard.
20437 2017-08-10 Martin Sebor <msebor@redhat.com>
20439 * print-tree.c (print_node): Print location using the established
20441 Replace spaces with colons.
20442 (debug_raw, debug): Ditto.
20444 2017-08-10 Martin Sebor <msebor@redhat.com>
20447 * pretty-print.c (pp_format): Correct the handling of %s precision.
20449 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20452 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20454 (ix86_finalize_stack_frame_flags): This. Also clear
20455 frame_pointer_needed if -fno-omit-frame-pointer is used without
20457 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20458 with ix86_finalize_stack_frame_flags.
20459 (ix86_expand_epilogue): Likewise.
20460 (ix86_expand_split_stack_prologue): Likewise.
20461 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20463 2017-08-10 Martin Liska <mliska@suse.cz>
20466 * c-attribs.c (handle_target_attribute):
20467 Report warning for an empty string argument of target attribute.
20469 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20472 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20474 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20475 or DECL_NONLOCAL labels.
20476 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20477 or DECL_NONLOCAL labels here.
20479 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20481 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20482 to indicate when early gimple folding has been disabled.
20483 (rs6000_gimple_fold_builtin): Add debug content.
20484 (rs6000_invalid_builtin): Fix whitespace.
20485 (rs6000_expand_builtin): Fix whitespace.
20486 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20488 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20491 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20492 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20494 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20496 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20497 (qdf24xx): Likewise.
20498 * config/aarch64/aarch64.md: Include falkor.md.
20499 * config/aarch64/falkor.md: New.
20501 2017-08-09 Marek Polacek <polacek@redhat.com>
20504 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20505 * diagnostic.c (emit_diagnostic): Add a comment.
20506 (emit_diagnostic_valist): New function.
20508 2017-08-09 Marek Polacek <polacek@redhat.com>
20511 * input.c (make_location): New overload.
20512 * input.h (make_location): Declare.
20514 2017-08-08 Alan Modra <amodra@gmail.com>
20515 H.J. Lu <hongjiu.lu@intel.com>
20518 * gcc.c (NO_PIE_SPEC): Delete.
20519 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20521 (LINK_PIE_SPEC): ..to here.
20522 (LINK_COMMAND_SPEC): Support -no-pie.
20523 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20524 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20526 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20527 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20528 (ENDFILE_CRTEND_SPEC): Similarly.
20530 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20533 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20534 (mstack-protector-guard-offset=): Ditto.
20535 * config/i386/i386.c (ix86_option_override): Handle
20536 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20538 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20539 ix86_stack_protect_guard_offset variables.
20540 (TARGET_STACK_PROTECT_GUARD): Always define.
20541 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20542 and -mstack-protector-guard-offset= options.
20544 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20546 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20547 boundary case for the last candidate.
20549 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20551 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20552 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20554 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20556 PR middle-end/19706
20557 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20558 * config/aarch64/aarch64-builtins.c
20559 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20560 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20561 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20563 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20564 Andrew Pinski <pinskia@gmail.com>
20566 PR middle-end/19706
20567 * internal-fn.def (XORSIGN): New.
20568 * optabs.def (xorsign_optab): New.
20569 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20570 (convert_expand_mult_copysign): New.
20571 (pass_optimize_widening_mul::execute): Call
20572 convert_expand_mult_copysign.
20574 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20576 PR tree-optimization/81354
20577 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20578 Insert on edges rather than explicitly creating landing pads.
20579 (analyze_candidates_and_replace): Commit edge inserts.
20581 2017-08-08 Richard Biener <rguenther@suse.de>
20583 PR middle-end/81719
20584 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20585 (expand_simple_operations): Also look through ADDR_EXPRs with
20586 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20588 2017-08-08 Richard Biener <rguenther@suse.de>
20590 PR tree-optimization/81723
20591 * tree-vect-slp.c (struct bst_traits): New hash traits.
20592 (bst_fail): New global.
20593 (vect_build_slp_tree_2): New worker, split out from ...
20594 (vect_build_slp_tree): ... this now wrapping it with using
20595 bst_fail set to cache SLP tree build fails. Properly handle
20597 (vect_analyze_slp_instance): Allocate and free bst_fail.
20599 2017-08-08 Martin Liska <mliska@suse.cz>
20602 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20603 LABEL_DECLs that can be from a different function.
20605 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20607 PR tree-optimization/81744
20608 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20609 loop's number of iterations.
20611 2017-08-08 Martin Liska <mliska@suse.cz>
20613 * asan.c: Include header files.
20614 * attribs.c (build_decl_attribute_variant): New function moved
20616 (build_type_attribute_qual_variant): Likewise.
20617 (cmp_attrib_identifiers): Likewise.
20618 (simple_cst_list_equal): Likewise.
20619 (omp_declare_simd_clauses_equal): Likewise.
20620 (attribute_value_equal): Likewise.
20621 (comp_type_attributes): Likewise.
20622 (build_type_attribute_variant): Likewise.
20623 (lookup_ident_attribute): Likewise.
20624 (remove_attribute): Likewise.
20625 (merge_attributes): Likewise.
20626 (merge_type_attributes): Likewise.
20627 (merge_decl_attributes): Likewise.
20628 (merge_dllimport_decl_attributes): Likewise.
20629 (handle_dll_attribute): Likewise.
20630 (attribute_list_equal): Likewise.
20631 (attribute_list_contained): Likewise.
20632 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20633 (lookup_attribute_by_prefix): Likewise.
20634 * bb-reorder.c: Include header files.
20635 * builtins.c: Likewise.
20636 * calls.c: Likewise.
20637 * cfgexpand.c: Likewise.
20638 * cgraph.c: Likewise.
20639 * cgraphunit.c: Likewise.
20640 * convert.c: Likewise.
20641 * dwarf2out.c: Likewise.
20642 * final.c: Likewise.
20643 * fold-const.c: Likewise.
20644 * function.c: Likewise.
20645 * gimple-expr.c: Likewise.
20646 * gimple-fold.c: Likewise.
20647 * gimple-pretty-print.c: Likewise.
20648 * gimple.c: Likewise.
20649 * gimplify.c: Likewise.
20650 * hsa-common.c: Likewise.
20651 * hsa-gen.c: Likewise.
20652 * internal-fn.c: Likewise.
20653 * ipa-chkp.c: Likewise.
20654 * ipa-cp.c: Likewise.
20655 * ipa-devirt.c: Likewise.
20656 * ipa-fnsummary.c: Likewise.
20657 * ipa-inline.c: Likewise.
20658 * ipa-visibility.c: Likewise.
20660 * lto-cgraph.c: Likewise.
20661 * omp-expand.c: Likewise.
20662 * omp-general.c: Likewise.
20663 * omp-low.c: Likewise.
20664 * omp-offload.c: Likewise.
20665 * omp-simd-clone.c: Likewise.
20666 * opts-global.c: Likewise.
20667 * passes.c: Likewise.
20668 * predict.c: Likewise.
20669 * sancov.c: Likewise.
20670 * sanopt.c: Likewise.
20671 * symtab.c: Likewise.
20672 * toplev.c: Likewise.
20673 * trans-mem.c: Likewise.
20674 * tree-chkp.c: Likewise.
20675 * tree-eh.c: Likewise.
20676 * tree-into-ssa.c: Likewise.
20677 * tree-object-size.c: Likewise.
20678 * tree-parloops.c: Likewise.
20679 * tree-profile.c: Likewise.
20680 * tree-ssa-ccp.c: Likewise.
20681 * tree-ssa-live.c: Likewise.
20682 * tree-ssa-loop.c: Likewise.
20683 * tree-ssa-sccvn.c: Likewise.
20684 * tree-ssa-structalias.c: Likewise.
20685 * tree-ssa.c: Likewise.
20686 * tree-streamer-in.c: Likewise.
20687 * tree-vectorizer.c: Likewise.
20688 * tree-vrp.c: Likewise.
20689 * tsan.c: Likewise.
20690 * ubsan.c: Likewise.
20691 * varasm.c: Likewise.
20692 * varpool.c: Likewise.
20693 * tree.c: Remove functions moved to attribs.[ch].
20694 * tree.h: Likewise.
20695 * config/aarch64/aarch64.c: Add attrs.h header file.
20696 * config/alpha/alpha.c: Likewise.
20697 * config/arc/arc.c: Likewise.
20698 * config/arm/arm.c: Likewise.
20699 * config/avr/avr.c: Likewise.
20700 * config/bfin/bfin.c: Likewise.
20701 * config/c6x/c6x.c: Likewise.
20702 * config/cr16/cr16.c: Likewise.
20703 * config/cris/cris.c: Likewise.
20704 * config/darwin.c: Likewise.
20705 * config/epiphany/epiphany.c: Likewise.
20706 * config/fr30/fr30.c: Likewise.
20707 * config/frv/frv.c: Likewise.
20708 * config/ft32/ft32.c: Likewise.
20709 * config/h8300/h8300.c: Likewise.
20710 * config/i386/winnt.c: Likewise.
20711 * config/ia64/ia64.c: Likewise.
20712 * config/iq2000/iq2000.c: Likewise.
20713 * config/lm32/lm32.c: Likewise.
20714 * config/m32c/m32c.c: Likewise.
20715 * config/m32r/m32r.c: Likewise.
20716 * config/m68k/m68k.c: Likewise.
20717 * config/mcore/mcore.c: Likewise.
20718 * config/microblaze/microblaze.c: Likewise.
20719 * config/mips/mips.c: Likewise.
20720 * config/mmix/mmix.c: Likewise.
20721 * config/mn10300/mn10300.c: Likewise.
20722 * config/moxie/moxie.c: Likewise.
20723 * config/msp430/msp430.c: Likewise.
20724 * config/nds32/nds32-isr.c: Likewise.
20725 * config/nds32/nds32.c: Likewise.
20726 * config/nios2/nios2.c: Likewise.
20727 * config/nvptx/nvptx.c: Likewise.
20728 * config/pa/pa.c: Likewise.
20729 * config/pdp11/pdp11.c: Likewise.
20730 * config/powerpcspe/powerpcspe.c: Likewise.
20731 * config/riscv/riscv.c: Likewise.
20732 * config/rl78/rl78.c: Likewise.
20733 * config/rx/rx.c: Likewise.
20734 * config/s390/s390.c: Likewise.
20735 * config/sh/sh.c: Likewise.
20736 * config/sol2.c: Likewise.
20737 * config/sparc/sparc.c: Likewise.
20738 * config/spu/spu.c: Likewise.
20739 * config/stormy16/stormy16.c: Likewise.
20740 * config/tilegx/tilegx.c: Likewise.
20741 * config/tilepro/tilepro.c: Likewise.
20742 * config/v850/v850.c: Likewise.
20743 * config/vax/vax.c: Likewise.
20744 * config/visium/visium.c: Likewise.
20745 * config/xtensa/xtensa.c: Likewise.
20747 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20750 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20751 constraints since the -mupper-regs-* switches have been
20753 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20754 into a vector from a double word element that was extracted from
20755 another vector, and eliminate extra XXPERMDI instructions.
20756 (vsx_concat_<mode>_2): Likewise.
20757 (vsx_concat_<mode>_3): Likewise.
20758 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20759 concat to allow optimizing inserts from previous extracts.
20761 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20763 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20764 memory reference to a SSP offset in TLS address space.
20765 (ix86_print_operand) <case '@'>: Remove.
20766 (ix86_print_operand_punct_valid_p): Remove '@' code.
20767 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20768 UNSPEC_SP_TLS_TEST.
20769 (stack_tls_protect_set_<mode>): Remove.
20770 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20771 (stack_tls_protect_test_<mode>): Remove.
20772 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20774 2017-08-07 Olivier Hainque <hainque@adacore.com>
20777 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20779 2017-08-07 Douglas Rupp <rupp@adacore.com>
20781 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20782 variable was referenced as multidir in command.
20784 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20787 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20790 2017-08-07 Martin Liska <mliska@suse.cz>
20792 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20793 * config/rl78/rl78.c: Add include of attribs.h.
20794 * config/sh/sh.c: Likewise.
20795 * config/v850/v850.c: Likewise.
20797 2017-08-07 Tom de Vries <tom@codesourcery.com>
20799 PR middle-end/78266
20800 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20802 2017-08-07 Martin Liska <mliska@suse.cz>
20804 * config/mips/mips.c: Include attribs.h.
20806 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20809 * doc/invoke.texi: Document change in behvaior for -Ofast for
20812 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20814 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20816 (aarch64_pop_regs): Likewise.
20817 (aarch64_gen_load_pair): Likewise.
20818 (aarch64_save_callee_saves): Likewise.
20819 (aarch64_restore_callee_saves): Likewise.
20821 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20823 * config/i386/i386.c: Revert the last change.
20825 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20828 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20830 (ix86_finalize_stack_frame_flags): This. Also clear
20831 frame_pointer_needed if -fno-omit-frame-pointer is used without
20833 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20834 with ix86_finalize_stack_frame_flags.
20835 (ix86_expand_epilogue): Likewise.
20836 (ix86_expand_split_stack_prologue): Likewise.
20838 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20841 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20842 to P_AES for Westmere.
20844 2017-08-07 Jonathan Yong <10walls@gmail.com>
20846 * config/i386/mingw.opt (fset-stack-executable): Removed.
20847 * config/i386/cygming.opt (fset-stack-executable): Moved
20849 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20851 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20853 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20855 2017-08-07 Marek Polacek <polacek@redhat.com>
20857 PR middle-end/81737
20858 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20860 2017-08-07 Martin Liska <mliska@suse.cz>
20862 * attribs.h (canonicalize_attr_name): New function.
20863 (cmp_attribs): Move from c-format.c and adjusted.
20864 (is_attribute_p): Moved from tree.h.
20865 * tree-inline.c: Add new includes.
20866 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20867 (private_is_attribute_p): Remove.
20868 (private_lookup_attribute): Likewise.
20869 (private_lookup_attribute_by_prefix): Simplify.
20870 (remove_attribute): Use is_attribute_p.
20871 * tree.h: Remove removed declarations.
20873 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20875 PR middle-end/81698
20876 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20877 instead of computing it in the function. Formatting fix.
20878 (expand_case): Don't rely on default_edge being the first edge,
20879 clear it if removing it, pass default_edge to
20880 emit_case_dispatch_table.
20881 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20884 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20886 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20887 insn in the function, emit NOP after the insn.
20889 2017-08-06 Tom de Vries <tom@codesourcery.com>
20891 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20894 2017-08-06 Tom de Vries <tom@codesourcery.com>
20896 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20899 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20901 PR tree-optimization/57371
20902 * match.pd: New pattern.
20904 2017-08-04 Marek Polacek <polacek@redhat.com>
20906 PR middle-end/81695
20907 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20908 perform the computation in offset_int.
20910 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20912 PR tree-optimization/81136
20913 * tree-vectorizer.h: Include tree-hash-traits.h.
20914 (vec_base_alignments): New typedef.
20915 (vec_info): Add a base_alignments field.
20916 (vect_record_base_alignments): Declare.
20917 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20919 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20920 (create_data_ref): Add an is_conditional_in_stmt argument.
20921 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20922 the is_conditional_in_stmt field.
20923 (data_ref_loc): Add an is_conditional_in_stmt field.
20924 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20925 (find_data_references_in_stmt): Update call to create_data_ref.
20926 (graphite_find_data_references_in_stmt): Likewise.
20927 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20928 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20929 (vect_record_base_alignment): New function.
20930 (vect_record_base_alignments): Likewise.
20931 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20932 for nested statements even if we fail to compute a misalignment.
20933 Use pooled base alignments for unconditional references.
20934 (vect_find_same_alignment_drs): Compare base addresses instead
20936 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20937 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20939 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20941 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20942 Add an explicit name for the enum. Use auto_vec for slp_instances
20943 and grouped_stores.
20944 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20946 (_bb_vec_info): Add a constructor and destructor.
20947 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20948 (destroy_loop_vec_info): Delete.
20949 (vect_destroy_datarefs): Likewise.
20950 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20951 (vec_info::vec_info): New function.
20952 (vec_info::~vec_info): Likewise.
20953 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20954 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20955 destroy_loop_vec_info.
20956 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20957 (_loop_vec_info::_loop_vec_info): ...this.
20958 (destroy_loop_vec_info): Replace with...
20959 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20960 the stmt_vec_infos. Leave handling of vec_info information to its
20961 destructor. Remove explicit vector releases.
20962 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20963 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20964 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20965 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20966 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20967 (destroy_bb_vec_info): Replace with...
20968 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20969 information to its destructor.
20970 (vect_slp_analyze_bb_1): Use new and delete instead of
20971 new_bb_vec_info and destroy_bb_vec_info.
20972 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20975 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20977 * tree-data-ref.h (subscript): Add access_fn field.
20978 (data_dependence_relation): Add could_be_independent_p.
20979 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20980 (same_access_functions): Move to tree-data-ref.c.
20981 * tree-data-ref.c (ref_contains_union_access_p): New function.
20982 (access_fn_component_p): Likewise.
20983 (access_fn_components_comparable_p): Likewise.
20984 (dr_analyze_indices): Add a reference to access_fn_component_p.
20985 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20987 (constant_access_functions): Likewise.
20988 (add_other_self_distances): Likewise.
20989 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20990 (initialize_data_dependence_relation): Use XCNEW and remove
20991 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20992 of access functions that have the same type. Allow the
20993 subsequence to end with different bases in some circumstances.
20994 Record the chosen access functions in SUB_ACCESS_FN.
20995 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20996 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20997 (subscript_dependence_tester_1): Likewise dra and drb.
20998 (build_classic_dist_vector): Update calls accordingly.
20999 (subscript_dependence_tester): Likewise.
21000 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21001 DDR_COULD_BE_INDEPENDENT_P.
21002 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21003 comp_alias_ddrs instead of may_alias_ddrs.
21004 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21006 (vect_analyze_data_ref_dependence): Use it if
21007 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21008 distance vectors if that fails.
21009 (dependence_distance_ge_vf): New function.
21010 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21011 LOOP_VINFO_MAY_ALIAS_DDRS.
21013 2017-08-04 Richard Biener <rguenther@suse.de>
21015 PR middle-end/81705
21016 * fold-const.c (fold_binary_loc): Properly restrict
21017 minus_var0 && minus_var1 case when associating undefined overflow
21020 2017-08-04 Tom de Vries <tom@codesourcery.com>
21022 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21024 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21026 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21027 Don't start diagnostic messages with a capital letter.
21028 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21030 (rs6000_invalid_builtin): Likewise.
21031 (rs6000_trampoline_init): Likewise.
21033 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21036 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21037 after setting changeable df flags.
21039 2017-08-03 Richard Biener <rguenther@suse.de>
21041 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21042 up if the use is in USE - X.
21044 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21046 * toplev.c (dumpfile.h): New include.
21047 (internal_error_reentered): New static function. Use it...
21048 (internal_error_function): ...here to handle reentered internal_error.
21050 2017-08-03 Richard Biener <rguenther@suse.de>
21052 PR middle-end/81148
21053 * fold-const.c (split_tree): Add minus_var and minus_con
21054 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21055 here but always use minus_*.
21056 (associate_trees): Assert we never associate with MINUS_EXPR
21057 and NULL first operand. Do not recurse for PLUS_EXPR operands
21058 when associating as MINUS_EXPR either.
21059 (fold_binary_loc): Track minus_var and minus_con.
21061 2017-08-03 Tom de Vries <tom@codesourcery.com>
21064 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21065 ACCEL_COMPILER, apply finish_options on
21066 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21068 2017-08-03 Tom de Vries <tom@codesourcery.com>
21071 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21072 function_entry_patch_area_size > 0.
21074 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21077 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21078 instead of 10??LU, perform unit multiplication in wide_int,
21079 don't change alloc_object_size_limit if the limit is larger
21082 PR tree-optimization/81655
21083 PR tree-optimization/81588
21084 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21085 the case when ranges[i].low and high are 1 for unsigned type with
21088 PR middle-end/81052
21089 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21090 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21092 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21094 * tree-vrp.h: Add include guard.
21096 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21099 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21100 (ud2): New insn pattern.
21101 * config/i386/i386.c (ix86_expand_epilogue):
21102 For naked functions, generate ud2 instead of trap insn.
21104 2017-08-02 Marek Polacek <polacek@redhat.com>
21107 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21109 2017-08-02 Tom de Vries <tom@codesourcery.com>
21110 Cesar Philippidis <cesar@codesourcery.com>
21112 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21113 Add missing edge probabilities.
21115 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21117 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21118 Correct endianness.
21120 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21122 PR middle-end/79499
21123 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21124 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21125 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21126 prologue_seq sequences - if any.
21128 2017-08-02 Richard Biener <rguenther@suse.de>
21130 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21131 via vectors if supported, integer extracts via punning if supported
21132 or otherwise vector extracts.
21134 2017-08-02 Richard Biener <rguenther@suse.de>
21136 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21138 (bitmap_insert_into_set): ... this.
21140 2017-08-02 Richard Biener <rguenther@suse.de>
21142 PR tree-optimization/81633
21144 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21146 PR tree-optimization/71752
21147 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21149 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21151 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21152 (machine_function::call_ms2sysv_pad_out): Remove field.
21153 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21154 (ix86_compute_frame_layout): Likewise.
21156 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21159 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21160 attribute with interrupt attribute.
21162 2017-08-01 Andrew Pinski <apinski@cavium.com>
21164 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21165 BIT_INSERT_EXPR's operand 1
21166 to see if the types precision matches.
21168 2017-08-01 Martin Liska <mliska@suse.cz>
21170 PR middle-end/70140
21171 * builtins.c (expand_builtin_memcpy_args): Remove.
21172 (expand_builtin_memcpy): Call newly added function
21173 expand_builtin_memory_copy_args.
21174 (expand_builtin_memcpy_with_bounds): Likewise.
21175 (expand_builtin_mempcpy): Remove last argument.
21176 (expand_builtin_mempcpy_with_bounds): Likewise.
21177 (expand_builtin_memory_copy_args): New function created from
21178 expand_builtin_mempcpy_args with small modifications.
21179 (expand_builtin_mempcpy_args): Remove.
21180 (expand_builtin_stpcpy): Remove unused argument.
21181 (expand_builtin): Likewise.
21182 (expand_builtin_with_bounds): Likewise.
21184 2017-08-01 Martin Liska <mliska@suse.cz>
21187 Make mempcpy more optimal (PR middle-end/70140).
21189 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21192 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21193 __builtin_vec_cmpne verify both arguments are compatible vectors
21194 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21195 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21196 move computation of aligned to after checking the argument types.
21200 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21201 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21204 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21205 Doug Rupp <rupp@adacore.com>
21206 Olivier Hainque <hainque@adacore.com>
21208 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21209 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21211 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21212 for TARGET_OS_CPP_BUILTIN.
21213 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21214 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21216 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21217 passing required abi options to the assembler for EABI configurations.
21218 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21219 of .text.hot and .text.unlikely sections for kernel modules when
21220 using ARM style exceptions.
21221 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21222 options. Add EXTRA_CC1_SPEC.
21223 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21225 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21227 (ARM_TARGET2_DWARF_FORMAT): Define.
21228 * config/arm/t-vxworks: Adjust multilib control to removal of the
21229 Diab command line options.
21231 2017-08-01 Martin Liska <mliska@suse.cz>
21233 PR gcov-profile/81561
21234 * gcov.c (unblock): Make unblocking safe as we need to preserve
21235 index correspondence of blocks and block_lists.
21237 2017-08-01 Richard Biener <rguenther@suse.de>
21239 PR tree-optimization/81181
21240 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21241 (compute_antic): ... end of iteration here.
21243 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21245 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21246 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21247 (ftree-slp-vectorize): Likewise.
21248 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21249 can no longer be set independent of flag_tree_loop_vectorize.
21250 * omp-general.c (emp_max_vf): Likewise.
21251 * opts.c (enable_fdo_optimizations): Remove references to
21252 flag_tree_vectorize, these are now implicit.
21253 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21254 and leave it for the options machinery.
21256 2017-08-01 Martin Liska <mliska@suse.cz>
21258 PR middle-end/70140
21259 * builtins.c (expand_builtin_memcpy_args): Remove.
21260 (expand_builtin_memcpy): Call newly added function
21261 expand_builtin_memory_copy_args.
21262 (expand_builtin_memcpy_with_bounds): Likewise.
21263 (expand_builtin_mempcpy): Remove last argument.
21264 (expand_builtin_mempcpy_with_bounds): Likewise.
21265 (expand_builtin_memory_copy_args): New function created from
21266 expand_builtin_mempcpy_args with small modifications.
21267 (expand_builtin_mempcpy_args): Remove.
21268 (expand_builtin_stpcpy): Remove unused argument.
21269 (expand_builtin): Likewise.
21270 (expand_builtin_with_bounds): Likewise.
21272 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21275 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21276 print "ds:" only for immediates in generic address space.
21278 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21281 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21282 (ix86_function_ok_for_sibcall): Return false for naked functions.
21284 2017-08-01 Richard Biener <rguenther@suse.de>
21286 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21287 (compute_antic): Seed worklist with exit block predecessors.
21288 * cfganal.c (dfs_find_deadend): For a cycle return the source
21289 of the edge closing it.
21291 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21293 * config/aarch64/aarch64.c
21294 (aarch64_can_const_movi_rtx_p): Move 0 check.
21296 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21298 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21299 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21301 * match.pd: Ditto in address comparison pattern.
21303 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21305 PR tree-optimization/81627
21306 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21307 closed ssa form for store-store chain.
21309 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21311 PR tree-optimization/81620
21312 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21313 for store-store chain.
21315 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21317 PR tree-optimization/81588
21318 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21319 ranges[i].in_p, invert comparison code ccode. For >/>=,
21320 swap rhs1 and rhs2 and comparison code unconditionally,
21321 for </<= don't do that. Don't swap rhs1/rhs2 again if
21322 ranges[i].in_p, instead invert comparison code ccode if
21323 opcode or oe->rank is BIT_IOR_EXPR.
21326 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21327 a direct optab to conversion optab.
21328 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21329 with GET_MODE_INNER as last argument instead of optab_handler.
21330 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21331 vector extraction if possible and optab is available.
21332 * expr.c (store_constructor): Use convert_optab_handler instead
21333 of optab_handler. Use vector initialization from smaller
21334 vectors if possible and optab is available.
21335 * tree-vect-stmts.c (vectorizable_load): Likewise.
21336 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21337 now have two modes.
21338 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21339 of vec_init from half-sized vectors with the same element mode.
21340 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21341 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21342 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21343 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21344 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21345 after mode in gen_vec_extract* calls.
21346 (vec_extract<mode>): Renamed to ...
21347 (vec_extract<mode><ssescalarmodelower>): ... this.
21348 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21349 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21350 element mode after mode in gen_vec_init* calls.
21351 (VEC_INIT_HALF_MODE): New mode iterator.
21352 (vec_init<mode>): Renamed to ...
21353 (vec_init<mode><ssescalarmodelower>): ... this.
21354 (vec_init<mode><ssehalfvecmodelower>): New expander.
21355 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21356 (vec_extractv2sfsf): ... this.
21357 (vec_initv2sf): Renamed to ...
21358 (vec_initv2sfsf): ... this.
21359 (vec_extractv2si): Renamed to ...
21360 (vec_extractv2sisi): ... this.
21361 (vec_initv2si): Renamed to ...
21362 (vec_initv2sisi): ... this.
21363 (vec_extractv4hi): Renamed to ...
21364 (vec_extractv4hihi): ... this.
21365 (vec_initv4hi): Renamed to ...
21366 (vec_initv4hihi): ... this.
21367 (vec_extractv8qi): Renamed to ...
21368 (vec_extractv8qiqi): ... this.
21369 (vec_initv8qi): Renamed to ...
21370 (vec_initv8qiqi): ... this.
21371 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21372 (vec_init<mode>): Renamed to ...
21373 (vec_init<mode><VEC_base_l>): ... this.
21374 (vec_extract<mode>): Renamed to ...
21375 (vec_extract<mode><VEC_base_l>): ... this.
21376 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21377 (vec_initv2sfsf): ... this.
21378 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21379 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21380 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21381 element mode after mode in gen_vec_init* calls.
21382 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21383 (vec_init<mode><Vel>): ... this.
21384 (vec_extract<mode>): Renamed to ...
21385 (vec_extract<mode><Vel>): ... this.
21386 * config/aarch64/iterators.md (Vel): New mode attribute.
21387 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21388 Add element mode after mode in gen_vec_extract* calls.
21389 * config/s390/vector.md (non_vec_l): New mode attribute.
21390 (vec_extract<mode>): Renamed to ...
21391 (vec_extract<mode><non_vec_l>): ... this.
21392 (vec_init<mode>): Renamed to ...
21393 (vec_init<mode><non_vec_l>): ... this.
21394 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21395 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21397 * config/arm/iterators.md (V_elem_l): New mode attribute.
21398 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21399 (vec_extract<mode><V_elem_l>): ... this.
21400 (vec_extractv2di): Renamed to ...
21401 (vec_extractv2didi): ... this.
21402 (vec_init<mode>): Renamed to ...
21403 (vec_init<mode><V_elem_l>): ... this.
21404 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21405 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21406 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21407 Add element mode after gen_vec_extract* calls.
21408 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21409 (vec_init<mode><unitmode>): ... this.
21410 (vec_extract<mode>): Renamed to ...
21411 (vec_extract<mode><unitmode>): ... this.
21412 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21413 (vec_init<mode><unitmode>): ... this.
21414 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21415 (vec_initv2sfsf): ... this.
21416 (vec_extractv2sf): Renamed to ...
21417 (vec_extractv2sfsf): ... this.
21418 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21419 Add element mode after gen_vec_extract* calls.
21420 * config/mips/mips.md (unitmode): New mode iterator.
21421 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21422 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21423 * config/spu/spu.md (inner_l): New mode attribute.
21424 (vec_init<mode>): Renamed to ...
21425 (vec_init<mode><inner_l>): ... this.
21426 (vec_extract<mode>): Renamed to ...
21427 (vec_extract<mode><inner_l>): ... this.
21428 * config/sparc/sparc.md (veltmode): New mode iterator.
21429 (vec_init<VMALL:mode>): Renamed to ...
21430 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21431 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21432 (vec_initv2sisi): ... this.
21433 (vec_initv2sf): Renamed to ...
21434 (vec_initv2sfsf): ... this.
21435 (vec_extractv2sf): Renamed to ...
21436 (vec_extractv2sfsf): ... this.
21437 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21438 (vec_init<mode>): Renamed to ...
21439 (vec_init<mode><VEC_base_l>): ... this.
21440 (vec_extract<mode>): Renamed to ...
21441 (vec_extract<mode><VEC_base_l>): ... this.
21442 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21443 (vec_initv2sfsf): ... this.
21444 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21445 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21446 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21447 gen_vec_init* calls.
21449 2017-08-01 Richard Biener <rguenther@suse.de>
21451 PR tree-optimization/81297
21452 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21453 TREE_OVERFLOW from INTEGER_CSTs.
21455 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21457 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21459 2017-07-31 Carl Love <cel@us.ibm.com>
21461 * config/rs6000/rs6000-c: Add support for built-in functions
21462 vector signed char vec_xl_be (signed long long, signed char *);
21463 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21464 vector signed int vec_xl_be (signed long long, signed int *);
21465 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21466 vector signed long long vec_xl_be (signed long long, signed long long *);
21467 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21468 vector signed short vec_xl_be (signed long long, signed short *);
21469 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21470 vector double vec_xl_be (signed long long, double *);
21471 vector float vec_xl_be (signed long long, float *);
21472 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21473 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21474 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21476 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21477 (altivec_expand_builtin): Add switch statement to call
21478 altivec_expand_xl_be for each builtin.
21479 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21480 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21481 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21482 __builtin_vsx_le_be_v16qi.
21483 * doc/extend.texi: Update the built-in documentation file for the
21484 new built-in functions.
21486 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21489 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21491 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21493 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21495 * config.gcc: Add z14.
21496 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21497 CPU model numbers for z13s and z14.
21498 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21500 * config/s390/s390-opts.h (enum processor_type): Rename
21501 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21502 * config/s390/s390.c (processor_table): Add field for CPU name to
21503 be passed to Binutils.
21504 (s390_asm_output_machine_for_arch): Use the new field in
21505 processor_table for Binutils.
21506 (s390_expand_builtin): Replace arch12 with z14.
21507 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21508 (s390_get_sched_attrmask): Likewise.
21509 (s390_get_unit_mask): Likewise.
21510 * config/s390/s390.opt: Add z14 to processor_type enum.
21512 2017-07-31 Martin Jambor <mjambor@suse.cz>
21515 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21516 regardless of optimization level.
21518 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21519 Martin Liska <mliska@suse.cz>
21521 * predict.def: Remove old comment and adjust probability.
21522 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21523 PREDICT statements.
21525 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21528 * config/i386/i386.c (ix86_function_naked): New function.
21529 (ix86_can_use_return_insn_p): Return false for naked functions.
21530 (ix86_expand_prologue): Skip prologue for naked functions.
21531 (ix86_expand_epilogue): Skip epilogue for naked functions
21532 and emit trap instruction.
21533 (ix86_warn_func_return): New function.
21534 (ix86_attribute_table): Add "naked" attribute specification.
21535 (TARGET_WARN_FUNC_RETURN): Define.
21536 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21538 2017-07-31 Martin Liska <mliska@suse.cz>
21540 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21541 (dump_gimple_bb_header): Always dump BB info.
21542 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21544 2017-07-31 Martin Liska <mliska@suse.cz>
21547 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21548 also with current_function_decl non-null equality.
21550 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21553 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21554 change type to the element type, instead add eltype variable and
21555 use it where we are interested in the element type.
21557 PR tree-optimization/81603
21558 * ipa-polymorphic-call.c
21559 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21560 offset arithmetic in offset_int, bail out if the resulting bit offset
21561 doesn't fit into shwi.
21563 2017-07-31 Martin Liska <mliska@suse.cz>
21565 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21566 (gimplify_save_expr): Fix comment.
21568 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21571 * config/i386/i386.c (ix86_function_arg): Update arguments for
21573 (ix86_compute_frame_layout): Set the initial stack offset to
21574 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21575 INCOMING_FRAME_SP_OFFSET.
21576 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21577 stack before exception handler returns.
21578 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21579 the 'ERROR_CODE' for exception handler.
21581 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21583 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21584 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21585 (ASM_OUTPUT_REG_POP): Ditto.
21586 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21587 instead of asm_fprintf to output pure string.
21589 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21591 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21592 to imported_module_or_decl hook.
21593 (debug_nothing_tree_tree_tree_bool): Remove.
21594 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21595 * debug.c (do_nothing_debug_hooks): Use
21596 debug_nothing_tree_tree_tree_bool_bool instead of
21597 debug_nothing_tree_tree_tree_bool.
21598 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21599 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21600 * sdbout.c (sdb_debug_hooks): Likewise.
21601 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21602 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21604 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21605 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21606 attribute, don't add anything.
21608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21610 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21611 (fold_build2_stat_loc): Likewise.
21612 (fold_build3_stat_loc): Likewise.
21613 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21614 (fold_build1_loc): Remove macro.
21615 (fold_build2_loc): Likewise.
21616 (fold_build3_loc): Likewise.
21618 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21620 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21621 (gimple_build_debug_bind_source_stat): Likewise.
21622 * gimple.h (gimple_build_debug_bind): Remove macro.
21623 (gimple_build_debug_bind_source): Likewise.
21625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21627 * bitmap.c (bitmap_alloc): Adjust.
21628 (bitmap_gc_alloc): Likewise.
21629 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21631 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21633 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21634 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21635 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21636 (bitmap_gc_alloc_stat): Likewise.
21637 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21639 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21641 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21642 * rtl.h (shallow_copy_rtx): Remove macro.
21644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21646 * emit-rtl.c (gen_raw_REG): Adjust.
21647 * gengenrtl.c (gendef): Likewise.
21648 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21649 * rtl.h (rtx_alloc): Remove macro.
21651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21653 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21654 (build_tree_list_stat): Likewise.
21655 * tree.h (build_tree_list): Remove macro.
21656 (build_tree_list_vec): Likewise.
21658 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21660 * tree.c (make_vector_stat): Remove _stat from name.
21661 (build_vector_stat): Likewise.
21662 * tree.h (make_vector_stat): Remove macro.
21663 (build_vector_stat): Likewise.
21665 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21667 * tree.h (build_var_debug_value): Remove prototype.
21669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21671 * tree.c (tree_cons_stat): Remove _stat from name.
21672 * tree.h (tree_cons): Remove macro.
21674 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21676 * tree.c (build_vl_exp_stat): Remove _stat from name.
21677 * tree.h (build_vl_exp): Remove macro.
21679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21681 * tree.c (build_decl_stat): Remove _stat from name.
21682 * tree.h (build_decl): Remove macro.
21684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21686 * gimple.c (gimple_build_with_ops_stat): Adjust.
21687 (gimple_alloc_stat): Remove _stat from name.
21688 * gimple.h (gimple_alloc): Remove macro.
21690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21692 * tree.c (make_tree_vec_stat): Remove _stat from name.
21693 (grow_tree_vec_stat): Likewise.
21694 * tree.h (make_tree_vec_stat): Adjust prototype.
21695 (grow_tree_vec_stat): Likewise.
21696 (make_tree_vec): Remove macro.
21697 (grow_tree_vec): Likewise.
21699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21701 * fold-const.c (fold_build1_stat_loc): Adjust.
21702 (fold_build2_stat_loc): Likewise.
21703 (fold_build3_stat_loc): Likewise.
21704 * tree.c (build0_stat): Remove _stat from name.
21705 (build1_stat): Likewise.
21706 (build2_stat): Likewise.
21707 (build3_stat): Likewise.
21708 (build4_stat): Likewise.
21709 (build5_stat): Likewise.
21710 * tree.h (build1_loc): Remove macro, and rename _stat function
21712 (build2_loc): Likewise.
21713 (build3_loc): Likewise.
21714 (build4_loc): Likewise.
21715 (build5_loc): Likewise.
21717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21719 * tree.c (make_int_cst_stat): Remove _stat from name.
21720 * tree.h (make_int_cst_stat): Adjust prototype.
21721 (make_int_cst): Remove macro.
21723 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21725 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21726 * tree.h (make_tree_binfo_stat): Adjust prototype.
21727 (make_tree_binfo): Remove.
21729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21731 * tree.c (copy_node_stat): Rename to copy_node.
21732 (build_distinct_type_copy): Adjust.
21733 * tree.h (copy_node_stat): Adjust prototype.
21734 (copy_node): Remove macro.
21736 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21738 * tree.c (make_node_stat): rename to make_node.
21739 (build_tree_list_stat): Adjust.
21740 (build0_stat): Likewise.
21741 (build2_stat): Likewise.
21742 (build3_stat): Likewise.
21743 (build4_stat): Likewise.
21744 (build5_stat): Likewise.
21745 (build_decl_stat): Likewise.
21746 * tree.h (make_node_stat): Adjust prototype.
21747 (make_node): remove macro.
21749 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21751 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21752 (PPC_FEATURE2_SCV): Likewise.
21753 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21755 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21757 * config/aarch64/aarch64.c
21758 (aarch64_internal_mov_immediate): Add new special pattern.
21759 * config/aarch64/aarch64.md (*movdi_aarch64):
21760 Add reg/32bit const mov case.
21762 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21763 Richard Sandiford <richard.sandiford@linaro.org>
21765 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21766 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21767 Add integer and movi cases.
21768 (movi-split-hf-df-sf split, fp16): New.
21769 (enabled): Added TARGET_FP_F16INST.
21770 * config/aarch64/iterators.md (GPF_HF): New.
21771 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21773 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21775 * config/aarch64/aarch64.c
21776 (aarch64_simd_container_mode): Add prototype.
21777 (aarch64_expand_mov_immediate): Add HI support.
21778 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21779 (aarch64_can_const_movi_rtx_p): New.
21780 (aarch64_preferred_reload_class):
21781 Remove restrictions of using FP registers for certain SIMD operations.
21782 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21783 (aarch64_valid_floating_const): Add integer move validation.
21784 (aarch64_simd_imm_scalar_p): Remove.
21785 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21786 (aarch64_legitimate_constant_p): Expand list of supported cases.
21787 * config/aarch64/aarch64-protos.h
21788 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21789 (aarch64_reinterpret_float_as_int): New.
21790 (aarch64_simd_imm_scalar_p): Remove.
21791 * config/aarch64/constraints.md (Uvi): New.
21792 (Dd): Split into Ds and new Dd.
21793 * config/aarch64/aarch64.md (*movsi_aarch64):
21795 (*movdi_aarch64): Add SIMD mov case.
21797 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21799 * tree-predcom.c: (struct chain): Handle store-store chain in which
21800 stores for elimination only store loop invariant values.
21801 (execute_pred_commoning_chain): Ditto.
21802 (prepare_initializers_chain_store_elim): Ditto.
21803 (prepare_finalizers): Ditto.
21804 (is_inv_store_elimination_chain): New function.
21805 (initialize_root_vars_store_elim_1): New function.
21807 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21809 * tree-predcom.c: Revise general description of the pass.
21810 (enum chain_type): New enum type for store elimination.
21811 (struct chain): New field supporting store elimination.
21812 (struct component): Ditto.
21813 (dump_chain): Dump store-stores chain.
21814 (release_chain): Release resources.
21815 (split_data_refs_to_components): Compute and create component
21816 contains only stores for elimination.
21817 (get_chain_last_ref_at): New function.
21818 (make_invariant_chain): Initialization.
21819 (make_rooted_chain): Specify chain type in parameter and record it.
21820 (add_looparound_copies): Skip for store-stores chain.
21821 (determine_roots_comp): Compute type of chain and pass it to
21823 (initialize_root_vars_store_elim_2): New function.
21824 (finalize_eliminated_stores): New function.
21825 (remove_stmt): Handle store for elimination.
21826 (execute_pred_commoning_chain): Execute predictive commoning on
21827 store-store chains.
21828 (determine_unroll_factor): Skip unroll for store-stores chain.
21829 (prepare_initializers_chain_store_elim): New function.
21830 (prepare_initializers_chain): Hanlde store-store chain.
21831 (prepare_finalizers_chain, prepare_finalizers): New function.
21832 (tree_predictive_commoning_loop): Return integer value indicating
21833 if loop is unrolled or lcssa form is corrupted.
21834 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21836 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21838 * tree-predcom.c (initialize_root): Delete.
21839 (execute_pred_commoning_chain): Initialize root vars and replace
21840 reference of non-combined chain directly, rather than call above
21843 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21845 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21846 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21848 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21850 * tree-predcom.c (struct chain): New field init_seq.
21851 (release_chain): Release init_seq.
21852 (prepare_initializers_chain): Record intialization stmts in above
21854 (insert_init_seqs): New function.
21855 (tree_predictive_commoning_loop): Call insert_init_seqs.
21857 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21859 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21861 2017-07-28 Richard Biener <rguenther@suse.de>
21863 * match.pd: Remove superfluous :c.
21864 * genmatch.c (simplify::id): Add member.
21865 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21867 (current_id): New global.
21868 (dt_node::parent): Move from ...
21869 (dt_operand::parent): ... here. Add for_id member.
21870 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21871 (decision_tree::find_node): Relax order requirement when
21872 merging DT_TRUE nodes to ones inbetween the current simplify
21873 and the one we try to merge with. Add diagnostic whenever
21874 we need to enforce pattern order by not merging.
21875 (decision_tree::insert): Set current_id.
21876 (decision_tree::print_node): Dump parent node and for_id.
21877 (parser::last_id): Add member.
21878 (parser::push_simplify): Assign unique id.
21879 (parser::parser): Initialize last_id.
21881 2017-07-28 Martin Liska <mliska@suse.cz>
21884 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21885 gimple_build_debug_bind.
21887 2017-07-28 Richard Biener <rguenther@suse.de>
21889 PR tree-optimization/81502
21890 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21892 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21894 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21895 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21896 for BIT_FIELD_REF args.
21897 * fold-const.c (make_bit_field_ref): Likewise.
21898 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21900 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21903 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21904 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21905 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21906 Or it into SANITIZER_UNDEFINED.
21907 * ubsan.c: Include gimple-fold.h and varasm.h.
21908 (ubsan_expand_ptr_ifn): New function.
21909 (instrument_pointer_overflow): New function.
21910 (maybe_instrument_pointer_overflow): New function.
21911 (instrument_object_size): Formatting fix.
21912 (pass_ubsan::execute): Call instrument_pointer_overflow
21913 and maybe_instrument_pointer_overflow.
21914 * internal-fn.c (expand_UBSAN_PTR): New function.
21915 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21916 * sanitizer.def (__ubsan_handle_pointer_overflow,
21917 __ubsan_handle_pointer_overflow_abort): New builtins.
21918 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21919 * internal-fn.def (UBSAN_PTR): New internal function.
21920 * opts.c (sanitizer_opts): Add pointer-overflow.
21921 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21922 * fold-const.c (build_range_check): Compute pointer range check in
21923 integral type if pointer arithmetics would be needed. Formatting
21926 2017-07-28 Martin Liska <mliska@suse.cz>
21929 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21930 parameters that are of a variable-length.
21932 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21934 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21936 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21937 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21938 (CRT_CALL_STATIC_FUNCTION): Likewise.
21939 (ASM_DEFAULT_SPEC): New define.
21940 (ASM_SPEC32): Likewise.
21941 (ASM_SPEC64): Likewise.
21942 (ASM_SPEC_COMMON): Likewise.
21943 (ASM_SPEC): Likewise.
21944 (INVALID_64BIT): Likewise.
21945 (LINK_OS_DEFAULT_SPEC): Likewise.
21946 (LINK_OS_SPEC32): Likewise.
21947 (LINK_OS_SPEC64): Likewise.
21948 (POWERPC_LINUX): Likewise.
21949 (PTRDIFF_TYPE): Likewise.
21950 (RESTORE_FP_PREFIX): Likewise.
21951 (RESTORE_FP_SUFFIX): Likewise.
21952 (SAVE_FP_PREFIX): Likewise.
21953 (SAVE_FP_SUFFIX): Likewise.
21954 (SIZE_TYPE): Likewise.
21955 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21956 (TARGET_64BIT): Likewise.
21957 (TARGET_64BIT): Likewise.
21958 (TARGET_AIX): Likewise.
21959 (WCHAR_TYPE_SIZE): Likewise.
21960 (WCHAR_TYPE): Undefine.
21961 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21962 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21963 (CPP_OS_RTEMS_SPEC): Delete.
21964 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21965 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21967 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21969 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21971 PR tree-optimization/81578
21972 * tree-parloops.c (build_new_reduction): Bail out if
21973 reduction_code isn't one of the standard OpenMP reductions.
21974 Move the details printing after that decision.
21976 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21978 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21979 related to reload_in_progress.
21980 (splat_input_operand): Likewise.
21981 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21983 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21985 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21986 (TARGET_INSTANTIATE_DECLS): Likewise.
21987 (legitimate_indexed_address_p): Delete reload_in_progress code.
21988 (rs6000_debug_legitimate_address_p): Likewise.
21989 (rs6000_eliminate_indexed_memrefs): Likewise.
21990 (rs6000_emit_le_vsx_store): Likewise.
21991 (rs6000_emit_move_si_sf_subreg): Likewise.
21992 (rs6000_emit_move): Likewise.
21993 (register_to_reg_type): Likewise.
21994 (rs6000_pre_atomic_barrier): Likewise.
21995 (rs6000_machopic_legitimize_pic_address): Likewise.
21996 (rs6000_allocate_stack_temp): Likewise.
21997 (rs6000_address_for_fpconvert): Likewise.
21998 (rs6000_address_for_altivec): Likewise.
21999 (rs6000_secondary_memory_needed_rtx): Delete function.
22000 (rs6000_check_sdmode): Likewise.
22001 (rs6000_alloc_sdmode_stack_slot): Likewise.
22002 (rs6000_instantiate_decls): Likewise.
22003 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22004 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22005 Delete reload_in_progress.
22006 (*vec_reload_and_plus_<mptrsize>): Likewise.
22007 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22008 (vsx_div_v2di): Likewise.
22009 (vsx_udiv_v2di): Likewise.
22011 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22013 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22014 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22015 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22016 (rs6000_debug_reg_global): Delete print of LRA status.
22017 (rs6000_option_override_internal): Delete dead LRA related code.
22018 (rs6000_lra_p): Delete function.
22019 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22021 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22023 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22024 * config/riscv/rtems.h: New file.
22026 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22027 Sudakshina Das <sudi.das@arm.com>
22029 * config/aarch64/aarch64.md
22030 (define_split for and<mode>3nr_compare): Move
22031 non aarch64_logical_operand to a register.
22032 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22033 register immediate operand to a register.
22034 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22036 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22038 PR middle-end/81564
22039 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22041 2017-07-27 Richard Biener <rguenther@suse.de>
22043 PR tree-optimization/81573
22044 PR tree-optimization/81494
22045 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22046 multi defuse cycle case.
22048 2017-07-27 Richard Biener <rguenther@suse.de>
22050 PR tree-optimization/81571
22051 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22054 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22056 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22057 earlier and only if MASK_FPU is set. Adjust formatting.
22059 2017-07-27 Martin Liska <mliska@suse.cz>
22061 * opt-functions.awk: Add validation of value of Init.
22062 * optc-gen.awk: Pass new argument.
22064 2017-07-27 Martin Liska <mliska@suse.cz>
22066 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22067 Fix wrong condition.
22069 2017-07-27 Martin Liska <mliska@suse.cz>
22071 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22072 BBs and edges seen by autoFDO.
22074 2017-07-27 Richard Biener <rguenther@suse.de>
22076 PR tree-optimization/81502
22077 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22078 with incompatible but same sized type.
22079 (execute_update_addresses_taken): Likewise.
22081 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22083 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22084 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22086 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22089 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22090 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22091 Change s_operand to memory_operand.
22093 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22095 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22096 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22097 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22098 Emit instructions rather than returning an expression. Handle TFmode
22099 and KFmode by casting to TImode.
22100 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22101 (rs6000_emit_le_vsx_store): Likewise.
22102 * config/rs6000/vsx.md (VSX_TI): New iterator.
22103 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22104 (*vsx_le_undo_permute_<mode>): Likewise.
22105 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22106 emit the split sequence.
22107 (*vsx_le_perm_store_<mode>): Likewise.
22109 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22111 PR tree-optimization/81555
22112 PR tree-optimization/81556
22113 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22114 if true, force CHANGED for the recursive invocation.
22115 (reassociate_bb): Remember original length of ops array, pass
22116 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22118 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22119 attributes for noipa attribute. For naked attribute use
22120 lookup_attribute first before lookup_attribute_spec.
22121 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22123 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22124 for functions with noipa attribute.
22125 (cgraph_externally_visible_p): Return true for functions with noipa
22127 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22128 for functions with noipa attribute.
22129 * doc/extend.texi: Document noipa function attribute.
22130 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22131 also for functions with noipa attribute.
22132 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22134 2017-07-26 Andrew Pinski <apinski@cavium.com>
22136 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22137 vec_unalign_load_cost and vec_unalign_store_cost.
22139 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22141 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22142 -mvsx-small-integer option.
22143 (ISA_3_0_MASKS_IEEE): Likewise.
22144 (OTHER_VSX_VECTOR_MASKS): Likewise.
22145 (POWERPC_MASKS): Likewise.
22146 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22147 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22148 code, only testing for DImode being allowed in non-VSX floating
22150 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22151 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22153 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22154 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22155 TARGET_P8_VECTOR test.
22156 (rs6000_secondary_reload_simple_move): Likewise.
22157 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22158 since TARGET_P9_VECTOR was already tested.
22159 (rs6000_opt_masks): Remove -mvsx-small-integer.
22160 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22161 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22163 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22164 test for TARGET_VEXTRACTUB was used, and that uses
22166 (p9 extract splitter): Likewise.
22167 (vsx_extract_<mode>_di_p9): Likewise.
22168 (vsx_extract_<mode>_store_p9): Likewise.
22169 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22170 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22171 the elimination of TARGET_VSX_SMALL_INTEGER.
22172 (vsx_extract_<mode>_p8): Likewise.
22173 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22174 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22175 (vsx_set_<mode>_p9): Likewise.
22176 (vsx_set_v4sf_p9): Likewise.
22177 (vsx_set_v4sf_p9_zero): Likewise.
22178 (vsx_insert_extract_v4sf_p9): Likewise.
22179 (vsx_insert_extract_v4sf_p9_2): Likewise.
22180 * config/rs6000/rs6000.md (sign extend splitter): Change
22181 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22182 (floatsi<mode>2_lfiwax_mem): Likewise.
22183 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22184 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22185 since a test for TARGET_P9_VECTOR was used.
22186 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22187 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22188 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22189 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22190 TARGET_P8_VECTOR test.
22191 (fix_trunc<mode>si2_stfiwx): Likewise.
22192 (fix_trunc<mode>si2_internal): Likewise.
22193 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22194 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22196 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22197 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22198 TARGET_P8_VECTOR test.
22199 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22200 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22201 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22203 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22204 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22205 since a test for TARGET_P9_VECTOR was used.
22206 (splitter for loading small constants): Likewise.
22208 2017-07-26 Andrew Pinski <apinski@cavium.com>
22210 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22213 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22216 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22217 (fp_valid_at): Likewise.
22219 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22221 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22222 (qdf24xx_addrcost_table): Likewise.
22223 (cortexa57_tunings): Update to use generic_branch_cost.
22224 (cortexa72_tunings): Likewise.
22225 (cortexa73_tunings): Likewise.
22226 (qdf24xx_tunings): Likewise.
22228 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22230 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22231 (thunderx2t99_branch_cost): Likewise.
22232 (cortexa35_tunings): Update to use generic_branch_cost.
22233 (cortexa53_tunings): Likewise.
22234 (cortexa57_tunings): Likewise.
22235 (cortexa72_tunings): Likewise.
22236 (cortexa73_tunings): Likewise.
22237 (thunderx2t99_tunings): Likewise.
22239 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22241 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22242 (sparc_option_override): Honour MASK_FSMULD.
22243 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22244 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22245 * config/sparc/sparc.opt (mfsmuld): New option.
22246 * doc/invoke.texi (mfsmuld): Document option.
22248 2017-07-26 Marek Polacek <polacek@redhat.com>
22250 PR middle-end/70992
22251 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22253 2017-07-26 Richard Biener <rguenther@suse.de>
22255 * gimple-match-head.c (do_valueize): Return OP if valueize
22257 (get_def): New helper to get at the def stmt of a SSA name
22258 if valueize allows.
22259 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22260 do_valueize to get at the def stmt.
22261 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22263 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22265 PR middle-end/46932
22266 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22268 2017-07-26 Martin Liska <mliska@suse.cz>
22271 * function.c (expand_function_start): Make expansion of
22272 nonlocal_goto_save_area after parm_birth_insn.
22274 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22276 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22277 from all CPU target flags enable members.
22279 2017-07-26 Richard Biener <rguenther@suse.de>
22281 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22282 (decision_tree::gen): Make 'type' const.
22283 (write_predicate): Likewise.
22285 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22287 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22288 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22289 (rs6000_option_override_internal): Likewise.
22290 (rs6000_expand_vector_set): Likewise.
22291 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22292 (TARGET_UPPER_REGS_SF): Likewise.
22293 (TARGET_UPPER_REGS_DI): Likewise.
22294 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22295 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22296 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22297 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22298 (Splitters for DI constants in Altivec registers): Likewise.
22299 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22300 (vsx_set_v4sf_p9): Likewise.
22301 (vsx_set_v4sf_p9_zero): Likewise.
22302 (vsx_insert_extract_v4sf_p9): Likewise.
22303 (vsx_insert_extract_v4sf_p9_2): Likewise.
22305 2017-07-25 Carl Love <cel@us.ibm.com>
22307 * doc/extend.texi: Update the built-in documentation file for the
22308 existing built-in functions
22309 vector signed char vec_cnttz (vector signed char);
22310 vector unsigned char vec_cnttz (vector unsigned char);
22311 vector signed short vec_cnttz (vector signed short);
22312 vector unsigned short vec_cnttz (vector unsigned short);
22313 vector signed int vec_cnttz (vector signed int);
22314 vector unsigned int vec_cnttz (vector unsigned int);
22315 vector signed long long vec_cnttz (vector signed long long);
22316 vector unsigned long long vec_cnttz (vector unsigned long long);
22318 2017-07-25 Andrew Pinski <apinski@cavium.com>
22320 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22321 accesses where the use is for the first operand of a BIT_INSERT.
22323 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22326 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22327 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22329 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22331 * config/i386/gstabs.h: Delete.
22332 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22334 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22336 * config/i386/i386.c (ix86_decompose_address): Do not check for
22337 register RTX when looking at index_reg or base_reg.
22338 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22340 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22342 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22343 to update EH info here.
22345 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22347 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22349 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22351 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22353 2017-07-25 Torsten Duwe <duwe@suse.de>
22355 * common.opt: Introduce -fpatchable-function-entry
22356 command line option, and its variables function_entry_patch_area_size
22357 and function_entry_patch_area_start.
22358 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22359 including a two-value parser.
22360 * target.def (print_patchable_function_entry): New target hook.
22361 * targhooks.h (default_print_patchable_function_entry): New function.
22362 * targhooks.c (default_print_patchable_function_entry): Likewise.
22363 * toplev.c (process_options): Switch off IPA-RA if
22364 patchable function entries are being generated.
22365 * varasm.c (assemble_start_function): Look at the
22366 patchable-function-entry command line switch and current
22367 function attributes and maybe generate NOP instructions by
22368 calling the print_patchable_function_entry hook.
22369 * doc/extend.texi: Document patchable_function_entry attribute.
22370 * doc/invoke.texi: Document -fpatchable_function_entry
22371 command line option.
22372 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22374 * doc/tm.texi: Re-generate.
22376 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22379 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22380 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22382 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22384 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22386 2017-07-25 Richard Biener <rguenther@suse.de>
22388 PR tree-optimization/81455
22389 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22390 not walk in cycles when looking for guards.
22392 2017-07-25 Richard Biener <rguenther@suse.de>
22394 PR tree-optimization/81529
22395 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22396 when optimizing backedge uses.
22398 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22400 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22402 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22403 to dl_section_ref. On AIX, append an expression to subtract
22404 the size of the section length to dl_section_ref.
22406 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22408 * configure.ac: If any of the config.* scripts fail, exit 1.
22409 * configure: Regenerate.
22411 2017-07-25 Richard Biener <rguenther@suse.de>
22413 PR middle-end/81546
22414 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22415 of immediate uses, be more verbose on errors.
22417 2017-07-25 Richard Biener <rguenther@suse.de>
22419 PR tree-optimization/81510
22420 * tree-vect-loop.c (vect_is_simple_reduction): When the
22421 reduction stmt is not inside the loop bail out.
22423 2017-07-25 Richard Biener <rguenther@suse.de>
22425 PR tree-optimization/81303
22426 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22427 profitability check against LOOP_VINFO_NITERSM1.
22429 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22431 * domwalk.c (cmp_bb_postorder): Simplify.
22432 (sort_bbs_postorder): New function. Use it...
22433 (dom_walker::walk): ...here to optimize common cases.
22435 2017-07-25 Martin Liska <mliska@suse.cz>
22438 * ipa-visibility.c (function_and_variable_visibility): Make the
22439 redirection just on target that supports aliasing.
22440 Fix GNU coding style.
22442 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22445 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22447 * config/arm/rtems.h: Likewise.
22448 * config/bfin/rtems.h: Likewise.
22449 * config/i386/rtemself.h: Likewise.
22450 * config/lm32/rtems.h: Likewise.
22451 * config/m32c/rtems.h: Likewise.
22452 * config/m68k/rtemself.h: Likewise.
22453 * config/microblaze/rtems.h: Likewise.
22454 * config/mips/rtems.h: Likewise.
22455 * config/moxie/rtems.h: Likewise.
22456 * config/nios2/rtems.h: Likewise.
22457 * config/powerpcspe/rtems.h: Likewise.
22458 * config/rs6000/rtems.h: Likewise.
22459 * config/rtems.h: Likewise.
22460 * config/sh/rtems.h: Likewise.
22461 * config/sh/rtemself.h: Likewise.
22462 * config/sparc/rtemself.h: Likewise.
22464 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22467 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22468 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22469 * tree-ssa-structalias.c (alias_get_name): Same.
22471 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22474 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22475 instructions if no du chain is found.
22477 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22479 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22481 2017-07-25 Richard Biener <rguenther@suse.de>
22483 PR middle-end/81505
22484 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22487 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22489 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22490 upper-regs options.
22491 (ISA_2_7_MASKS_SERVER): Likewise.
22492 (ISA_3_0_MASKS_IEEE): Likewise.
22493 (OTHER_P8_VECTOR_MASKS): Likewise.
22494 (OTHER_VSX_VECTOR_MASKS): Likewise.
22495 (POWERPC_MASKS): Likewise.
22496 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22497 duplicate list of options.
22498 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22499 explicit -mupper-regs options.
22500 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22501 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22502 alias for -mupper-regs-df.
22503 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22504 (rs6000_init_hard_regno_mode_ok): Likewise.
22505 (rs6000_option_override_internal): Likewise.
22506 (rs6000_opt_masks): Likewise.
22507 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22508 options in terms of whether -mvsx or -mpower8-vector was used.
22509 (TARGET_UPPER_REGS_DI): Likewise.
22510 (TARGET_UPPER_REGS_SF): Likewise.
22511 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22512 -mupper-regs-* options.
22514 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22516 * passes.c (emergency_dump_function): Print some empty lines and a
22517 header before the RTL dump.
22519 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22521 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22523 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22526 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22527 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22529 2017-07-24 Carl Love <cel@us.ibm.com>
22531 * config/rs6000/rs6000-c.c: Add support for built-in functions
22532 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22533 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22534 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22535 vec_extract_fp_from_shortl): Add defines for the two builtins.
22536 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22537 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22539 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22540 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22541 * doc/extend.texi: Update the built-in documentation file for the
22542 new built-in function.
22544 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22547 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22549 * doc/generic.texi: Likewise.
22550 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22551 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22553 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22555 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22556 (aarch64_mls_elt_merge<mode>): Likewise.
22558 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22560 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22561 having __cxa_atexit.
22563 2017-07-23 Michael Collison <michael.collison@arm.com>
22565 * config/arm/arm.c (arm_option_override): Deprecate
22566 use of -mstructure-size-boundary.
22567 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22568 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22570 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22573 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22574 Reduce cost estimate for direct moves.
22576 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22579 * config/i386/i386.c (ix86_option_override_internal): Disable
22580 BMI, BMI2 and TBM instructions for -m16.
22582 2017-07-21 Carl Love <cel@us.ibm.com>
22584 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22585 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22586 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22587 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22588 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22589 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22590 VMULOSW): New enum "unspec" values.
22591 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22592 altivec_vmulosw): New patterns.
22593 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22594 VMULOSW): Add definitions.
22596 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22598 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22599 (qdf24xx): Likewise.
22600 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22601 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22602 (AARCH64_FL_V8_1): Renumber.
22603 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22604 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22605 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22606 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22607 rdma to feature modifiers list.
22609 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22611 PR middle-end/56727
22612 * ipa-visibility (function_and_variable_visibility): Convert
22613 recursive PLT call to direct call if appropriate.
22615 2017-07-21 Andrew Pinski <apinski@cavium.com>
22617 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22618 operand 1 to see if the types precision matches.
22619 * fold-const.c (operand_equal_p): Likewise.
22621 2017-07-21 Richard Biener <rguenther@suse.de>
22623 PR tree-optimization/81303
22624 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22625 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22626 (vect_peeling_hash_get_lowest_cost): Adjust.
22627 (vect_enhance_data_refs_alignment): Likewise. Use
22628 vect_get_peeling_costs_all_drs to compute the penalty for no
22629 peeling to match up costs.
22631 2017-07-21 Richard Biener <rguenther@suse.de>
22633 PR tree-optimization/81500
22634 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22635 we didn't identify a reduction path.
22637 2017-07-21 Tom de Vries <tom@codesourcery.com>
22638 Cesar Philippidis <cesar@codesourcery.com>
22640 PR gcov-profile/81442
22641 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22644 2017-07-21 Tom de Vries <tom@codesourcery.com>
22647 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22649 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22650 nvptx_override_options_after_change.
22652 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22654 * dwarf2out.c (output_file_names): Avoid double testing for
22655 dwarf_version >= 5.
22657 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22659 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22661 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22663 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22665 (try_crossjump_to_edge): Do not punt on partitioned functions.
22667 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22669 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22670 Put all BBs reachable only via paths crossing cold region to cold
22672 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22674 2016-07-21 Richard Biener <rguenther@suse.de>
22676 PR tree-optimization/81303
22677 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22678 into account prologue and epilogue iterations when raising
22679 min_profitable_iters to sth at least covering one vector iteration.
22681 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22683 * config/arm/arm.c (arm_test_cpu_arch_dat):
22686 2017-07-20 Nathan Sidwell <nathan@acm.org>
22688 Remove TYPE_METHODS.
22689 * tree.h (TYPE_METHODS): Delete.
22690 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22691 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22692 (dbxout_type_methods): Scan TYPE_FIELDS.
22693 (dbxout_type): Don't check TYPE_METHODS here.
22694 * function.c (use_register_for_decl): Always ignore register for
22695 class types when not optimizing.
22696 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22697 * tree.c (free_lang_data_in_type): Stitch out member functions and
22698 templates from TYPE_FIELDS.
22699 (build_distinct_type_copy, verify_type_variant,
22700 verify_type): Member fns are on TYPE_FIELDS.
22701 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22702 * tree-pretty-print.c (dump_generic_node): Likewise.
22704 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22707 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22708 V2TImode and V4TImode.
22709 (ix86_expand_vector_extract): Likewise.
22710 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22711 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22712 (ssescalarmode): Handle V4TImode and V2TImode.
22713 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22714 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22715 (VEXTRACTI128_MODE): New mode iterator.
22716 (splitter for *vec_extractv?ti first element): New.
22717 (VEC_INIT_MODE): New mode iterator.
22718 (vec_init<mode>): Consolidate 3 expanders into one using
22719 VEC_INIT_MODE mode iterator.
22721 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22723 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22724 non_spilled_static_chain_regno_p.
22726 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22728 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22730 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22732 * bb-reorder.c (connect_traces): Allow copying of blocks within
22735 2017-07-20 Richard Biener <rguenther@suse.de>
22737 * gimple.h (gimple_phi_result): Add gphi * overload.
22738 (gimple_phi_result_ptr): Likewise.
22739 (gimple_phi_arg): Likewise. Adjust index assert to only
22740 allow actual argument accesses rather than all slots available
22742 (gimple_phi_arg_def): Add gphi * overload.
22743 * tree-phinodes.c (make_phi_node): Initialize only actual
22745 (resize_phi_node): Clear memory not covered by old node,
22746 do not initialize excess argument slots.
22747 (reserve_phi_args_for_new_edge): Initialize new argument slot
22750 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22752 PR tree-optimization/81388
22754 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22756 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22757 by removing computation of may_be_zero.
22759 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22760 Tom de Vries <tom@codesourcery.com>
22762 PR middle-end/81030
22763 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22764 when gimple level profile disagrees with what RTL expander did.
22766 2017-07-20 Richard Biener <rguenther@suse.de>
22768 PR tree-optimization/61171
22769 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22770 (vect_analyze_stmt): Add slp instance parameter.
22771 (vectorizable_reduction): Likewise.
22772 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22773 (vect_is_simple_reduction): Deal with chains not detected
22774 as SLP reduction chain, specifically not properly associated
22775 chains containing a mix of plus/minus.
22776 (get_reduction_op): Remove.
22777 (get_initial_defs_for_reduction): Simplify, pass in whether
22778 this is a reduction chain, pass in the SLP node for the PHIs.
22779 (vect_create_epilog_for_reduction): Get the SLP instance as
22781 (vectorizable_reduction): Get the SLP instance as arg.
22782 During analysis remember the SLP node with the PHIs in the
22783 instance. Simplify getting at the vectorized reduction PHIs.
22784 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22785 through SLP instance.
22786 (vect_slp_analyze_operations): Likewise.
22787 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22788 (vect_transform_stmt): Likewise.
22790 2017-07-20 Tom de Vries <tom@codesourcery.com>
22792 PR tree-optimization/81489
22793 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22794 read of phi arg location to before loop that modifies phi.
22796 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22798 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22801 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22803 PR middle-end/81331
22804 * except.c (execute): Fix ordering issue.
22806 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22808 PR rtl-optimization/81423
22809 * combine.c (make_compound_operation_int): Don't try to optimize
22810 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22812 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22814 PR rtl-optimization/81423
22815 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22816 with a constant that is -1 in the truncated to mode.
22818 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22820 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22821 (determine_unlikely_bbs): ... here.
22822 * predict.h (propagate_unlikely_bbs_forward): Declare.
22823 * cfgexpand.c (pass_expand::execute): Use it.
22824 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22826 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22827 propagate_unlikely_bbs_forward.
22829 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22831 PR middle-end/81331
22832 * except.c (maybe_add_nop_after_section_switch): New function.
22835 2017-07-19 Tom de Vries <tom@codesourcery.com>
22837 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22839 2017-07-19 Tom de Vries <tom@codesourcery.com>
22841 * gimple.h (gimple_phi_arg): Make assert more strict.
22843 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22845 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22846 * config/rs6000/mmintrin.h: New file.
22847 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22849 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22851 PR tree-optimization/81346
22852 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22854 2017-07-19 Tom de Vries <tom@codesourcery.com>
22856 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22858 2017-07-19 Tom de Vries <tom@codesourcery.com>
22860 * config/nvptx/nvptx-modes.def: Add V2DImode.
22861 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22862 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22863 (nvptx_output_mov_insn): Handle lack of mov.b128.
22864 (nvptx_print_operand): Handle 'H' and 'L' codes.
22865 (nvptx_vector_mode_supported): Allow V2DImode.
22866 (nvptx_preferred_simd_mode): New function.
22867 (nvptx_data_alignment): New function.
22868 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22869 nvptx_preferred_simd_mode.
22870 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22872 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22874 2017-07-19 Tom de Vries <tom@codesourcery.com>
22876 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22877 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22878 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22879 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22880 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22881 (mov<VECIM>_insn): New define_insn.
22882 (define_expand "mov<VECIM>): New define_expand.
22884 2017-07-19 Tom de Vries <tom@codesourcery.com>
22886 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22888 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22890 PR tree-optimization/81346
22891 * fold-const.h (fold_div_compare, range_check_type): Declare.
22892 * fold-const.c (range_check_type): New function.
22893 (build_range_check): Use range_check_type.
22894 (fold_div_compare): No longer static, rewritten into
22895 a match.pd helper function.
22896 (fold_comparison): Don't call fold_div_compare here.
22897 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22898 as helper function.
22900 2017-07-19 Nathan Sidwell <nathan@acm.org>
22902 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22903 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22904 * tree.c (find_decls_types_r, verify_type): Use
22905 TYPE_{MIN,MAX}_VALUE_RAW.
22906 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22907 (hash_tree): Likewise.
22908 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22910 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22913 2017-07-18 Tom de Vries <tom@codesourcery.com>
22915 PR middle-end/81464
22916 * omp-expand.c (expand_omp_for_static_chunk): Handle
22917 equal-argument loop exit phi.
22919 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22922 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22923 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22924 operand 2 predicate.
22925 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22926 operand 2 predicate.
22927 (ror,rol -> rorx splitters): Use const_int_operand as
22928 operand 2 predicate.
22930 2017-06-18 Richard Biener <rguenther@suse.de>
22932 PR tree-optimization/81410
22933 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22934 the gap in the ! slp_perm SLP case after each group.
22936 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22938 PR middle-end/81463
22939 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22942 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22944 PR middle-end/81462
22945 * predict.c (set_even_probabilities): Cleanup; do not affect
22946 probabilities that are already known.
22947 (combine_predictions_for_bb): Call even when count is set.
22949 2017-07-18 Nathan Sidwell <nathan@acm.org>
22951 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22954 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22957 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22958 optimization for loop niter analysis.
22960 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22963 * config/avr/avr.c (avr_optimize_casesi): Don't use
22964 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22966 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22968 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22969 body_cost_vec from _vect_peel_extended_info.
22970 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22971 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22974 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22976 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22978 2017-07-18 Richard Biener <rguenther@suse.de>
22980 PR tree-optimization/80620
22981 PR tree-optimization/81403
22982 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22983 info when re-using a VN table entry.
22985 2017-07-18 Richard Biener <rguenther@suse.de>
22987 PR tree-optimization/81418
22988 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22989 vectype_in. Verify that with lane-reducing reduction operations
22990 we have a single def-use cycle.
22992 2017-07-17 Carl Love <cel@us.ibm.com>
22994 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22997 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22998 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22999 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23000 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23001 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23002 VMULOSW): New enum "unspec" values.
23003 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23004 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23005 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23006 altivec_vmulosw): New patterns.
23007 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23008 VMULOSW): Add definitions.
23010 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23012 * config/alpha/alpha.c: Include predict.h.
23014 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23016 * tree-vrp.c (compare_assert_loc): Fix comparison function
23017 to return predictable results.
23019 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23021 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23023 (subdi3): Likewise.
23024 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23025 * doc/invoke.texi (mexpand-adddi): Update text.
23027 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23029 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23030 that also clobbers the CC register. The old expand code is moved
23032 (*arc_clzsi2): ... here.
23033 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23034 the CC register. The old expand code is moved to ...
23035 (arc_ctzsi2): ... here.
23037 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23039 * config/arc/arc.opt (mindexed-loads): Use initial value
23040 TARGET_INDEXED_LOADS_DEFAULT.
23041 (mauto-modify-reg): Use initial value
23042 TARGET_AUTO_MODIFY_REG_DEFAULT.
23043 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23044 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23045 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23046 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23048 2017-07-17 Martin Liska <mliska@suse.cz>
23051 * opts.c (finish_options): Do not allow -fgnu-tm
23052 w/ -fsanitize={kernel-,}address. Say sorry.
23054 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23057 * tree-loop-distribution.c (classify_partition): Only assert on
23058 numer of iterations.
23059 (merge_dep_scc_partitions): Delete prameter. Update function call.
23060 (distribute_loop): Remove code handling loop with unknown niters.
23061 (pass_loop_distribution::execute): Skip loop with unknown niters.
23063 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23066 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23067 function sort_partitions_by_post_order.
23069 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23071 PR tree-optimization/81374
23072 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23073 the max index of basic blocks, rather than number of basic blocks.
23075 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23077 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23079 (arc_legitimate_pic_operand_p): Likewise.
23080 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23082 (arc_needs_pcl_p): Likewise.
23083 (arc_legitimate_pc_offset_p): Likewise.
23084 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23085 function is also used in constrains.md.
23086 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23087 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23088 PLUS. Only return true/false in known cases, otherwise assert.
23089 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23090 is already called in arc_legitimate_constant_p.
23091 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23093 (LEGITIMATE_PIC_OPERAND_P): Use
23094 arc_raw_symbolic_reference_mentioned_p function.
23095 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23100 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23101 Andrew Burgess <andrew.burgess@embecosm.com>
23103 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23104 (arc_return_address_register): New function.
23105 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23106 (arc_handle_fndecl_attribute): Add naked attribute.
23107 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23108 (TARGET_WARN_FUNC_RETURN): Likewise.
23109 (arc_allocate_stack_slots_for_args): New function.
23110 (arc_warn_func_return): Likewise.
23111 (machine_function): Change type fn_type.
23112 (arc_compute_function_type): Consider new naked function type,
23113 change function return type.
23114 (arc_must_save_register): Adapt to handle new
23115 arc_compute_function_type's return type.
23116 (arc_expand_prologue): Likewise.
23117 (arc_expand_epilogue): Likewise.
23118 (arc_return_address_regs): Delete.
23119 (arc_return_address_register): New function.
23120 (arc_epilogue_uses): Use above function.
23121 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23122 (arc_function_type): Change encoding, add naked type.
23123 (ARC_INTERRUPT_P): Change to handle the new encoding.
23124 (ARC_FAST_INTERRUPT_P): Likewise.
23125 (ARC_NORMAL_P): Define.
23126 (ARC_NAKED_P): Likewise.
23127 (arc_compute_function_type): Delete prototype.
23128 * config/arc/arc.md (in_ret_delay_slot): Use
23129 arc_return_address_register function.
23130 (simple_return): Likewise.
23131 (p_return_i): Likewise.
23133 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23135 PR tree-optimization/81428
23136 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23137 can't be built for those types.
23139 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23141 Remove stuff dead since r239246.
23143 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23144 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23145 (avr_inform_devices): Remove dead stuff.
23147 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23149 * config/arm/arm_neon.h: Fix softp typo.
23151 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23153 PR tree-optimization/81365
23154 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23155 aggregate moves onto bb predecessor edges, make sure there are no
23156 loads that could alias the lhs in between the start of bb and the
23159 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23162 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23163 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23164 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23166 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23168 PR tree-optimization/81396
23169 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23170 (init_symbolic_number): Initialize it to 1.
23171 (perform_symbolic_merge): Add n_ops from both operands into the new
23173 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23174 without base_addr as useless if they need more than one operation.
23175 (bswap_replace): Handle !bswap case for NULL base_addr.
23177 2017-07-17 Tom de Vries <tom@codesourcery.com>
23180 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23183 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23185 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23186 conditional builtin define __FIX_LEON3FT_B2BST.
23188 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23190 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23191 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23194 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23196 PR rtl-optimization/81424
23197 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23198 to remove potential trapping from operands if -fnon-call-exceptions.
23200 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23202 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23203 profile_proability for scalling.
23204 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23206 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23208 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23210 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23212 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23213 fixpoint arithmetics.
23215 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23217 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23218 fixpoint arithmetics.
23220 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23222 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23223 fixpoint arithmetics.
23225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23227 * profile-count.h (profile_probability::from_reg_br_prob_note,
23228 profile_probability::to_reg_br_prob_note): New functions.
23229 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23230 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23231 * predict.c (probability_reliable_p): Update.
23232 (edge_probability_reliable_p): Update.
23233 (br_prob_note_reliable_p): Update.
23234 (invert_br_probabilities): Update.
23235 (add_reg_br_prob_note): New function.
23236 (combine_predictions_for_insn): Update.
23237 * asan.c (asan_clear_shadow): Update.
23238 * cfgbuild.c (compute_outgoing_frequencies): Update.
23239 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23240 (update_br_prob_note): Update.
23241 (rtl_verify_edges): Update.
23242 (purge_dead_edges): Update.
23243 (fixup_reorder_chain): Update.
23244 * emit-rtl.c (try_split): Update.
23245 * ifcvt.c (cond_exec_process_insns): Update.
23246 (cond_exec_process_if_block): Update.
23247 (dead_or_predicable): Update.
23248 * internal-fn.c (expand_addsub_overflow): Update.
23249 (expand_neg_overflow): Update.
23250 (expand_mul_overflow): Update.
23251 * loop-doloop.c (doloop_modify): Update.
23252 * loop-unroll.c (compare_and_jump_seq): Update.
23253 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23254 * predict.h: Update.
23255 * reorg.c (mostly_true_jump): Update.
23257 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23258 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23259 * config/arc/arc.c: (emit_unlikely_jump): Update.
23260 * config/arm/arm.c: (emit_unlikely_jump): Update.
23261 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23262 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23263 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23264 (ix86_print_operand): Update.
23265 (ix86_split_fp_branch): Update.
23266 (predict_jump): Update.
23267 * config/ia64/ia64.c (ia64_print_operand): Update.
23268 * config/mmix/mmix.c (mmix_print_operand): Update.
23269 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23270 (rs6000_expand_split_stack_prologue): Update.
23271 * config/rs6000/rs6000.c: Update.
23272 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23273 (s390_expand_vec_movstr): Update.
23274 (s390_expand_cs_tdsi): Update.
23275 (s390_expand_split_stack_prologue): Update.
23276 * config/sh/sh.c (sh_print_operand): Update.
23277 (expand_cbranchsi4): Update.
23278 (expand_cbranchdi4): Update.
23279 * config/sparc/sparc.c (output_v9branch): Update.
23280 * config/spu/spu.c (get_branch_target): Update.
23281 (ea_load_store_inline): Update.
23282 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23283 * config/tilepro/tilepro.c: Update.
23285 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23287 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23288 (gimplify_save_expr): Likewise.
23290 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23292 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23294 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23296 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23298 * timevar.def (TV_IPA_FNSUMMARY): Define.
23300 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23302 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23303 to back store errata sensitive sequence from being generated.
23304 (sqrtdf2_fix): Likewise.
23306 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23308 * tree-ssa-threadupdate.c (compute_path_counts,
23309 update_joiner_offpath_counts): Use profile_probability.
23311 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23314 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23316 * config/arm/arm-c.c (arm_cpu_builtins): Define
23317 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23319 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23322 array entries to represent __ieee128 versions of the
23323 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23324 scalar_extract_sig, and scalar_insert_exp built-in functions.
23325 (altivec_resolve_overloaded_builtin): Add special case handling
23326 for the __builtin_scalar_insert_exp function, as represented by
23327 the P9V_BUILTIN_VEC_VSIEDP constant.
23328 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23329 exponent support for __ieee128 argument.
23330 (VSESQP): Add scalar extract signature support for __ieee128
23332 (VSTDCNQP): Add scalar test negative support for __ieee128
23334 (VSIEQP): Add scalar insert exponent support for __int128 argument
23335 with __ieee128 result.
23336 (VSIEQPF): Add scalar insert exponent support for __ieee128
23337 argument with __ieee128 result.
23338 (VSTDCQP): Add scalar test data class support for __ieee128
23340 (VSTDCNQP): Add overload support for scalar test negative with
23341 __ieee128 argument.
23342 (VSTDCQP): Add overload support for scalar test data class
23343 __ieee128 argument.
23344 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23345 UNSPEC_VSX_SXSIGDP.
23346 (UNSPEC_VSX_SIEXPQP): New constant.
23347 (xsxexpqp): New insn for VSX scalar extract exponent quad
23349 (xsxsigqp): New insn for VSX scalar extract significand quad
23351 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23352 precision with floating point argument.
23353 (xststdcqp): New expand for VSX scalar test data class quad
23355 (xststdcnegqp): New expand for VSX scalar test negative quad
23357 (xststdcqp): New insn to match expansions for VSX scalar test data
23358 class quad precision and VSX scalar test negative quad precision.
23359 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23360 special case operand checking to enforce that second operand of
23361 VSX scalar test data class with quad precision argument is a 7-bit
23363 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23364 prototypes and descriptions of __ieee128 versions of
23365 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23366 scalar_test_data_class, and scalar_test_neg built-in functions.
23368 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23370 PR tree-optimization/81162
23371 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23372 replace a negate with an add.
23374 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23376 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23378 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23380 * config/arm/arm-c.c (arm_cpu_builtins): Define
23381 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23383 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23385 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23386 (armv8-r): Set ARM Cortex-R52 as default CPU.
23387 * config/arm/arm-tables.opt: Regenerate.
23388 * config/arm/arm-tune.md: Regenerate.
23389 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23391 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23392 extension for -mcpu=cortex-r52.
23394 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23396 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23397 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23398 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23399 (fp-armv8): Define it as FP_ARMv8 only.
23400 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23401 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23403 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23404 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23405 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23406 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23407 than TARGET_FPU_ARMV8.
23408 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23409 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23410 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23412 * config/arm/neon.md (neon_vrint): Likewise.
23413 (neon_vcvt): Likewise.
23414 (neon_<fmaxmin_op><mode>): Likewise.
23415 (<fmaxmin><mode>3): Likewise.
23416 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23417 * config/arm/predicates.md (arm_cond_move_operator): Check against
23418 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23420 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23422 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23423 to top of function.
23425 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23427 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23428 loop in comment with memset.
23430 2017-07-14 Martin Liska <mliska@suse.cz>
23432 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23433 * dwarf2out.c (is_java): Remove the function.
23434 (output_pubname): Remove usage of the function.
23435 (lower_bound_default): Remove usage of DW_LANG_Java.
23436 (gen_compile_unit_die): Likewise.
23437 * gcc.c: Remove compiler defaults for .java and .zip files.
23438 * gimple-expr.c (remove_suffix): Change as there's no longer
23439 extension than 4-letter one.
23440 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23441 (gimplify_save_expr): Likewise.
23442 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23443 as it's possible even for other languages than Java.
23444 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23445 * lto-opts.c (lto_write_options): Remove reference to Java.
23446 * opts.c (strip_off_ending): Update file extension handling.
23447 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23448 * tree-eh.c (lower_resx): Likewise.
23449 * tree.c (free_lang_data_in_type): Remove dead code.
23450 (find_decls_types_r): Likewise.
23451 (build_common_builtin_nodes): Remove Java from a comment.
23452 (verify_type): Remove dead code.
23453 * varasm.c (assemble_external): Remove Java from a comment.
23455 2017-07-14 Martin Liska <mliska@suse.cz>
23457 * opts.c (finish_options): Add quotes.
23458 (common_handle_option): Likewise.
23460 2017-07-14 Martin Liska <mliska@suse.cz>
23462 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23463 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23464 Remove N_SO_PASCAL.
23465 * dwarf2out.c (lower_bound_default): Do not handle
23467 (gen_compile_unit_die): Likewise.
23468 * gcc.c: Remove default extension binding for GNU Pascal.
23469 * stmt.c: Remove Pascal language from a comment.
23470 * xcoffout.c: Likewise.
23472 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23475 * diagnostic-show-locus.c (fixit_cmp): New function.
23476 (layout::layout): Sort m_fixit_hints.
23477 (column_range::column_range): Assert that the values are valid.
23478 (struct char_span): New struct.
23479 (correction::overwrite): New method.
23480 (struct source_line): New struct.
23481 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23482 calls in terms of classes source_line and char_span, and
23483 correction::overwrite.
23484 (selftest::test_overlapped_fixit_printing_2): New function.
23485 (selftest::diagnostic_show_locus_c_tests): Call it.
23487 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23490 early if there is no lhs.
23492 2017-07-13 Martin Liska <mliska@suse.cz>
23494 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23495 (gen_reference_type_die): Likewise.
23496 * stor-layout.c: Remove Pascal-related comment.
23498 2017-07-13 Martin Liska <mliska@suse.cz>
23500 * opts.c (finish_options): Add quotes to error messages.
23501 (parse_sanitizer_options): Likewise.
23503 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23505 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23507 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23509 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23511 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23513 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23515 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23517 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23520 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23521 provides the hardware capability bits, define the macro
23522 __BUILTIN_CPU_SUPPORTS__.
23523 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23524 if GLIBC does not provide the hardware capability bits. Add a
23525 gcc_unreachable call if the built-in cpu function is neither
23526 __builtin_cpu_is nor __builtin_cpu_supports.
23527 (rs6000_get_function_versions_dispatcher): Change the warning
23528 that an old GLIBC is used which does not export the capability
23529 bits to be an error.
23530 * doc/extend.texi (target_clones attribute): Document the
23531 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23532 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23533 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23534 the macros defined by GCC if the newer GLIBC is available.
23536 2017-07-12 Jeff Law <law@redhat.com>
23538 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23539 remaining includes slightly.
23540 * config/riscv/riscv-builtins.c: Include profile-count.h.
23542 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23545 * config/avr/avr.c (avr_set_current_function): In diagnostic
23546 messages: Quote keywords and (parts of) identifiers.
23547 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23550 2017-07-12 Carl Love <cel@us.ibm.com>
23552 * config/rs6000/rs6000-c.c: Add support for built-in functions
23553 vector bool char vec_revb (vector bool char);
23554 vector bool short vec_revb (vector short char);
23555 vector bool int vec_revb (vector bool int);
23556 vector bool long long vec_revb (vector bool long long);
23557 * doc/extend.texi: Update the built-in documentation file for the
23558 new built-in functions.
23560 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23562 * config/s390/s390.md: Remove movcc splitter.
23564 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23566 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23567 load/store on condition.
23569 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23572 * config/avr/avr.c (avr_encode_section_info)
23573 [progmem && !TREE_READONLY]: Error if progmem object needs
23576 2017-07-11 Michael Collison <michael.collison@arm.com>
23578 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23581 2017-07-11 Carl Love <cel@us.ibm.com>
23583 * config/rs6000/rs6000-c.c: Add support for builtins
23584 vector unsigned int vec_parity_lsbb (vector signed int);
23585 vector unsigned int vec_parity_lsbb (vector unsigned int);
23586 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23587 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23588 vector unsigned long long vec_parity_lsbb (vector signed long long);
23589 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23590 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23591 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23592 * doc/extend.texi: Update the built-in documentation file for the
23593 new built-in functions.
23595 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23597 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23598 (layout::m_primary_loc): New field.
23599 (layout::layout): Initialize new field. Move location filtering
23600 logic from here to...
23601 (layout::maybe_add_location_range): ...this new method. Add
23602 support for filtering to just the lines already specified by other
23604 (layout::will_show_line_p): New method.
23605 (gcc_rich_location::add_location_if_nearby): New method.
23606 (selftest::test_add_location_if_nearby): New test function.
23607 (selftest::diagnostic_show_locus_c_tests): Call it.
23608 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23611 2017-07-11 Tom de Vries <tom@codesourcery.com>
23613 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23614 (bb_first_real_insn): New function.
23615 (nvptx_single): Add extra initialization of broadcasted condition
23618 2017-07-11 Nathan Sidwell <nathan@acm.org>
23620 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23622 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23624 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23625 Binutils doc as TEXI will mess them up.
23626 * doc/invoke.texi (AVR Options): Same here.
23628 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23630 * config/sparc/sparc.opt (mfix-ut700): New option.
23631 (mfix-gr712rc): Likewise.
23632 (sparc_fix_b2bst): New variable.
23633 * doc/invoke.texi (SPARC options): Document them.
23634 (ARM options): Fix warnings.
23635 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23636 instructions to prevent sequences that can trigger the store-store
23637 errata for certain LEON3FT processors.
23638 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23639 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23640 * config/sparc/sparc.md (fix_b2bst): New attribute.
23641 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23643 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23646 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23648 (*rsqrtsf2_sse): Ditto.
23650 (div<mode>3): Macroize insn from divdf3 and divsf3
23651 using MODEF mode iterator.
23653 2017-07-10 Martin Sebor <msebor@redhat.com>
23655 PR tree-optimization/80397
23656 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23657 instead of testing for equality to INTEGER_TYPE.
23659 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23661 * config.gcc: Remove uclibc from arc target spec.
23663 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23665 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23667 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23670 * lto-wrapper.c (remove_option): New function.
23671 (merge_and_complain): Merge PIC/PIE options more realistically.
23673 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23675 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23679 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23680 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23681 * config.in: Regenerate.
23682 * configure: Regenerate.
23683 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23684 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23685 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23686 (TARGET_GASISR_PROLOGUES): ...target mask.
23687 * common/config/avr/avr-common.c
23688 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23689 Set -mgas-isr-prologues.
23690 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23691 INSERT_PASS_BEFORE for it.
23692 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23693 * config/avr/avr.c (avr_option_override)
23694 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23695 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23696 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23697 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23698 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23700 (make_avr_pass_pre_proep): New function.
23701 (emit_push_sfr) <treg>: Add argument to function and use it
23702 instead of TMP_REG.
23703 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23704 and set machine->gasisr.yes.
23705 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23706 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23707 __gcc_isr.n_pushed to .L__stack_usage.
23708 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23709 (avr_asm_final_postscan_insn): ...this new static function.
23710 * config/avr/avr.h (machine_function)
23711 <is_no_gccisr, use_L__stack_usage>: New fields.
23712 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23713 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23714 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23715 (gasisr, *gasisr): New expander and insn.
23716 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23717 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23718 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23720 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23722 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23725 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23727 Move jump-tables out of .text again.
23730 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23731 (ASM_OUTPUT_ADDR_VEC): New function.
23732 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23733 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23734 INSN_ADDRESSes as asm comment.
23735 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23736 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23737 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23738 * config/avr/avr.md (*tablejump): Adjust comment.
23739 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23740 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23742 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23743 (avr_output_addr_vec): New proto.
23744 (avr_log_t) <insn_addresses>: New field.
23746 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23749 * config/i386/i386.c (ix86_function_arg_advance): Set
23750 outgoing_args_on_stack to true if there are outgoing arguments
23752 (ix86_function_arg): Likewise.
23753 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23754 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23755 * config/i386/i386.h (machine_function): Add
23756 outgoing_args_on_stack.
23758 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23760 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23761 supporting pthreds.
23762 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23764 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23766 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23767 (REAL_H): Remove $(MACHMODE_H).
23768 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23770 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23771 $(MACHMODE_H) and double-int.h.
23772 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23774 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23777 2017-07-07 Andrew Pinski <apinski@cavium.com>
23779 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23780 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23782 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23784 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23785 Add warning if GCC was not configured to link against a GLIBC that
23786 exports the hardware capability bits.
23787 (make_resolver_func): Make resolver function private and not a
23788 COMDAT function. Create the name with clone_function_name instead
23789 of make_unique_name.
23792 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23793 correct operand in doing the split.
23795 2017-07-07 Carl Love <cel@us.ibm.com>
23797 * config/rs6000/rs6000-c: Add support for built-in function
23798 vector unsigned short vec_pack_to_short_fp32 (vector float,
23800 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23801 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23802 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23803 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23804 (convert_4f32_8i16): Add define_expand.
23805 * doc/extend.texi: Update the built-in documentation file for the
23806 new built-in function.
23808 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23810 * config/sparc/m8.md: New file.
23811 * config/sparc/sparc.md: Include m8.md.
23813 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23815 * config/sparc/sparc.opt: New option -mvis4b.
23816 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23817 (sparc_option_override): Handle VIS4B.
23818 (enum sparc_builtins): Define
23819 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23820 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23821 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23822 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23823 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23824 (check_constant_argument): New function.
23825 (sparc_vis_init_builtins): Define builtins
23826 __builtin_vis_dictunpack{8,16,32},
23827 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23828 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23829 __builtin_vis_fpcmpde{8,16,32}shl and
23830 __builtin_vis_fpcmpur{8,16,32}shl.
23831 (sparc_expand_builtin): Check that the constant operands to
23832 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23833 constant and in range.
23834 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23836 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23837 (SPARC_IMM5_P): Likewise.
23838 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23839 (enabled): Handle vis4b.
23840 (UNSPEC_DICTUNPACK): New unspec.
23841 (UNSPEC_FPCMPSHL): Likewise.
23842 (UNSPEC_FPUCMPSHL): Likewise.
23843 (UNSPEC_FPCMPDESHL): Likewise.
23844 (UNSPEC_FPCMPURSHL): Likewise.
23845 (cpu_feature): New CPU feature `vis4b'.
23846 (dictunpack{8,16,32}): New insns.
23847 (FPCSMODE): New mode iterator.
23848 (fpcscond): New code iterator.
23849 (fpcsucond): Likewise.
23850 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23851 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23852 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23853 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23854 * config/sparc/constraints.md: Define constraints `q' for unsigned
23855 2-bit integer constants and `t' for unsigned 5-bit integer
23857 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23859 (imm5_operand_dictunpack16): Likewise.
23860 (imm5_operand_dictunpack32): Likewise.
23861 (imm2_operand): Likewise.
23862 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23863 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23864 ditunpack* and fpcmp*shl builtins.
23866 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23868 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23869 * config.in: Add HAVE_AS_SPARC6 define.
23870 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23872 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23874 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23875 (CPP_CPU_SPEC): Handle m8.
23876 (ASM_CPU_SPEC): Likewise.
23877 * config/sparc/sparc-opts.h (enum processor_type): Add
23879 * config/sparc/sparc.c (m8_costs): New struct.
23880 (sparc_option_override): Handle TARGET_CPU_m8.
23881 (sparc32_initialize_trampoline): Likewise.
23882 (sparc64_initialize_trampoline): Likewise.
23883 (sparc_issue_rate): Likewise.
23884 (sparc_register_move_cost): Likewise.
23885 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23886 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23887 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23888 (CPP_CPU_SPEC): Handle M8.
23889 (ASM_CPU_SPEC): Likewise.
23890 (AS_M8_FLAG): Define.
23891 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23892 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23893 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23895 * configure: Regenerate.
23896 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23899 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23901 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23903 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23904 ("*movdi_insn_sp32"): Do not set v3pipe.
23905 ("*movsi_insn"): Likewise.
23906 ("*movdi_insn_sp64"): Likewise.
23907 ("*movsf_insn"): Likewise.
23908 ("*movdf_insn_sp32"): Likewise.
23909 ("*movdf_insn_sp64"): Likewise.
23910 ("*zero_extendsidi2_insn_sp64"): Likewise.
23911 ("*sign_extendsidi2_insn"): Likewise.
23912 ("*mov<VM32:mode>_insn"): Likewise.
23913 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23914 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23915 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23916 ("<vlop:code><VL:mode>3"): Likewise.
23917 ("*not_<vlop:code><VL:mode>3"): Likewise.
23918 ("*nand<VL:mode>_vis"): Likewise.
23919 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23920 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23921 ("one_cmpl<VL:mode>2"): Likewise.
23922 ("faligndata<VM64:mode>_vis"): Likewise.
23923 ("alignaddrsi_vis"): Likewise.
23924 ("alignaddrdi_vis"): Likweise.
23925 ("alignaddrlsi_vis"): Likewise.
23926 ("alignaddrldi_vis"): Likewise.
23927 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23928 ("bmaskdi_vis"): Likewise.
23929 ("bmasksi_vis"): Likewise.
23930 ("bshuffle<VM64:mode>_vis"): Likewise.
23931 ("cmask8<P:mode>_vis"): Likewise.
23932 ("cmask16<P:mode>_vis"): Likewise.
23933 ("cmask32<P:mode>_vis"): Likewise.
23934 ("pdistn<P:mode>_vis"): Likewise.
23935 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23937 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23939 * config/sparc/sparc.md ("subtype"): New insn attribute.
23940 ("*wrgsr_sp64"): Set insn subtype.
23941 ("*rdgsr_sp64"): Likewise.
23942 ("alignaddrsi_vis"): Likewise.
23943 ("alignaddrdi_vis"): Likewise.
23944 ("alignaddrlsi_vis"): Likewise.
23945 ("alignaddrldi_vis"): Likewise.
23946 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23947 ("fexpand_vis"): Likewise.
23948 ("fpmerge_vis"): Likewise.
23949 ("faligndata<VM64:mode>_vis"): Likewise.
23950 ("bshuffle<VM64:mode>_vis"): Likewise.
23951 ("cmask8<P:mode>_vis"): Likewise.
23952 ("cmask16<P:mode>_vis"): Likewise.
23953 ("cmask32<P:mode>_vis"): Likewise.
23954 ("fchksm16_vis"): Likewise.
23955 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23956 ("fmean16_vis"): Likewise.
23957 ("fp<plusminus_insn>64_vis"): Likewise.
23958 ("<plusminus_insn>v8qi3"): Likewise.
23959 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23960 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23961 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23962 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23963 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23964 ("*movqi_insn"): Likewise.
23965 ("*movhi_insn"): Likewise.
23966 ("*movsi_insn"): Likewise.
23967 ("movsi_pic_gotdata_op"): Likewise.
23968 ("*movdi_insn_sp32"): Likewise.
23969 ("*movdi_insn_sp64"): Likewise.
23970 ("movdi_pic_gotdata_op"): Likewise.
23971 ("*movsf_insn"): Likewise.
23972 ("*movdf_insn_sp32"): Likewise.
23973 ("*movdf_insn_sp64"): Likewise.
23974 ("*zero_extendhisi2_insn"): Likewise.
23975 ("*zero_extendqihi2_insn"): Likewise.
23976 ("*zero_extendqisi2_insn"): Likewise.
23977 ("*zero_extendqidi2_insn"): Likewise.
23978 ("*zero_extendhidi2_insn"): Likewise.
23979 ("*zero_extendsidi2_insn_sp64"): Likewise.
23980 ("ldfsr"): Likewise.
23981 ("prefetch_64"): Likewise.
23982 ("prefetch_32"): Likewise.
23983 ("tie_ld32"): Likewise.
23984 ("tie_ld64"): Likewise.
23985 ("*tldo_ldub_sp32"): Likewise.
23986 ("*tldo_ldub1_sp32"): Likewise.
23987 ("*tldo_ldub2_sp32"): Likewise.
23988 ("*tldo_ldub_sp64"): Likewise.
23989 ("*tldo_ldub1_sp64"): Likewise.
23990 ("*tldo_ldub2_sp64"): Likewise.
23991 ("*tldo_ldub3_sp64"): Likewise.
23992 ("*tldo_lduh_sp32"): Likewise.
23993 ("*tldo_lduh1_sp32"): Likewise.
23994 ("*tldo_lduh_sp64"): Likewise.
23995 ("*tldo_lduh1_sp64"): Likewise.
23996 ("*tldo_lduh2_sp64"): Likewise.
23997 ("*tldo_lduw_sp32"): Likewise.
23998 ("*tldo_lduw_sp64"): Likewise.
23999 ("*tldo_lduw1_sp64"): Likewise.
24000 ("*tldo_ldx_sp64"): Likewise.
24001 ("*mov<VM32:mode>_insn"): Likewise.
24002 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24003 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24005 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24007 * config/sparc/sparc.md ("type"): New insn type viscmp.
24008 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24010 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24011 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24012 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24013 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24015 ("n7_vis_logical_11cycle"): Likewise.
24016 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24017 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24018 * config/sparc/niagara.md ("niag_vis"): Likewise.
24019 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24020 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24022 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24024 * config/sparc/sparc.md: New instruction type `bmask'.
24025 (bmaskdi_vis): Use the `bmask' type.
24026 (bmasksi_vis): Likewise.
24027 * config/sparc/ultra3.md (us3_array): Likewise.
24028 * config/sparc/niagara7.md (n7_array): Likewise.
24029 * config/sparc/niagara4.md (n4_array): Likewise.
24030 * config/sparc/niagara2.md (niag2_vis): Likewise.
24031 (niag3_vis): Likewise.
24032 * config/sparc/niagara.md (niag_vis): Likewise.
24034 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24036 * ipa-comdats.c: Remove optimize check from gate.
24037 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24038 for functions not optimized.
24039 (ipa_fn_summary_read): Skip optimize check.
24040 (ipa_fn_summary_write): Likewise.
24041 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24043 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24045 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24046 (check_callers): Give up on caller which is not optimized.
24047 (inline_small_functions): Likewise.
24048 (ipa_inline): Do not give up when not optimizing.
24049 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24050 away unoptimizes cdtors.
24051 (whole_program_function_and_variable_visibility): Do
24052 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24053 * ipa.c (process_references): Do not check optimize.
24054 (symbol_table::remove_unreachable_nodes): Update optimize check.
24055 (set_writeonly_bit): Update optimize check.
24056 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24057 (pass_ipa_single_use::gate): Remove.
24059 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24061 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24062 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24063 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24064 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24065 permute_load, permute_store, adjust_extract, adjust_splat,
24066 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24067 replace_swap_with_copy, dump_swap_insn_table,
24068 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24069 recombine_lvx_pattern, recombine_stvx_pattern,
24070 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24071 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24072 to file rs6000-p8swap.c.
24073 * config/rs6000/rs6000-p8swap.c: New file.
24074 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24075 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24076 and rs6000*-*-* targets.
24078 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24080 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24082 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24084 * lto-wrapper.c (merge_and_complain): Do not merge
24085 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24086 fsigned_zeros, ftrapping_math, fwrapv.
24087 (append_compiler_options): Do not track these options.
24088 (append_linker_options): Likewie
24090 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24092 * cgraphunit.c (cgraph_node::finalize_function): When
24093 !flag_toplevel_reorde set no_reorder flag.
24094 (varpool_node::finalize_decl): Likewise.
24095 (symbol_table::compile): Drop no toplevel reorder path.
24097 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24099 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24100 edges; zero probability is not better than uninitialized.
24102 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24104 * asan.h (asan_sanitize_allocas_p): Declare.
24105 * asan.c (asan_sanitize_allocas_p): New function.
24106 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24107 (handle_builtin_alloca): Likewise.
24108 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24109 if !asan_sanitize_allocas_p.
24110 * params.def (asan-instrument-allocas): Add new option.
24111 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24112 * opts.c (common_handle_option): Disable allocas sanitization for
24115 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24117 * asan.c: Include gimple-fold.h.
24118 (get_last_alloca_addr): New function.
24119 (handle_builtin_stackrestore): Likewise.
24120 (handle_builtin_alloca): Likewise.
24121 (asan_emit_allocas_unpoison): Likewise.
24122 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24123 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24124 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24125 (instrument_builtin_call): Pass gimple iterator to
24126 get_mem_refs_of_builtin_call.
24127 (last_alloca_addr): New global.
24128 * asan.h (asan_emit_allocas_unpoison): Declare.
24129 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24130 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24131 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24132 if function calls alloca.
24133 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24134 * gimple-fold.h (replace_call_with_value): Declare.
24135 * internal-fn.c: Include asan.h.
24136 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24137 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24139 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24141 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24142 (C_SELFTEST_FLAGS): New.
24143 (CPP_SELFTEST_FLAGS): New.
24144 (SELFTEST_DEPS): New, from deps of s-selftest.
24145 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24146 (CPP_SELFTEST_DEPS): New.
24147 (selftest): Add dependency on s-selftest-c++.
24148 (s-selftest): Rename to...
24149 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24150 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24151 than SELFTEST_FLAGS.
24152 (selftest-gdb): Rename to...
24153 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24155 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24156 (selftest-valgrind): Rename to...
24157 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24159 (selftest-valgrind): Reintroduce as an alias for
24160 selftest-c-valgrind.
24161 (s-selftest-c++): New.
24162 (selftest-c++-gdb): New.
24163 (selftest-c++-valgrind): New.
24165 2017-07-06 Olivier Hainque <hainque@adacore.com>
24167 * gcc.c (process_command): When deciding if undefined variables
24168 should be ignored when processing specs, accept "gcc -v" as well.
24170 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24172 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24173 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24175 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24177 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24178 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24179 * config/arm/arm-tables.opt: Regenerate.
24180 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24182 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24184 2017-07-06 Carl Love <cel@us.ibm.com>
24186 * ChangeLog: Clean up from mid air collision
24188 2017-07-06 Carl Love <cel@us.ibm.com>
24190 * config/rs6000/rs6000-c.c: Add support for built-in functions
24191 vector signed int vec_subc (vector signed int, vector signed int);
24192 vector signed __int128 vec_subc (vector signed __int128,
24193 vector signed __int128);
24194 vector unsigned __int128 vec_subc (vector unsigned __int128,
24195 vector unsigned __int128);
24196 vector signed int vec_sube (vector signed int, vector signed int,
24197 vector signed int);
24198 vector unsigned int vec_sube (vector unsigned int,
24199 vector unsigned int,
24200 vector unsigned int);
24201 vector signed __int128 vec_sube (vector signed __int128,
24202 vector signed __int128,
24203 vector signed__int128);
24204 vector unsigned __int128 vec_sube (vector unsigned __int128,
24205 vector unsigned __int128,
24206 vector unsigned __int128);
24207 vector signed int vec_subec (vector signed int, vector signed int,
24208 vector signed int);
24209 vector unsigned int vec_subec (vector unsigned int,
24210 vector unsigned int,
24211 vector unsigned int);
24212 vector signed __int128 vec_subec (vector signed __int128,
24213 vector signed __int128,
24214 vector signed__int128);
24215 vector unsigned __int128 vec_subec (vector unsigned __int128,
24216 vector unsigned __int128,
24217 vector unsigned __int128);
24218 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24219 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24220 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24221 BU_ALTIVEC_OVERLOAD_X definitions.
24222 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24223 * doc/extend.texi: Update the built-in documentation file for the new
24224 built-in functions.
24226 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24229 * diagnostic-show-locus.c (layout::layout): Use start and finish
24230 spelling location for the start and finish of each range.
24231 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24232 Add unused aspect param.
24233 * input.c (expand_location_1): Add "aspect" param, and use it
24234 to access the correct part of the location.
24235 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24237 (expand_location_to_spelling_point): Likewise.
24238 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24239 param, and pass it to expand_location_1.
24241 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24243 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24244 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24245 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24246 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24247 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24248 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24249 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24250 _mm_maskz_getmant_ss): New intrinsics.
24251 (__builtin_ia32_getexpss128_mask): Changed to ...
24252 __builtin_ia32_getexpss128_round ... this.
24253 (__builtin_ia32_getexpsd128_mask): Changed to ...
24254 __builtin_ia32_getexpsd128_round ... this.
24255 * config/i386/i386-builtin-types.def
24256 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24257 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24258 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24259 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24260 __builtin_ia32_getmantss_mask_round): New builtins.
24261 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24262 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24263 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24264 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24265 * config/i386/sse.md
24266 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24267 avx512f_sgetexp<mode><mask_scalar_name>
24268 <round_saeonly_scalar_name> ... this.
24269 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24270 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24271 vgetexp<ssescalarmodesuffix>
24272 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24273 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24274 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24275 avx512f_vgetmant<mode><mask_scalar_name>
24276 <round_saeonly_scalar_name> ... this.
24277 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24278 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24279 vgetmant<ssescalarmodesuffix>
24280 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24281 %0<mask_scalar_operand4>, %1, %2
24282 <round_saeonly_scalar_mask_op4>, %3} ... this.
24283 * config/i386/subst.md (mask_scalar_operand4,
24284 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24285 round_saeonly_scalar_nimm_predicate): New subst attributes.
24287 2017-07-06 Julia Koval <julia.koval@intel.com>
24289 * config/i386/i386.c (ix86_erase_embedded_rounding):
24290 Remove code for old rounding pattern.
24292 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24294 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24296 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24298 * doc/sourcebuild.texi (Test Directives, Variants of
24299 dg-require-support): Add documentation for dg-require-stack-check.
24301 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24303 * config/i386/subst.md (mask_scalar, round_scalar,
24304 round_saeonly_scalar): New meta-templates.
24305 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24306 round_scalar_mask_operand3, round_scalar_mask_op3,
24307 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24308 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24309 round_saeonly_scalar_constraint,
24310 round_saeonly_scalar_prefix): New subst attribute.
24311 * config/i386/sse.md
24312 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24313 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24314 <round_scalar_name> ... this.
24315 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24316 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24317 <round_scalar_name> ... this.
24318 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24319 <sse>_vm<code><mode>3<mask_scalar_name>
24320 <round_saeonly_scalar_name> ... this.
24321 (v<plusminus_mnemonic><ssescalarmodesuffix>
24322 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24323 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24324 v<plusminus_mnemonic><ssescalarmodesuffix>
24325 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24326 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24327 (v<multdiv_mnemonic><ssescalarmodesuffix>
24328 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24329 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24330 v<multdiv_mnemonic><ssescalarmodesuffix>
24331 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24332 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24333 (v<maxmin_float><ssescalarmodesuffix>
24334 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24335 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24336 v<maxmin_float><ssescalarmodesuffix>
24337 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24338 %0<mask_scalar_operand3>, %1, %<iptr>2
24339 <round_saeonly_scalar_mask_op3>} ... this.
24341 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24343 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24344 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24346 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24347 Alan Hayward <alan.hayward@arm.com>
24348 David Sherwood <david.sherwood@arm.com>
24350 * combine.c (simplify_if_then_else): Remove "enum" before
24352 * compare-elim.c (can_eliminate_compare): Likewise.
24353 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24355 (aarch64_lookup_simd_builtin_type): Likewise.
24356 (aarch64_simd_builtin_type): Likewise.
24357 (aarch64_init_simd_builtin_types): Likewise.
24358 (aarch64_simd_expand_args): Likewise.
24359 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24361 (aarch64_reverse_mask): Likewise.
24362 (aarch64_simd_emit_reg_reg_move): Likewise.
24363 (aarch64_gen_adjusted_ldpstp): Likewise.
24364 (aarch64_ccmp_mode_to_code): Likewise.
24365 (aarch64_operands_ok_for_ldpstp): Likewise.
24366 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24367 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24369 (aarch64_min_divisions_for_recip_mul): Likewise.
24370 (aarch64_reassociation_width): Likewise.
24371 (aarch64_get_condition_code_1): Likewise.
24372 (aarch64_simd_emit_reg_reg_move): Likewise.
24373 (aarch64_simd_attr_length_rglist): Likewise.
24374 (aarch64_reverse_mask): Likewise.
24375 (aarch64_operands_ok_for_ldpstp): Likewise.
24376 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24377 (aarch64_gen_adjusted_ldpstp): Likewise.
24378 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24380 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24381 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24382 (arm_lookup_simd_builtin_type): Likewise.
24383 (arm_simd_builtin_type): Likewise.
24384 (arm_init_simd_builtin_types): Likewise.
24385 (arm_expand_builtin_args): Likewise.
24386 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24387 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24388 (ft32_setup_incoming_varargs): Likewise.
24389 (ft32_function_arg): Likewise.
24390 (ft32_function_arg_advance): Likewise.
24391 (ft32_pass_by_reference): Likewise.
24392 (ft32_arg_partial_bytes): Likewise.
24393 (ft32_valid_pointer_mode): Likewise.
24394 (ft32_addr_space_pointer_mode): Likewise.
24395 (ft32_addr_space_legitimate_address_p): Likewise.
24396 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24398 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24399 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24400 (iamcu_alignment): Likewise.
24401 (canonicalize_vector_int_perm): Likewise.
24402 (ix86_noce_conversion_profitable_p): Likewise.
24403 (ix86_mpx_bound_mode): Likewise.
24404 (ix86_operands_ok_for_move_multiple): Likewise.
24405 * config/microblaze/microblaze-protos.h
24406 (microblaze_expand_conditional_branch_reg): Likewise.
24407 * config/microblaze/microblaze.c
24408 (microblaze_expand_conditional_branch_reg): Likewise.
24409 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24411 (rs6000_reassociation_width): Likewise.
24412 (rs6000_invalid_binary_op): Likewise.
24413 (fusion_p9_p): Likewise.
24414 (emit_fusion_p9_load): Likewise.
24415 (emit_fusion_p9_store): Likewise.
24416 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24418 (riscv_hard_regno_mode_ok_p): Likewise.
24419 (riscv_address_insns): Likewise.
24420 (riscv_split_symbol): Likewise.
24421 (riscv_legitimize_move): Likewise.
24422 (riscv_function_value): Likewise.
24423 (riscv_hard_regno_nregs): Likewise.
24424 (riscv_expand_builtin): Likewise.
24425 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24426 (riscv_build_integer): Likewise.
24427 (riscv_split_integer): Likewise.
24428 (riscv_legitimate_constant_p): Likewise.
24429 (riscv_cannot_force_const_mem): Likewise.
24430 (riscv_regno_mode_ok_for_base_p): Likewise.
24431 (riscv_valid_base_register_p): Likewise.
24432 (riscv_valid_offset_p): Likewise.
24433 (riscv_valid_lo_sum_p): Likewise.
24434 (riscv_classify_address): Likewise.
24435 (riscv_legitimate_address_p): Likewise.
24436 (riscv_address_insns): Likewise.
24437 (riscv_load_store_insns): Likewise.
24438 (riscv_force_binary): Likewise.
24439 (riscv_split_symbol): Likewise.
24440 (riscv_force_address): Likewise.
24441 (riscv_legitimize_address): Likewise.
24442 (riscv_move_integer): Likewise.
24443 (riscv_legitimize_const_move): Likewise.
24444 (riscv_legitimize_move): Likewise.
24445 (riscv_address_cost): Likewise.
24446 (riscv_subword): Likewise.
24447 (riscv_output_move): Likewise.
24448 (riscv_canonicalize_int_order_test): Likewise.
24449 (riscv_emit_int_order_test): Likewise.
24450 (riscv_function_arg_boundary): Likewise.
24451 (riscv_pass_mode_in_fpr_p): Likewise.
24452 (riscv_pass_fpr_single): Likewise.
24453 (riscv_pass_fpr_pair): Likewise.
24454 (riscv_get_arg_info): Likewise.
24455 (riscv_function_arg): Likewise.
24456 (riscv_function_arg_advance): Likewise.
24457 (riscv_arg_partial_bytes): Likewise.
24458 (riscv_function_value): Likewise.
24459 (riscv_pass_by_reference): Likewise.
24460 (riscv_setup_incoming_varargs): Likewise.
24461 (riscv_print_operand): Likewise.
24462 (riscv_elf_select_rtx_section): Likewise.
24463 (riscv_save_restore_reg): Likewise.
24464 (riscv_for_each_saved_reg): Likewise.
24465 (riscv_register_move_cost): Likewise.
24466 (riscv_hard_regno_mode_ok_p): Likewise.
24467 (riscv_hard_regno_nregs): Likewise.
24468 (riscv_class_max_nregs): Likewise.
24469 (riscv_memory_move_cost): Likewise.
24470 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24471 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24472 (rl78_addr_space_address_mode): Likewise.
24473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24475 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24476 (rs6000_reassociation_width): Likewise.
24477 (rs6000_invalid_binary_op): Likewise.
24478 (fusion_p9_p): Likewise.
24479 (emit_fusion_p9_load): Likewise.
24480 (emit_fusion_p9_store): Likewise.
24481 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24482 (ok_for_simple_move_operands): Likewise.
24483 (ok_for_simple_move_strict_operands): Likewise.
24484 (ok_for_simple_arith_logic_operands): Likewise.
24485 (visium_legitimize_reload_address): Likewise.
24486 (visium_select_cc_mode): Likewise.
24487 (output_cbranch): Likewise.
24488 (visium_split_double_move): Likewise.
24489 (visium_expand_copysign): Likewise.
24490 (visium_expand_int_cstore): Likewise.
24491 (visium_expand_fp_cstore): Likewise.
24492 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24493 (visium_function_arg): Likewise.
24494 (visium_function_arg_advance): Likewise.
24495 (visium_libcall_value): Likewise.
24496 (visium_setup_incoming_varargs): Likewise.
24497 (visium_legitimate_constant_p): Likewise.
24498 (visium_legitimate_address_p): Likewise.
24499 (visium_legitimize_address): Likewise.
24500 (visium_secondary_reload): Likewise.
24501 (visium_register_move_cost): Likewise.
24502 (visium_memory_move_cost): Likewise.
24503 (prepare_move_operands): Likewise.
24504 (ok_for_simple_move_operands): Likewise.
24505 (ok_for_simple_move_strict_operands): Likewise.
24506 (ok_for_simple_arith_logic_operands): Likewise.
24507 (visium_function_value_1): Likewise.
24508 (rtx_ok_for_offset_p): Likewise.
24509 (visium_legitimize_reload_address): Likewise.
24510 (visium_split_double_move): Likewise.
24511 (visium_expand_copysign): Likewise.
24512 (visium_expand_int_cstore): Likewise.
24513 (visium_expand_fp_cstore): Likewise.
24514 (visium_split_cstore): Likewise.
24515 (visium_select_cc_mode): Likewise.
24516 (visium_split_cbranch): Likewise.
24517 (output_cbranch): Likewise.
24518 (visium_print_operand_address): Likewise.
24519 * expmed.c (flip_storage_order): Likewise.
24520 * expmed.h (emit_cstore): Likewise.
24521 (flip_storage_order): Likewise.
24522 * genrecog.c (validate_pattern): Likewise.
24523 * hsa-gen.c (gen_hsa_addr): Likewise.
24524 * internal-fn.c (expand_arith_overflow): Likewise.
24525 * ira-color.c (allocno_copy_cost_saving): Likewise.
24526 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24527 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24528 (process_invariant_for_inheritance): Likewise.
24529 * lra-eliminations.c (move_plus_up): Likewise.
24530 * omp-low.c (lower_oacc_reductions): Likewise.
24531 * simplify-rtx.c (simplify_subreg): Likewise.
24532 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24533 (TARGET_CHKP_BOUND_MODE): Likewise..
24534 * targhooks.c (default_chkp_bound_mode): Likewise.
24535 (default_setup_incoming_vararg_bounds): Likewise.
24536 * targhooks.h (default_chkp_bound_mode): Likewise.
24537 (default_setup_incoming_vararg_bounds): Likewise.
24538 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24539 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24540 (have_whole_vector_shift): Likewise.
24541 * tree-vect-stmts.c (vectorizable_load): Likewise.
24542 * doc/tm.texi: Regenerate.
24544 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24546 Graceful degrade if Binutils PR21472 is not available.
24549 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24550 .rodata in flash test fails.
24551 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24552 * confgure: Regenerate.
24553 * config.in: Regenerate.
24554 * config/avr/avr.c (avr_asm_named_section)
24555 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24556 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24557 (avr_asm_init_sections): Same.
24559 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24561 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24562 (fma<VH:mode>4_intrinsic): Likewise.
24563 (*fmsub<VCVTF:mode>4): Likewise.
24564 (*fmsub<VH:mode>4_intrinsic): Likewise.
24566 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24569 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24570 Don't depend on "optimize > 0".
24571 (out_movhi_r_mr, out_movqi_mr_r): Same.
24572 (out_movhi_mr_r, out_movqi_r_mr): Same.
24573 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24574 io_address_operand on "optimize > 0".
24576 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24578 * tree-loop-distribution.c: Add general explanantion on the pass.
24579 (generate_loops_for_partition): Mark distributed loop.
24580 (pg_add_dependence_edges): New parameter. Handle alias data
24581 dependence specially and record it in the parameter if asked.
24582 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24583 (init_partition_graph_vertices, add_partition_graph_edge): New.
24584 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24585 (free_partition_graph_vdata, build_partition_graph): New.
24586 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24587 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24588 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24589 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24590 (version_for_distribution_p, finalize_partitions): New.
24591 (distribute_loop): Handle alias data dependence specially. Factor
24592 out loop fusion code as functions and call these functions.
24594 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24596 * tree-loop-distribution.c (classify_partition): New parameter and
24597 better handle reduction statement.
24598 (rdg_build_partitions): Revise comment.
24599 (distribute_loop): Compute statements in all partitions and pass it
24600 to classify_partition.
24602 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24604 * tree-loop-distribution.c (enum partition_type): New.
24605 (struct partition): New field type.
24606 (partition_merge_into): Add parameter. Update partition type.
24607 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24608 (build_rdg_partition_for_vertex): Compute partition type.
24609 (rdg_build_partitions): Dump partition type.
24610 (distribute_loop): Update calls to partition_merge_into.
24612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24614 * tree-loop-distribution.c (struct ddr_hasher): New.
24615 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24617 (classify_partition): Call get_data_dependence.
24618 (pg_add_dependence_edges): Ditto.
24619 (distribute_loop): Release data dependence hash table.
24621 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24623 * tree-loop-distribution.c (ref_base_address): Delete.
24624 (similar_memory_accesses): Rename ...
24625 (share_memory_accesses): ... to this. Check if partitions access
24626 the same memory reference.
24627 (distribute_loop): Call share_memory_accesses.
24629 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24631 * tree-loop-distribution.c (struct partition): New field recording
24632 its data reference.
24633 (partition_alloc, partition_free): Init and release data refs.
24634 (partition_merge_into): Merge data refs.
24635 (build_rdg_partition_for_vertex): Collect data refs for partition.
24636 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24638 (distribute_loop): Remve data refs from vertice data of partition
24641 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24643 * tree-loop-distribution.c (params.h): Include header file.
24644 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24645 (datarefs_vec): New global var.
24646 (create_rdg_vertices): Use datarefs_vec directly.
24647 (free_rdg): Don't free data references.
24648 (build_rdg): Update use. Don't free data references.
24649 (distribute_loop): Compute global variable for data references.
24650 Bail out if there are too many data references.
24652 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24654 * tree-loop-distribution.c (loop_nest): New global var.
24655 (build_rdg): Use loop directly, rather than loop nest.
24656 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24658 (distribute_loop): Compute global variable loop nest. Update use.
24660 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24662 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24663 (partition_merge_into): New parameter. Dump reason for fusion.
24664 (distribute_loop): Update use of partition_merge_into.
24666 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24668 * tree-loop-distribution.c (bb_top_order_index): New.
24669 (bb_top_order_index_size, bb_top_order_cmp): New.
24670 (stmts_from_loop): Use topological order.
24671 (pass_loop_distribution::execute): Compute and release topological
24672 order for basic blocks.
24674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24676 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24679 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24681 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24682 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24683 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24684 * internal-fn.def (LOOP_DIST_ALIAS): New.
24685 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24686 (fold_loop_internal_call): ... this.
24687 (vect_loop_dist_alias_call): New function.
24688 (set_uid_loop_bbs): Call fold_loop_internal_call.
24689 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24692 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24695 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24696 Require dead FLAGS_REG at the beginning of a peephole.
24698 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24701 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24702 arguments in the call to __builtin_ia32_sbb_u32.
24703 (_subborrow_u64): Swap _X and _Y arguments in the call to
24704 __builtin_ia32_sbb_u64.
24706 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24709 * tree-vrp.c (compare_assert_loc): Turn into a function template
24710 with stable template parameter. Only test if a->e is NULL,
24711 !a->e == !b->e has been verified already. Use e == NULL or
24712 e != NULL instead of e or ! e tests. If stable is true, don't use
24713 iterative_hash_expr, on the other side allow a or b or both NULL
24714 and sort the NULLs last.
24715 (process_assert_insertions): Sort using compare_assert_loc<false>
24716 instead of compare_assert_loc, later sort using
24717 compare_assert_loc<true> before calling process_assert_insertions_for
24718 in a loop. Use break instead of continue once seen NULL pointer.
24720 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24722 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24723 Cortex-R7 and Cortex-R8 processors.
24725 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24727 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24728 uninitialized while src is not.
24730 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24732 * common/config/arm/arm-common.c: Adjust include path for
24734 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24735 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24736 (arm-cpu-data.h): Likewise.
24737 (arm-cpu-cdata.h): Likewise.
24738 * config/arm/arm-cpu.h: Delete.
24739 * config/arm/arm-cpu-cdata.h: Delete.
24740 * config/arm/arm-cpu-data.h: Delete.
24742 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24744 * config/arm/arm-cpus.in (cortex-a55): New.
24745 (cortex-a75): Likewise.
24746 (cortex-a75.cortex-a55): Likewise.
24747 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24749 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24750 * config/arm/arm-cpu-cdata.h: Regenerate.
24751 * config/arm/arm-cpu-data.h: Regenerate.
24752 * config/arm/arm-cpu.h: Regenerate.
24753 * config/arm/arm-tables.opt: Regenerate.
24754 * config/arm/arm-tune.md: Regenerate.
24756 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24758 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24760 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24762 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24765 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24767 PR tree-optimization/81292
24768 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24769 full_string_p, also call adjust_related_strinfos if the adjustment
24770 is simple, otherwise invalidate related strinfos.
24772 2017-07-04 Martin Liska <mliska@suse.cz>
24775 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24776 newly created variable as DECL_IGNORED_P.
24778 2017-07-04 Martin Liska <mliska@suse.cz>
24781 * ipa-inline.c (inline_small_functions):
24782 Use xstrdup_for_dump.
24784 2017-07-04 Tom de Vries <tom@codesourcery.com>
24786 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24788 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24791 * config/darwin.c (darwin_function_switched_text_sections):
24794 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24796 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24798 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24800 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24802 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24804 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24805 min_profitable_iters, and th as inclusive lower bounds.
24806 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24807 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24808 for min_profitable_iters and min_profitable_estimate.
24809 (vect_transform_loop): Treat th as an inclusive lower bound.
24810 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24812 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24815 * config/darwin.c (darwin_function_switched_text_sections):
24816 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24817 in two pieces, and suppress the use of buf.
24819 2017-07-03 Nathan Sidwell <nathan@acm.org>
24821 * hash-table.h (hash_table_mod1): Fix indentation.
24823 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24825 PR middle-end/81290
24826 * predict.c (force_edge_cold): Be more careful about propagation
24828 * profile-count.h (profile_probability::guessed,
24829 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24831 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24833 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24835 * doc/invoke.texi (rcpc architecture extension): Document it.
24837 2017-07-03 Richard Biener <rguenther@suse.de>
24839 PR tree-optimization/60510
24840 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24841 the scalar reduction PHI and use it.
24842 (vectorizable_reduction): Properly guard the single_defuse_cycle
24843 path for non-SLP reduction chains where we cannot use it.
24844 Rework reduc_def/index and vector type deduction. Rework
24845 vector operand gathering during reduction op code-gen.
24846 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24847 chains dissolve the chain and leave it to non-SLP reduction
24850 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24852 * tree-data-ref.h (dr_alignment): Declare.
24853 * tree-data-ref.c (dr_alignment): New function.
24854 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24855 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24857 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24859 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24861 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24862 and base_misalignment fields.
24863 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24864 * tree-data-ref.c: Include builtins.h.
24865 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24866 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24867 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24868 * tree-vect-data-refs.c: Include tree-cfg.h.
24869 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24870 fields instead of calculating an alignment here.
24871 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24872 innermost_loop_behavior fields.
24874 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24876 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24878 (DR_STEP_ALIGNMENT): New macro.
24879 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24880 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24881 (create_data_ref): Print it.
24882 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24883 to tell whether the step preserves vector (mis)alignment.
24884 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24885 Move the check for an integer step and generalise to all INTEGER_CST.
24886 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24887 Print the outer step alignment.
24889 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24891 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24892 with offset_alignment.
24893 (DR_ALIGNED_TO): Delete.
24894 (DR_OFFSET_ALIGNMENT): New macro.
24895 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24896 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24897 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24898 (create_data_ref): Likewise.
24899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24900 (vect_analyze_data_refs): Likewise.
24901 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24902 creating dummy innermost behavior.
24904 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24906 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24907 with a "innermost_loop_behavior *" and refeence tree.
24908 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24909 (create_data_ref): Update call accordingly.
24910 * tree-predcom.c (find_looparound_phi): Likewise.
24912 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24914 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24915 fields with dr_wrt_vec_loop.
24916 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24917 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24918 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24919 (vect_dr_behavior): New function.
24920 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24921 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24922 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24923 track whether the step preserves the misalignment.
24924 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24925 Use vect_dr_behavior.
24926 (vect_setup_realignment): Update call accordingly.
24927 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24928 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24929 call to vect_create_addr_base_for_vector_ref.
24930 (vect_create_cond_for_align_checks): Likewise.
24931 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24932 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24933 (vect_recog_mask_conversion_pattern): Likewise.
24934 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24935 (new_stmt_vec_info): Remove redundant zeroing.
24937 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24939 * common/config/arm/arm-common.c (arm_be8_option): New function.
24940 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24941 (ISA_ARMv6): Add isa_bit_be8.
24942 * config/arm/arm.h (arm_be8_option): Add prototype.
24943 (BE8_SPEC_FUNCTION): New define.
24944 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24945 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24946 (mlittle-endian): Similarly.
24947 (mbe8, mbe32): New options.
24948 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24949 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24951 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24953 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24955 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24957 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24958 (cleanup_tree_cfg_bb): Use it.
24959 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24961 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24963 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24966 * loop-doloop.c (add_test): Update profile.
24968 2017-07-03 Martin Liska <mliska@suse.cz>
24971 * sanopt.c (rewrite_usage_of_param): New function.
24972 (sanitize_rewrite_addressable_params): Likewise.
24973 (pass_sanopt::execute): Call rewrite_usage_of_param.
24975 2017-07-03 Richard Biener <rguenther@suse.de>
24977 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24978 back to using VIEW_CONVERT_EXPR.
24980 2017-07-03 Martin Liska <mliska@suse.cz>
24983 * doc/extend.texi: Document when a resolver function is
24984 generated for target_clones.
24986 2017-07-03 Martin Liska <mliska@suse.cz>
24988 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24989 and shadow memory of auto variables which are subject of
24990 use-after-scope sanitization.
24991 (asan_expand_mark_ifn): Add do set only when is_poison.
24993 2016-07-03 Richard Biener <rguenther@suse.de>
24995 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24997 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24998 (vectorizable_reduction): Perform reduction PHI creation when
24999 visiting a reduction PHI and adjust and simplify code generation
25000 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25001 (vect_transform_loop): Visit reduction PHIs.
25002 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25003 defs into the SLP tree.
25004 (vect_build_slp_tree): Reduction defs terminate the recursion.
25005 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25007 (vect_get_vec_defs_for_stmt_copy): Export.
25008 (vect_get_vec_defs): Likewise.
25009 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25011 (vect_get_vec_defs_for_stmt_copy): Declare.
25012 (vect_get_vec_defs): Likewise.
25014 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25016 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25017 parameter with a "loop" parameter and use it instead of the
25018 loop containing DR_STMT. Don't check simple_iv when doing
25019 BB analysis. Describe the two analysis modes in the comment.
25021 2017-07-03 Tom de Vries <tom@codesourcery.com>
25023 PR tree-optimization/69468
25024 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25025 (find_same_succ_bb): Handle ignore_edge_flags.
25027 2017-07-03 Tom de Vries <tom@codesourcery.com>
25029 PR tree-optimization/81192
25030 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25032 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25034 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25036 2017-07-03 Tom de Vries <tom@codesourcery.com>
25038 PR tree-optimization/81192
25039 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25040 BB_SAME_SUCC (bb) == NULL.
25042 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25044 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25049 * dumpfile.c: Include profile-count.h
25050 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25052 (insert_cond_bb): Update profile.
25053 * tree-cfg.h (insert_cond_bb): Update prototype.
25054 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25055 * tree-dump.c: Do not include tree-cfg.
25057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25059 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25061 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25063 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25066 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25068 * tree-complex.c (expand_complex_div_wide): update profile.
25070 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25071 Alan Hayward <alan.hayward@arm.com>
25072 David Sherwood <david.sherwood@arm.com>
25074 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25075 (CORETYPES_H): New define.
25076 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25077 (insn-modes-inline.h, s-modes-inline-h): New rules.
25078 (generated_files): Add insn-modes-inline.h.
25079 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25080 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25081 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25082 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25083 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25084 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25085 (build/gencodes.o, build/genconditions.o): Likewise.
25086 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25087 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25088 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25089 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25090 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25091 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25092 * coretypes.h: Include everything up to real.h for generators.
25093 Include insn-modes.h first. Include wide-int-print.h after
25094 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25095 * machmode.h: Don't include insn-modes.h here.
25096 * function-tests.c: Remove includes of signop.h, machmode.h,
25097 double-int.h and wide-int.h.
25099 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25101 * optc-save-gen.awk: Likewise.
25102 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25103 * godump.c: Remove include of wide-int-print.h.
25104 * pretty-print.h: Likewise.
25105 * wide-int-print.cc: Likewise.
25106 * wide-int.cc: Likewise.
25107 * hash-map-tests.c: Remove include of signop.h.
25108 * hash-set-tests.c: Likewise.
25109 * rtl-tests.c: Likewise.
25110 * mkconfig.sh: Remove include of machmode.h.
25111 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25113 (emit_insn_modes_inline_h): ...this new function. Emit the code
25114 into an insn-modes-inline.h header file, adding appropriate
25115 include guards and end comments.
25116 (emit_insn_modes_c_header): Remove include of machmode.h.
25117 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25119 (main): Handle -i flag and call emit_insn_modes_inline_h when
25122 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25124 * tree-ssa-strlen.c (strinfo): Rename the length field to
25125 nonzero_chars. Add a full_string_p field.
25126 (compare_nonzero_chars, zero_length_string_p): New functions.
25127 (get_addr_stridx): Add an offset_out parameter.
25128 Use compare_nonzero_chars.
25129 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25130 (new_strinfo): Update after above changes to strinfo.
25131 (set_endptr_and_length): Set full_string_p.
25132 (get_string_length): Update after above changes to strinfo.
25133 (unshare_strinfo): Update call to new_strinfo.
25134 (maybe_invalidate): Likewise.
25135 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25136 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25137 as a uhwi instead of an shwi. Update after above changes to
25138 strinfo and new_strinfo.
25139 (zero_length_string): Assert that chainsi contains full strings.
25140 Use zero_length_string_p. Update call to new_strinfo.
25141 (adjust_related_strinfos): Update after above changes to strinfo.
25142 Copy full_string_p from origsi.
25143 (adjust_last_stmt): Use zero_length_string_p.
25144 (handle_builtin_strlen): Update after above changes to strinfo and
25145 new_strinfo. Install the lhs as the string length if the previous
25146 entry didn't describe a full string.
25147 (handle_builtin_strchr): Update after above changes to strinfo
25149 (handle_builtin_strcpy): Likewise.
25150 (handle_builtin_strcat): Likewise.
25151 (handle_builtin_malloc): Likewise.
25152 (handle_pointer_plus): Likewise.
25153 (handle_builtin_memcpy): Likewise. Track nonzero characters
25154 that aren't necessarily followed by a nul terminator.
25155 (handle_char_store): Likewise.
25157 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25159 PR tree-optimization/80769
25160 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25161 for malloc and calloc. Document the new invariant that all related
25162 strinfos have delayed lengths or none do.
25163 (verify_related_strinfos): Move earlier in file.
25164 (set_endptr_and_length): New function, split out from...
25165 (get_string_length): ...here. Also set the lengths of related
25167 (zero_length_string): Assert that chainsi has known (rather than
25169 (adjust_related_strinfos): Likewise.
25171 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25173 PR tree-optimization/81136
25174 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25175 assert that two references with the same misalignment have the same
25176 compile-time misalignment if those compile-time misalignments
25179 2017-07-01 Andi Kleen <ak@linux.intel.com>
25181 * print-tree.c (print_node): Print all attributes.
25183 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25185 * cfg.c (scale_bbs_frequencies): New function.
25186 * cfg.h (scale_bbs_frequencies): Declare it.
25187 * cfgloopanal.c (single_likely_exit): Cleanup.
25188 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25190 (scale_loop_profile): Likewise.
25191 (loop_version): Likewise.
25192 (create_empty_loop_on_edge): Update.
25193 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25194 scale_loop_frequencies, scale_loop_profile, loopify,
25195 loop_version): Update prototypes.
25196 * modulo-sched.c (sms_schedule): Update.
25197 * predict.c (unlikely_executed_edge_p): Also check probability.
25198 (probably_never_executed_edge_p): Fix typo.
25199 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25200 * tree-parloops.c (gen_parallel_loop): Update.
25201 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25202 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25203 * tree-ssa-loop-split.c (split_loop): Update.
25204 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25205 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25206 (vect_loop_versioning): Update.
25207 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25209 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25211 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25213 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25215 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25216 to keep profile consistent.
25218 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25220 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25221 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25222 * profile-count.h (max_safe_multiplier): Make unsigned.
25223 (profile_count::guessed_zero): New.
25225 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25227 * bb-reorder.c (fix_up_crossing_landing_pad,
25228 fix_crossing_conditional_branches): Use make_single_succ_edge
25229 to keep profile consistent.
25231 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25233 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25236 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25239 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25240 the right scopes, make sure cond_jump isn't preserved between multiple
25241 iterations. Search for fallthru edge whenever there are 3+ edges and
25242 use find_fallthru_edge for it.
25244 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25246 Patch by Alexander Monakov <amonakov@ispras.ru>
25247 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25248 probabilities consistently.
25250 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25252 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25253 emit_cmp_and_jump_insns.
25255 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25258 * tree-inline.c (expand_call_inline): Combine profile statuses.
25260 2017-06-30 Andrew Pinski <apinski@cavium.com>
25262 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25263 fold_stmt returned true.
25265 2017-06-30 Nathan Sidwell <nathan@acm.org>
25267 * ggc.h (empty_string): Delete.
25268 * cfgexpand.c (expand_asm_stmt): Use plain "".
25269 * optabs.c (expand_asm_memory_barrier): Likewise.
25270 * stringpool.c (empty_string): Delete.
25271 (digit_vector, digit_string): Delete.
25272 (ggc_alloc_string): Use plain "", don't optimize single digit
25273 strings. Use ggc_alloc_atomic.
25275 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25277 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25278 comparison set and one other set, use the cost of the non-comparison
25281 2017-06-30 Nathan Sidwell <nathan@acm.org>
25283 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25286 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25288 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25289 loops. Remove now unneeded calls to gimple_switch_set_label() that
25290 just set removed labels to NULL_TREE.
25292 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25294 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25295 (set_range_info): ...here. Only call set_range_info_raw if domain
25297 (set_nonzero_bits): Call set_range_info_raw.
25298 * tree-ssanames.h (set_range_info_raw): New.
25300 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25303 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25304 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25305 of nonimmediate_operand and <store_mask_constraint> instead of m
25306 for the input operand. For V8FI iterator, always split if input
25307 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25308 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25309 <store_mask_predicate> instead of register_operand and
25310 <store_mask_constraint> instead of v for the input operand. Make
25311 sure both operands aren't MEMs for if not <mask_applied>.
25313 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25315 * lto-wrapper.c (copy_file) Close both file descriptors before
25318 2017-06-30 Martin Liska <mliska@suse.cz>
25321 * multiple_target.c (create_dispatcher_calls): Make ifunc
25322 also for function that don't have calls or are not referenced.
25324 2017-06-30 Richard Biener <rguenther@suse.de>
25326 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25327 analyze the first scalar stmt. Move vector type computation
25328 for the BB case here from ...
25329 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25330 live operation processing in the SLP case properly.
25332 2017-06-30 Richard Biener <rguenther@suse.de>
25334 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25336 2017-06-30 Martin Liska <mliska@suse.cz>
25339 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25340 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25342 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25344 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25347 2017-06-30 Richard Biener <rguenther@suse.de>
25349 PR tree-optimization/81249
25350 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25351 condition reduction result to original scalar type.
25353 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25355 * profile-count.h (enum profile_quality): Fix typos and whitespace
25358 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25360 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25361 type for branch probabilities.
25363 2017-06-29 Julian Brown <julian@codesourcery.com>
25364 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25366 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25367 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25368 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25369 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25371 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25373 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25374 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25375 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25376 CC usage from generic code to here.
25377 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25378 CC usage into the target macros.
25380 2017-06-29 Maya Rashish <coypu@sdf.org>
25382 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25385 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25387 * arm/arm-builtins.c: Include profile-count.h
25388 * except.c (sjlj_emit_function_enter): Use
25389 profile_probability::unlikely.
25391 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25393 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25394 and tocrel_offset be pointer args rather than implicitly using
25396 (legitimate_constant_pool_address_p, rs6000_emit_move,
25397 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25398 tocrel_offset and use in toc_relative_expr_p call.
25399 (print_operand, print_operand_address): Use static tocrel_base_oac
25400 and tocrel_offset_oac.
25401 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25404 2017-06-29 Maya Rashish <coypu@sdf.org>
25406 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25408 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25410 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25411 objects, take into account only the alignment of 'op0' and 'mode1' if
25414 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25416 * ccmp.c (ccmp_tree_comparison_p): New function.
25417 (ccmp_candidate_p): Update to use above function.
25418 (get_compare_parts): New function.
25419 (expand_ccmp_next): Update to use new functions.
25420 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25422 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25423 take mode as argument.
25424 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25425 * expr.c (expand_expr_real_1): Pass mode as argument.
25427 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25429 * combine.c (combine_instructions): Print insns to dump_file, together
25432 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25434 * asan.c (asan_emit_stack_protection): Update.
25435 (create_cond_insert_point): Update.
25436 * auto-profile.c (afdo_propagate_circuit): Update.
25437 * basic-block.h (struct edge_def): Turn probability to
25438 profile_probability.
25439 (EDGE_FREQUENCY): Update.
25440 * bb-reorder.c (find_traces_1_round): Update.
25441 (better_edge_p): Update.
25442 (sanitize_hot_paths): Update.
25443 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25444 (make_single_succ_edge): Update.
25445 (check_bb_profile): Update.
25446 (dump_edge_info): Update.
25447 (update_bb_profile_for_threading): Update.
25448 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25449 probabilitycount to 0.
25450 * cfgbuild.c (compute_outgoing_frequencies): Update.
25451 * cfgcleanup.c (try_forward_edges): Update.
25452 (outgoing_edges_match): Update.
25453 (try_crossjump_to_edge): Update.
25454 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25455 (expand_gimple_tailcall): Update.
25456 (construct_init_block): Use make_single_succ_edge.
25457 (construct_exit_block): Use make_single_succ_edge.
25458 * cfghooks.c (verify_flow_info): Update.
25459 (redirect_edge_succ_nodup): Update.
25460 (split_edge): Update.
25461 (account_profile_record): Update.
25462 * cfgloopanal.c (single_likely_exit): Update.
25463 * cfgloopmanip.c (scale_loop_profile): Update.
25464 (set_zero_probability): Remove.
25465 (duplicate_loop_to_header_edge): Update.
25466 * cfgloopmanip.h (loop_version): Update prototype.
25467 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25468 (force_nonfallthru_and_redirect): Update.
25469 (update_br_prob_note): Update.
25470 (rtl_verify_edges): Update.
25471 (purge_dead_edges): Update.
25472 (rtl_lv_add_condition_to_bb): Update.
25473 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25474 * cgraphunit.c (init_lowered_empty_function): Update.
25475 (cgraph_node::expand_thunk): Update.
25476 * cilk-common.c: Include profile-count.h
25477 * dojump.c (inv): Remove.
25478 (jumpifnot): Update.
25479 (jumpifnot_1): Update.
25480 (do_jump_1): Update.
25482 (do_jump_by_parts_greater_rtx): Update.
25483 (do_compare_rtx_and_jump): Update.
25484 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25485 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25486 * dwarf2cfi.c: Include profile-count.h
25487 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25488 (sjlj_emit_dispatch_table): Likewise.
25489 * explow.c: Include profile-count.h
25490 * expmed.c (emit_store_flag_force): Update.
25491 (do_cmp_and_jump): Update.
25492 * expr.c (compare_by_pieces_d::generate): Update.
25493 (compare_by_pieces_d::finish_mode): Update.
25494 (emit_block_move_via_loop): Update.
25495 (store_expr_with_bounds): Update.
25496 (store_constructor): Update.
25497 (expand_expr_real_2): Update.
25498 (expand_expr_real_1): Update.
25499 * expr.h (try_casesi, try_tablejump): Update prototypes.
25500 * gimple-pretty-print.c (dump_probability): Update.
25501 (dump_profile): New.
25502 (dump_gimple_label): Update.
25503 (dump_gimple_bb_header): Update.
25504 * graph.c (draw_cfg_node_succ_edges): Update.
25505 * hsa-gen.c (convert_switch_statements): Update.
25506 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25507 (find_if_case_1): Update.
25508 (find_if_case_2): Update.
25509 * internal-fn.c (expand_arith_overflow_result_store): Update.
25510 (expand_addsub_overflow): Update.
25511 (expand_neg_overflow): Update.
25512 (expand_mul_overflow): Update.
25513 (expand_vector_ubsan_overflow): Update.
25514 * ipa-cp.c (good_cloning_opportunity_p): Update.
25515 * ipa-split.c (split_function): Use make_single_succ_edge.
25516 * ipa-utils.c (ipa_merge_profiles): Update.
25517 * loop-doloop.c (add_test): Update.
25518 (doloop_modify): Update.
25519 * loop-unroll.c (compare_and_jump_seq): Update.
25520 (unroll_loop_runtime_iterations): Update.
25521 * lra-constraints.c (lra_inheritance): Update.
25522 * lto-streamer-in.c (input_cfg): Update.
25523 * lto-streamer-out.c (output_cfg): Update.
25524 * mcf.c (adjust_cfg_counts): Update.
25525 * modulo-sched.c (sms_schedule): Update.
25526 * omp-expand.c (expand_omp_for_init_counts): Update.
25527 (extract_omp_for_update_vars): Update.
25528 (expand_omp_ordered_sink): Update.
25529 (expand_omp_for_ordered_loops): Update.
25530 (expand_omp_for_generic): Update.
25531 (expand_omp_for_static_nochunk): Update.
25532 (expand_omp_for_static_chunk): Update.
25533 (expand_cilk_for): Update.
25534 (expand_omp_simd): Update.
25535 (expand_omp_taskloop_for_outer): Update.
25536 (expand_omp_taskloop_for_inner): Update.
25537 * omp-simd-clone.c (simd_clone_adjust): Update.
25538 * optabs.c (expand_doubleword_shift): Update.
25539 (expand_abs): Update.
25540 (emit_cmp_and_jump_insn_1): Update.
25541 (expand_compare_and_swap_loop): Update.
25542 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25543 * predict.c (predictable_edge_p): Update.
25544 (edge_probability_reliable_p): Update.
25545 (set_even_probabilities): Update.
25546 (combine_predictions_for_insn): Update.
25547 (combine_predictions_for_bb): Update.
25548 (propagate_freq): Update.
25549 (estimate_bb_frequencies): Update.
25550 (force_edge_cold): Update.
25551 * profile-count.c (profile_count::dump): Add missing space into dump.
25552 (profile_count::debug): Add newline.
25553 (profile_count::differs_from_p): Explicitly convert to unsigned.
25554 (profile_count::stream_in): Update.
25555 (profile_probability::dump): New member function.
25556 (profile_probability::debug): New member function.
25557 (profile_probability::differs_from_p): New member function.
25558 (profile_probability::differs_lot_from_p): New member function.
25559 (profile_probability::stream_in): New member function.
25560 (profile_probability::stream_out): New member function.
25561 * profile-count.h (profile_count_quality): Rename to ...
25562 (profile_quality): ... this one.
25563 (profile_probability): New.
25564 (profile_count): Update.
25565 * profile.c (compute_branch_probabilities): Update.
25566 * recog.c (peep2_attempt): Update.
25567 * sched-ebb.c (schedule_ebbs): Update.
25568 * sched-rgn.c (find_single_block_region): Update.
25569 (compute_dom_prob_ps): Update.
25570 (schedule_region): Update.
25571 * sel-sched-ir.c (compute_succs_info): Update.
25572 * stmt.c (struct case_node): Update.
25573 (do_jump_if_equal): Update.
25574 (get_outgoing_edge_probs): Update.
25575 (conditional_probability): Update.
25576 (emit_case_dispatch_table): Update.
25577 (expand_case): Update.
25578 (expand_sjlj_dispatch_table): Update.
25579 (emit_case_nodes): Update.
25580 * targhooks.c: Update.
25581 * tracer.c (better_p): Update.
25582 (find_best_successor): Update.
25583 * trans-mem.c (expand_transaction): Update.
25584 * tree-call-cdce.c: Update.
25585 * tree-cfg.c (gimple_split_edge): Upate.
25586 (move_sese_region_to_fn): Upate.
25587 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25588 * tree-eh.c (lower_resx): Upate.
25589 (cleanup_empty_eh_move_lp): Upate.
25590 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25591 * tree-inline.c (copy_edges_for_bb): Update.
25592 (copy_cfg_body): Update.
25593 * tree-parloops.c (gen_parallel_loop): Update.
25594 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25595 (gimple_gen_time_profiler): Update.
25596 * tree-ssa-dce.c (remove_dead_stmt): Update.
25597 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25598 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25599 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25600 (unloop_loops): Update.
25601 (try_peel_loop): Update.
25602 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25603 * tree-ssa-loop-split.c (connect_loops): Update.
25604 (split_loop): Update.
25605 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25606 (hoist_guard): Update.
25607 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25608 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25609 (value_replacement): Update.
25610 * tree-ssa-reassoc.c (branch_fixup): Update.
25611 * tree-ssa-tail-merge.c (replace_block_by): Update.
25612 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25613 (create_edge_and_update_destination_phis): Update.
25614 (compute_path_counts): Update.
25615 (recompute_probabilities): Update.
25616 (update_joiner_offpath_counts): Update.
25617 (freqs_to_counts_path): Update.
25618 (duplicate_thread_path): Update.
25619 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25620 (struct switch_conv_info): Update.
25621 (gen_inbound_check): Update.
25622 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25623 (vect_do_peeling): Update.
25624 (vect_loop_versioning): Update.
25625 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25626 (optimize_mask_stores): Update.
25627 * ubsan.c (ubsan_expand_null_ifn): Update.
25628 * value-prof.c (gimple_divmod_fixed_value): Update.
25629 (gimple_divmod_fixed_value_transform): Update.
25630 (gimple_mod_pow2): Update.
25631 (gimple_mod_pow2_value_transform): Update.
25632 (gimple_mod_subtract): Update.
25633 (gimple_mod_subtract_transform): Update.
25634 (gimple_ic): Update.
25635 (gimple_stringop_fixed_value): Update.
25636 (gimple_stringops_transform): Update.
25637 * value-prof.h: Update.
25639 2017-06-29 Carl Love <cel@us.ibm.com>
25641 * config/rs6000/rs6000-c.c: Add support for built-in functions
25642 vector signed int vec_signed (vector float);
25643 vector signed long long vec_signed (vector double);
25644 vector signed int vec_signed2 (vector double, vector double);
25645 vector signed int vec_signede (vector double);
25646 vector signed int vec_signedo (vector double);
25647 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25648 instruction generator.
25649 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25650 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25651 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25653 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25654 vunsignede_v2df): Add define_expands.
25655 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25656 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25657 VEC_UNSIGNEDO): Add definitions.
25658 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25659 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25660 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25661 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25662 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25663 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25664 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25665 vec_unsignede, vec_unsignedo): Add builtin defines.
25666 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25668 * doc/extend.texi: Update the built-in documentation file for the
25669 new built-in functions.
25671 2017-06-29 Richard Biener <rguenther@suse.de>
25673 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25674 reduction chains to LOOP_VINFO_REDUCTIONS.
25675 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25676 SLP reductions after processing reduction chains.
25678 2017-06-29 Nathan Sidwell <nathan@acm.org>
25680 * builtins.c (fold_builtin_FUNCTION): Use
25681 lang_hooks.decl_printable_name.
25683 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25685 PR middle-end/81194
25686 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25687 with only one label.
25688 * stmt.c (expand_case): Assert NCASES is greater than one.
25690 2017-06-29 Richard Biener <rguenther@suse.de>
25692 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25694 (group_case_labels): Likewise.
25695 (find_taken_edge): Push sanity checking on val to workers...
25696 (find_taken_edge_cond_expr): ... here
25697 (find_taken_edge_switch_expr): ... and here, handle cases
25698 with just a default label.
25699 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25700 (group_case_labels): Likewise.
25701 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25702 group_case_labels does anything cleanup the CFG again.
25704 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25706 PR tree-optimization/81196
25707 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25708 exit condition comparing two IVs.
25710 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25712 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25713 profile to the dummy entry at the end of the list of architectures.
25714 * config/arm/arm-cpu-cdata.h: Regenerated.
25716 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25717 Michael Collison <michael.collison@arm.com>
25720 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25722 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25723 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25724 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25725 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25726 of shift when the shift amount is masked with constant equal to
25727 the size of the mode.
25728 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25731 2017-06-29 Martin Liska <mliska@suse.cz>
25733 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25735 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25738 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25741 (edge_growth_cache_entry::edge_growth_cache_entry): New
25743 (reset_edge_growth_cache): Update to use constructor.
25745 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25747 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25748 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25749 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25751 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25753 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25754 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25756 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25758 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25759 (*-linux-uclibc*): Add t-uclibc tmake_file.
25760 * config/t-musl: New.
25761 * config/t-uclibc: New.
25763 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25765 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25767 (gen_comm_data): Emit architectural setting of arch_prof.
25768 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25770 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25771 (armv8-m.base, armv8-m.main): Likewise.
25772 * arm-protos.h (arm_build_target): Add profile field.
25773 (arch_option): Likewise.
25774 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25776 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25777 arm_active_target.profile.
25779 2017-06-28 Richard Biener <rguenther@suse.de>
25781 PR middle-end/81227
25782 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25783 TYPE_OVERFLOW_WRAPS.
25784 * match.pd (negate_expr_p): Likewise.
25785 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25786 fold_build2, not fold_binary.
25788 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25790 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25791 Convert memory address to Pmode.
25792 (aarch64_print_operand): Assert MEM operands are always Pmode.
25794 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25797 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25798 Remove redundant if.
25799 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25800 * config/arm/aarch-common-protos.h
25801 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25802 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25804 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25807 * multiple_target.c (create_dispatcher_calls): Set the default
25808 clone to be static, not public.
25810 2017-06-28 Richard Biener <rguenther@suse.de>
25812 * tree-vect-loop.c (vectorizable_reduction): Move special
25813 cond reduction IV var creation ...
25814 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25815 parameter. Use STMT_VINFO_VECTYPE.
25816 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25819 2017-06-28 Martin Liska <mliska@suse.cz>
25822 * ipa-visibility.c (non_local_p): Handle visibility.
25824 2017-06-28 Martin Liska <mliska@suse.cz>
25827 * common.opt: Add IntegerRange to various options.
25828 * opt-functions.awk (integer_range_info): New function.
25829 * optc-gen.awk: Add integer_range_info to cl_options struct.
25830 * opts-common.c (decode_cmdline_option): Handle
25831 CL_ERR_INT_RANGE_ARG.
25832 (cmdline_handle_error): Likewise.
25833 * opts.c (print_filtered_help): Show valid interval in
25834 when --help is provided.
25835 * opts.h (struct cl_option): Add range_min and range_max fields.
25836 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25838 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25840 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25841 (x * C EQ/NE y * C): New transformation.
25843 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25845 * genmultilib (combination_space): Accept '+' in option names.
25847 2017-06-28 Martin Liska <mliska@suse.cz>
25850 * asan.c (instrument_derefs): Bail out inner references
25851 that are hard register variables.
25853 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25856 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25857 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25859 2017-06-28 Richard Biener <rguenther@suse.de>
25861 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25862 (vect_get_slp_defs): Adjust.
25863 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25865 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25867 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25868 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25869 (vectorizable_reduction): Adjust.
25870 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25872 (vect_get_slp_defs): Likewise.
25873 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25874 (vectorizable_bswap): Adjust.
25875 (vectorizable_call): Likewise.
25876 (vectorizable_conversion): Likewise.
25877 (vectorizable_assignment): Likewise.
25878 (vectorizable_shift): Likewise.
25879 (vectorizable_operation): Likewise.
25880 (vectorizable_store): Likewise.
25881 (vectorizable_condition): Likewise.
25882 (vectorizable_comparison): Likewise.
25884 2017-06-28 Michael Collison <michael.collison@arm.com>
25887 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25889 (arm_gen_movmemqi): Removed unused variable 'i'.
25890 Convert 'for' loop into 'while' loop.
25891 (arm_expand_prologue): Remove last unnecessary set of insn.
25892 (thumb_pop): Remove unused variable 'pushed_words'.
25893 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25895 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25897 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25898 * config/s390/s390-protos.h: Add prototype of
25899 s390_rel_address_ok_p.
25900 * config/s390/s390.c (s390_got_symbol): New function.
25901 (s390_rel_address_ok_p): New function.
25902 (legitimize_pic_address): Use s390_rel_address_ok_p.
25903 (s390_load_got): Use s390_got_symbol.
25904 (s390_option_override): Issue error if
25905 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25906 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25908 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25910 2017-06-27 Andrew Pinski <apinski@cavium.com>
25912 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25913 (X * copysign (1.0, X)): New pattern.
25914 (X * copysign (1.0, -X)): New pattern.
25915 (copysign (-1.0, CST)): New pattern.
25917 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25919 * genmultilib (combination_space): Remove variable.
25920 Validate reuse rules against regular expression for any sequence
25921 of multilib options in any order.
25923 2017-06-27 Michael Collison <michael.collison@arm.com>
25925 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25926 call aarch64_split_simd_combine.
25927 * (aarch64_combine_internal<mode>): Delete pattern.
25928 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25929 Allow register and subreg operands.
25931 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25933 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25934 specific need, just fallback on defaults.
25935 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25937 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25938 Olivier Hainque <hainque@adacore.com>
25940 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25942 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25943 targets. Pick a default if no particular attempt applied.
25944 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25947 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25949 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25950 (x86_64-wrs-vxworks7): Likewise.
25952 2017-06-27 Marek Polacek <polacek@redhat.com>
25955 * ubsan.c (instrument_null): Check get_base_address's result for null.
25957 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25959 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25961 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25963 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25964 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25965 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25966 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25967 New function types.
25968 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25969 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25970 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25971 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25972 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25973 BUILT_IN_FEUPDATEENV): New builtins.
25974 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25975 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25976 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25977 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25979 (builtin_structptr_types): Adjust size.
25980 * tree.c (builtin_structptr_types): Add four entries.
25982 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25983 Olivier Hainque <hainque@adacore.com>
25985 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25986 (TLS_SYM): New local macro, forcing reference to __tls__ on
25987 link command lines for VxWorks 7 RTPs, triggering initialization
25989 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25990 OS features TLS support, true for RTPs on VxWorks 7.
25991 * config/vxworks.c (vxworks_override_options): Setup emutls
25994 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25996 * predict.c (test_prediction_value_range): Use -1U instead of -1
25997 to avoid narrowing conversion warning.
25998 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25999 to avoid narrowing conversion warning.
26000 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26002 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26004 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26006 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26007 64bit configurations.
26008 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26009 (SIZE_TYPE): Likewise.
26010 * config/vxworks.c (vxworks_emutls_var_fields): Use
26011 long_unsigned_type_node instead of unsigned_type_node as the offset
26012 field type, which is "pointer" mode in emutls.c.
26014 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26017 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26019 PR middle-end/81207
26020 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26021 gimple_vuse copying separately from gimple_vdef copying.
26023 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26025 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26028 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26029 Olivier Hainque <hainque@adacore.com>
26031 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26032 for all vxworks7 targets.
26033 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26034 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26035 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26036 variations for VX6/VX7 and 32/64bits later on in ...
26037 (VXWORKS_LIB_SPEC): Leverage new macros.
26038 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26039 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26041 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26043 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26044 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26046 2017-06-26 Carl Love <cel@us.ibm.com>
26048 * config/rs6000/rs6000-c.c: Add support for built-in functions
26049 vector bool char vec_reve (vector bool char);
26050 vector signed char vec_reve (vector signed char);
26051 vector unsigned char vec_reve (vector unsigned char);
26052 vector bool int vec_reve (vector bool int);
26053 vector signed int vec_reve (vector signed int);
26054 vector unsigned int vec_reve (vector unsigned int);
26055 vector bool long long vec_reve (vector bool long long);
26056 vector signed long long vec_reve (vector signed long long);
26057 vector unsigned long long vec_reve (vector unsigned long long);
26058 vector bool short vec_reve (vector bool short);
26059 vector signed short vec_reve (vector signed short);
26060 vector double vec_reve (vector double);
26061 vector float vec_reve (vector float);
26062 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26063 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26064 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26065 (altivec_vreve): New pattern.
26066 * config/rs6000/altivec.h (vec_reve): New define.
26067 * doc/extend.texi (vec_rev): Update the built-in documentation file
26068 for the new built-in functions.
26070 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26072 PR tree-optimization/71815
26073 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26075 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26077 (slsr_process_phi): Likewise.
26078 (replace_uncond_cands_and_profitable_phis): Don't replace a
26079 multiply candidate with a stride of 1 (copy or cast).
26080 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26082 (lowest_cost_path): Likewise.
26083 (total_savings): Likewise.
26085 2017-06-26 Richard Biener <rguenther@suse.de>
26088 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26089 Use def_builtin_pure for all gather builtins.
26091 2017-06-26 Richard Biener <rguenther@suse.de>
26093 PR tree-optimization/81203
26094 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26095 non-dominating BBs.
26097 2017-06-26 Marek Polacek <polacek@redhat.com>
26100 * doc/invoke.texi: Document -Wmultistatement-macros.
26102 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26104 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26105 arm_neon_ok_no_float_abi effective target.
26107 2017-06-26 Richard Biener <rguenther@suse.de>
26109 PR tree-optimization/80928
26110 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26111 (copy_bbs): Set BB_DUPLICATED flag early.
26112 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26114 (execute_on_shrinking_pred): Likewise.
26115 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26116 BB_DUPLICATED blocks.
26117 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26118 iterate over all PHIs considering removal of *gsi.
26120 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26122 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26125 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26127 * config/rs6000/rs6000-string.c: (expand_block_clear,
26128 do_load_for_compare, select_block_compare_mode,
26129 compute_current_alignment, expand_block_compare,
26130 expand_strncmp_align_check, expand_strn_compare,
26131 expand_block_move, rs6000_output_load_multiple)
26132 Move functions related to string/block move/compare
26133 to a separate file.
26134 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26135 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26136 for this function which is now used in two files.
26137 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26138 * config.gcc: Add rs6000-string.o to extra_objs for
26139 targets powerpc*-*-* and rs6000*-*-*.
26141 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26144 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26145 32-bit, since indexed is not valid for DImode.
26146 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26147 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26148 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26149 (define_peephole2 for Altivec d-form store): Likewise.
26152 * multiple_target.c (create_dispatcher_calls): Only create the
26153 dispatcher call if the function is the default clone of a
26154 versioned function.
26156 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26158 PR middle-end/80902
26159 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26160 a call, force the call to not be a tail call.
26162 2017-06-23 Jeff Law <law@redhat.com>
26164 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26167 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26169 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26170 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26171 handling for early expansion of vector shifts (sl,sr,sra,rl).
26172 (builtin_function_type): Add vector shift right instructions
26173 to the unsigned argument list.
26175 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26177 rtl-optimizatoin/79286
26178 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26179 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26180 trap. PIC register plus a const unspec without offset can never trap.
26182 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26184 * tree.h (builtin_structptr_type): New type.
26185 (builtin_structptr_types): Declare new array.
26186 * tree.c (builtin_structptr_types): New array.
26187 (free_lang_data, build_common_tree_nodes): Use it.
26189 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26192 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26195 2017-06-22 Matt Turner <mattst88@gmail.com>
26197 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26198 Lake models to skylake case. Assume skylake for unknown
26199 models with clflushopt.
26201 2017-06-22 Jeff Law <law@redhat.com>
26203 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26204 frame sizes that do not satisfy aarch64_uimm12_shift.
26206 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26208 * profile-count.h (apply_probability,
26209 apply_scale, probability_in): Fix checks for zero.
26211 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26213 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26214 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26216 2016-06-22 Richard Biener <rguenther@suse.de>
26218 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26219 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26220 REDUC_MAX_EXPR support.
26221 (vectorizable_reduction): Likewise.
26222 (vect_create_epilog_for_reduction): Likewise.
26224 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26226 * match.pd (A / (1 << B) -> A >> B): New.
26227 * generic-match-head.c: Include optabs-tree.h.
26228 * gimple-match-head.c: Likewise.
26229 * optabs-tree.h (target_supports_op_p): New.
26230 * optabs-tree.c (target_supports_op_p): New.
26232 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26234 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26235 $gcc_cv_ld --help output.
26236 (gcc_cv_ld_demangle): Likewise.
26237 (gcc_cv_ld_eh_frame_hdr): Likewise.
26238 (gcc_cv_ld_pie): Likewise.
26239 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26240 (gcc_cv_ld_buildid): Likewise.
26241 (gcc_cv_ld_sysroot): Likewise.
26242 (ld_bndplt_support): Likewise.
26243 (ld_pushpopstate_support): Likewise.
26244 * configure: Regenerate.
26245 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26247 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26250 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26251 operands indexes to avoid gap between operands and match_dups.
26253 2017-06-21 Andrew Pinski <apinski@cavium.com>
26255 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26256 Increment Arith_shift and Arith_shift_reg by 1.
26257 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26259 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26260 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26261 (aarch64_strip_extend): Add new argument and test for it.
26262 (aarch64_cheap_mult_shift_p): New function.
26263 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26264 add a cost if it is true.
26265 Update calls to aarch64_strip_extend.
26266 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26268 2017-06-21 Andrew Pinski <apinski@cavium.com>
26270 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26272 (thunderxt88): Likewise.
26273 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26274 (thunderx_prefetch_tune): New variable.
26275 (thunderx2t99_prefetch_tune): Update for the correct values.
26276 (thunderxt88_tunings): New variable.
26277 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26278 generic_prefetch_tune.
26279 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26281 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26283 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26284 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26285 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26286 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26287 (aarch64_atomic_cas<mode>, GPI): Likewise.
26289 2017-06-21 Martin Liska <mliska@suse.cz>
26291 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26292 statements on cold and hot labels.
26293 * predict.c (tree_estimate_probability_bb): Remove the
26294 prediction from this place.
26296 2017-06-21 Martin Liska <mliska@suse.cz>
26298 PR tree-optimization/79489
26299 * gimplify.c (maybe_add_early_return_predict_stmt): New
26301 (gimplify_return_expr): Call the function.
26302 * predict.c (tree_estimate_probability_bb): Remove handling
26304 * predict.def: Update comment about early return predictor.
26305 * gimple-predict.h (is_gimple_predict): New function.
26306 * predict.def: Change default value of early return to 66.
26307 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26309 * passes.def: Put pass_strip_predict_hints to the beginning of
26312 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26314 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26315 FUNCTION_DECL declarations.
26316 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26318 (dwaf2out_decl): Likewise.
26319 * godump.c (go_early_global_decl): Skip call to the real debug hook
26320 for FUNCTION_DECL declarations.
26321 * passes.c (rest_of_decl_compilation): Skip call to the
26322 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26323 -fdump-go-spec is passed.
26325 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26327 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26328 Reorder for compactness.
26329 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26330 (def_builtin_pure, def_builtin_pure2): New functions.
26331 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26333 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26335 * match.pd (nop_convert): New predicate.
26336 ((A +- CST1) +- CST2): Allow some NOP conversions.
26338 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26341 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26342 with ctors/dtors if GOVD_SHARED is set.
26344 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26346 * config/aarch64/aarch64.md (movti_aarch64):
26347 Emit mov rather than orr.
26348 (movtf_aarch64): Likewise.
26349 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26350 Emit mov rather than orr.
26352 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26354 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26355 Swap alternatives, make integer dup more expensive.
26357 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26359 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26360 Return true for non-tls symbols.
26362 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26364 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26365 (cortex-a75): Likewise.
26366 (cortex-a75.cortex-a55): Likewise.
26367 * config/aarch64/aarch64-tune.md: Regenerate.
26368 * doc/invoke.texi (-mtune): Document new values for -mtune.
26370 2017-06-21 Tom de Vries <tom@codesourcery.com>
26372 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26373 stack_size feature.
26374 (Effective-Target Keywords, Other attributes): Suggest using
26375 dg-add-options stack_size feature to get stack limit in stack_size
26376 effective target documentation.
26378 2017-06-21 Julian Brown <julian@codesourcery.com>
26379 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26381 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26382 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26383 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26385 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26386 attribute type list for neon_multiply.
26387 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26388 attribute type list for neon_multiply.
26389 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26390 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26391 attribute type list for neon_multiply.
26392 * config/arm/types.md (crypto_pmull): Add.
26393 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26394 attribute type list.
26396 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26398 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26401 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26403 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26404 to make sure not to dereference a NULL cost_classes_ptr pointer.
26406 2017-06-20 Carl Love <cel@us.ibm.com>
26408 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26409 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26410 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26411 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26412 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26413 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26414 VMULOSW): New enum "unspec" values.
26415 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26416 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26417 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26418 altivec_vmulosw): New patterns.
26419 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26420 VMULOSW): Add definitions.
26422 2017-06-20 Julia Koval <julia.koval@intel.com>
26424 * config/i386/i386.c: Fix rounding expand for new pattern.
26425 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26427 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26429 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26430 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26432 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26434 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26437 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26439 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26440 architecture, then by implementer ID.
26441 * config/aarch64/aarch64-tune.md: Regenerate.
26443 2017-06-20 Richard Biener <rguenther@suse.de>
26445 PR middle-end/81097
26446 * fold-const.c (split_tree): Fold to type before negating.
26448 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26450 * diagnostic-show-locus.c
26451 (selftest::test_fixit_deletion_affecting_newline): New function.
26452 (selftest::diagnostic_show_locus_c_tests): Call it.
26454 2017-06-20 Andreas Schwab <schwab@suse.de>
26457 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26460 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26462 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26463 __ARM_FEATURE_COPROC according to support.
26465 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26467 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26468 Rewritten to avoid overflow for > 32-bit pointers.
26471 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26472 by removing enum keyword.
26473 (ubsan_type_descriptor): Likewise. Formatting fix.
26476 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26477 splitter): Require TARGET_SSE2 in the condition.
26479 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26482 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26483 for doing vector set of SFmode on ISA 3.0.
26484 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26485 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26487 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26488 SFmode value into a V4SF variable that was extracted from another
26489 V4SF variable without converting the element to double precision
26490 and back to single precision vector format.
26491 (vsx_insert_extract_v4sf_p9_2): Likewise.
26493 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26495 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26496 in UWHI to avoid undefined overflow.
26499 * ubsan.h (enum ubsan_encode_value_phase): New.
26500 (ubsan_encode_value): Change second argument to
26501 enum ubsan_encode_value_phase with default value of
26502 UBSAN_ENCODE_VALUE_GENERIC.
26503 * ubsan.c (ubsan_encode_value): Change second argument to
26504 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26505 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26506 create_tmp_var_raw instead of create_tmp_var and use a
26508 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26509 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26510 ubsan_encode_value callers.
26513 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26514 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26515 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26517 2017-06-19 Richard Biener <rguenther@suse.de>
26519 PR middle-end/81118
26520 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26521 estimates if we changed anything.
26523 2017-06-19 Richard Biener <rguenther@suse.de>
26525 PR tree-optimization/80887
26526 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26527 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26528 simplified lookups, then reset mprts_hook.
26529 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26531 (try_to_simplify): Likewise.
26533 2017-06-19 Martin Liska <mliska@suse.cz>
26536 * gimplify.c (gimplify_switch_expr):
26537 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26539 2017-06-19 Martin Liska <mliska@suse.cz>
26541 * doc/install.texi: Document that PGO runs in 4 stages.
26543 2017-06-19 Martin Liska <mliska@suse.cz>
26546 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26547 to dispatcher function name.
26548 * multiple_target.c (replace_function_decl): New function.
26549 (create_dispatcher_calls): Redirect both edges and references.
26551 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26553 * profile-count.c (profile_count::dump): Dump quality.
26554 (profile_count::differs_from_p): Update for unsigned val.
26555 * profile-count.h (profile_count_quality): New enum.
26556 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26558 2017-06-19 Richard Biener <rguenther@suse.de>
26560 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26561 struct function as arg.
26562 (estimate_numbers_of_iterations): Export overload with loop arg.
26563 (free_numbers_of_iterations_estimates_loop): Use an overload of
26564 free_numbers_of_iterations_estimates instead.
26565 * tree-cfg.c (remove_bb): Adjust.
26566 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26567 * tree-parloops.c (gen_parallel_loop): Likewise.
26568 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26570 (tree_unroll_loops_completely): Likewise.
26571 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26572 Use an overload instead and export.
26573 (estimated_loop_iterations): Adjust.
26574 (max_loop_iterations): Likewise.
26575 (likely_max_loop_iterations): Likewise.
26576 (estimate_numbers_of_iterations): Take struct function as arg
26578 (loop_exits_before_overflow): Adjust.
26579 (free_numbers_of_iterations_estimates_loop): Use an overload.
26580 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26581 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26583 2017-06-19 Richard Biener <rguenther@suse.de>
26586 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26587 RANGE_EXPR conservatively.
26589 2017-06-16 Carl Love <cel@us.ibm.com>
26591 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26592 definitions for vec_float, vec_float2, vec_floato,
26593 vec_floate built-ins.
26594 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26595 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26597 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26598 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26599 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26600 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26601 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26602 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26603 vec_floato): Add builtin defines.
26604 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26605 Update the built-in documentation file for the new built-in
26608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26610 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26611 (mthumb): Mark as the negative of -marm.
26613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26615 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26617 (ARM Options, -mtune): Document that this accepts the same
26618 extension options as -mcpu.
26619 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26621 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26623 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26624 permitted extensions.
26626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26628 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26629 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26630 (armv8-m.main): Add option +nodsp.
26631 * config/arm/arm-cpu-cdata.h: Regenerated.
26633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26635 * config/arm/t-fuchsia: New file.
26636 * config.gcc (arm*-*-fuchsia*): Use it.
26638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26640 * config/arm/t-symbian: Rewrite for new option infrastructure.
26642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26644 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26645 (MULTILIB_REQUIRED): Likewise.
26647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26649 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26650 (MULTILIB_RESUE): Likewise.
26651 (MULTILIB_MATCHES): Likewise.
26652 (MULTLIB_REQUIRED): Likewise.
26654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26656 * config/arm/t-rtems: Rewrite for new option framework.
26658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26660 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26661 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26662 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26663 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26664 * config/arm/t-multilib: ... here.
26665 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26666 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26667 armv7-a and armv8*-a when A-profile libraries have not been built.
26668 * config/arm/t-rmprofile: Rewrite.
26670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26672 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26673 with a backslash. Remove the backslash after substituting unescaped
26675 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26679 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26680 the driver to pass through the default setting of -mfloat-abi.
26681 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26683 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26684 (all_feat_combs): New rule.
26685 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26687 * config/arm/t-aprofile: Rewrite.
26689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26691 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26692 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26693 fpu is not specified by the user/command-line.
26694 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26695 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26696 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26697 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26698 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26699 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26703 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26704 * config/arm/t-arm-elf: Rewritten.
26706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26708 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26709 have some floating-point instructions.
26710 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26711 (TARGET_MAYBE_HARD_FLOAT): New macro.
26712 * config/arm/arm-builtins.c (arm_init_builtins): Use
26713 TARGET_MAYBE_HARD_FLOAT.
26714 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26718 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26719 (configargs.h): Include it.
26720 (arm_print_hint_for_fpu_option): New function.
26721 (arm_parse_fpu_option): New function.
26722 (candidate_extension): New class.
26723 (arm_canon_for_multilib): New function.
26724 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26725 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26726 (ARCH_CANONICAL_SPECS): New macro.
26727 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26729 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26731 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26732 are set after handling multilib fragments. Set target_cpu_default2
26735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26737 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26739 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26740 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26742 (check_arch): Likewise.
26743 * config/arm/arm.c (arm_configure_build_target): Handle
26744 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26745 options in the default.
26747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26749 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26750 when an option is an alias of another.
26751 * config/arm/parsecpu.awk (optalias): New parser token.
26752 (gen_comm_data): Mark non-alias options as such. Emit entries
26753 for extension aliases.
26754 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26755 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26756 (armv6kz, armv6zk, armv6t2): Likewise.
26757 (armv7): Make vfpv3-d16 an alias.
26758 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26760 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26761 Sort in canonical order.
26762 (armv8-a): Sort in canonical order.
26763 (armv8.1-a, armv8.2-a): Likewise.
26764 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26766 (cortex-a9): Sort in canonical order.
26767 * config/arm/arm.c (selftests.h): Include it.
26768 (arm_test_cpu_arch_data): New function.
26769 (arm_run_self_tests): New function.
26770 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26771 (targetm): Move declaration to the end of the file.
26772 * arm-cpu-cdata.h: Regenerated.
26774 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26776 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26777 call to target_mode_check describing the type of option passed.
26778 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26779 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26780 arm_parse_cpu_option_name to match parameters against list of
26782 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26783 arm_arch_core_flags data structure.
26784 * config/arm/arm-cpu_cdata.h: Regenerated.
26786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26788 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26790 (arm_print_hint_for_cpu_option): Likewise.
26791 (arm_print_hint_for_arch_option): Likewise.
26792 (arm_parse_cpu_option_name): Likewise.
26793 (arm_parse_arch_option_name): Likewise.
26794 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26795 of entries in the all_fpus list.
26796 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26797 (arm_parse_cpu_option_name): Declare.
26798 (arm_parse_arch_option_name): Declare.
26799 (arm_parse_option_features): Declare.
26800 (arm_intialize_isa): Declare.
26801 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26803 (gen_comm_data): ... here. Make definitions non-static.
26804 * config/arm/arm-cpu-data.h: Regenerated.
26805 * config/arm/arm-cpu-cdata.h: Regenerated.
26807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26809 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26810 (cpu_arch_extension): New structure.
26811 (cpu_arch_option, arch_option, cpu_option): New structures.
26812 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26813 architecture types.
26814 (gen_data): Generate new format data tables.
26815 * config/arm/arm.c (cpu_tune): New structure.
26816 (cpu_option, processors): Delete.
26817 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26818 (arm_print_hint_for_cpu_option): ... this and ...
26819 (arm_print_hint_for_arch_option): ... this.
26820 (arm_parse_arch_cpu_name): Delete. Replace with ...
26821 (arm_parse_cpu_option_name): ... this and ...
26822 (arm_parse_arch_option_name): ... this.
26823 (arm_unrecognized_feature): Change type of target parameter to
26825 (arm_parse_arch_cpu_features): Delete. Replace with ...
26826 (arm_parse_option_features): ... this.
26827 (arm_configure_build_target): Rework to use new configuration data
26829 (arm_print_tune_info): Rework for new configuration data tables.
26830 * config/arm/arm-cpu-data.h: Regenerated.
26831 * config/arm/arm-cpu.h: Regenerated.
26833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26835 * Makefile.in (OBJS): Move sbitmap.o from here ...
26836 (OBJS-libcommon): ... to here.
26838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26840 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26841 (ISA_ALL_CRYPTO): New macro.
26842 (ISA_ALL_SIMD): New macro
26843 (ISA_ALL_FP): New macro.
26844 * config/arm/arm.c (fpu_bitlist): Update initializer.
26845 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26847 (arm9e): Add fpu. Add option for nofp
26848 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26849 (arm926ej-s, arm1026ej-s): Likewise.
26850 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26851 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26852 neon-fp16, neon-vfpv4, nofp and nosimd.
26853 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26854 (cortex-a8): Add fpu. Add option for nofp.
26855 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26856 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26857 (cortex-r4f): Add fpu.
26858 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26859 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26861 (cortex-r8): Likewise.
26862 (cortex-m4): Add fpu. Add option for nofp.
26863 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26864 (cortex-a17.cortex-a7): Likewise.
26865 (cortex-a32): Add fpu. Add options for crypto and nofp.
26866 (cortex-a35, cortex-a53): Likewise.
26867 (cortex-a57): Add fpu. Add option for crypto.
26868 (cortex-a72, cortex-a73): Likewise.
26869 (exynos-m1): Likewise.
26870 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26871 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26872 (cortex-m33): Add fpu. Add option for nofp.
26873 * config/arm/arm-cpu-cdata.h: Regenerated
26874 * config/arm/arm-cpu-data.h: Regenerated.
26876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26878 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26879 (armv5te, armv5tej): Likewise.
26880 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26881 (armv7): Add options fp and vfpv3-d16.
26882 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26883 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26885 (armv7ve): Likewise.
26886 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26887 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26888 (armv8-a): Add nocrypto option.
26889 (armv8.1-a, armv8.2-a): Likewise.
26890 (armv8-m.main): add options fp, fp.dp and nofp.
26892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26894 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26896 (armv8-a+crc): Delete.
26897 (armv8.1-a): Add options simd, crypto and nofp.
26898 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26899 (armv8.2-a+fp16): Delete.
26900 (armv8-m.main): Add option dsp.
26901 (armv8-m.main+dsp): Delete.
26902 (cortex-a8): Add fpu. Add nofp option.
26903 (cortex-a9): Add fpu. Add nofp and nosimd options.
26904 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26905 link to main cpu and architecture data structures.
26906 (gen_comm_data): Only put isa attributes from the main architecture
26908 (option): New statement for architecture and CPU entries.
26909 * arm.c (struct cpu_option): New structure.
26910 (struct processors): Add entry for options.
26911 (arm_unrecognized_feature): New function.
26912 (arm_parse_arch_cpu_name): Ignore any characters after the first
26914 (arm_parse_arch_cpu_feature): New function.
26915 (arm_configure_build_target): Separate out any CPU and architecture
26916 features and parse separately. Don't error out if -mfpu=auto is
26917 used with only an architecture string.
26918 (arm_print_asm_arch_directives): New function.
26919 (arm_file_start): Call it.
26920 * config/arm/arm-cpu-cdata.h: Regenerated.
26921 * config/arm/arm-cpu-data.h: Likewise.
26922 * config/arm/arm-tables.opt: Likewise.
26924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26926 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26927 assembler when it is not -mfpu=auto.
26929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26931 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26932 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26933 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26934 (ASM_CPU_SPEC): Rewrite.
26935 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26936 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26937 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26938 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26939 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26940 copied string is NUL-terminated. Also strip any characters prefixed
26942 (arm_rewrite_selected_arch): New function.
26943 (arm_rewrite_march): New function.
26945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26947 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26948 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26949 (march, mcpu, mtune): Convert to string-based options.
26950 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26951 (arm_parse_arch_cpu_name): New function.
26952 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26953 identify selected architecture or CPU.
26954 (arm_option_save): New function.
26955 (TARGET_OPTION_SAVE): Redefine.
26956 (arm_option_restore): Restore string options.
26957 (arm_option_print): Print string options.
26959 2017-06-16 Martin Sebor <msebor@redhat.com>
26961 PR tree-optimization/80933
26962 PR tree-optimization/80934
26963 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26964 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26965 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26966 (gimple_fold_builtin): Call them.
26968 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26970 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26971 as unlikely; update profile.
26973 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26975 * predict.c (force_edge_cold): Handle declaring edges impossible
26978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26980 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26982 (try_unroll_loop_completely): Fix reporting.
26984 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26986 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26988 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26991 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26992 if given a non-constant argument for an intrinsic which requires a
26995 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26997 * profile.c (compare_freqs): New function.
26998 (branch_prob): Sort edge list.
26999 (find_spanning_tree): Assume that the list is priority sorted.
27001 2017-06-16 Richard Biener <rguenther@suse.de>
27003 PR tree-optimization/81090
27004 * passes.def (pass_record_bounds): Remove.
27005 * tree-pass.h (make_pass_record_bounds): Likewise.
27006 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27007 make_pass_record_bounds): Likewise.
27008 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27009 not free niter estimates at the beginning but at the end.
27010 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27012 2017-06-16 Richard Biener <rguenther@suse.de>
27014 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27015 initializer to workaround ICE in host GCC 4.8.
27017 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27019 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27021 (clone_inlined_nodes): Update.
27023 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27025 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27026 prefetch settings, and enable prefetching by default at -O3.
27028 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27030 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27031 Set flag_prefetch_loop_arrays according to tuning data.
27033 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27035 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27036 New tune structure.
27037 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27038 [Unrelated to main purpose of the patch] Place the pointer field last
27039 to enable type checking errors when tune structure are wrongly merged.
27040 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27041 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27042 (thunderx2t99_prefetch_tune): New tune constants.
27043 (tune_params *_tunings): Update all tunings (no functional change).
27044 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27045 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27046 from tunings structures.
27048 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27051 * ubsan.c (instrument_null): Add T argument, use it instead
27052 of computing it based on IS_LHS.
27053 (instrument_object_size): Likewise.
27054 (pass_ubsan::execute): Adjust instrument_null and
27055 instrument_object_size callers to pass gimple_get_lhs or
27056 gimple_assign_rhs1 result to it. Use instrument_null instead of
27057 calling get_base_address and instrument_mem_ref. Handle
27058 aggregate call arguments for object-size sanitization.
27060 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27062 PR tree-optimization/81089
27063 * tree-vrp.c (is_masked_range_test): Validate operands of
27066 2017-06-15 Martin Sebor <msebor@redhat.com>
27069 * dumpfile.c (dump_register): Avoid calling memset to initialize
27070 a class with a default ctor.
27071 * gcc.c (struct compiler): Remove const qualification.
27072 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27073 * hash-table.h: Ditto.
27074 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27076 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27077 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27079 * params.h (struct param_info): Make struct members non-const.
27080 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27081 with default initialization.
27082 * vec.h (vec_copy_construct, vec_default_construct): New helper
27084 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27085 with vec_copy_construct.
27086 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27087 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27088 * doc/invoke.texi (-Wclass-memaccess): Document.
27090 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27092 * emit-rtl.h (is_leaf): Update comment about local
27093 register allocator.
27095 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27098 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27099 for a variable to have a section before checking if the section has a
27101 Set section to.persistent if persistent attribute is set.
27102 Warn if .persistent attribute is used on an automatic variable.
27104 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27106 PR rtl-optimization/80474
27107 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27109 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27111 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27114 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27116 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27117 (casesi): Emit bounds checking as RTL.
27118 (casesi_internal_mips16_<mode>): Remove bounds checking.
27120 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27122 * config/xtensa/xtensa.c (xtensa_option_override): Append
27123 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27124 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27125 xtensa_doloop_hooks): Define unconditionally.
27126 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27128 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27129 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27130 for it in xtensa_option_override.
27131 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27132 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27134 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27136 * doc/cppopts.texi: Document '-' special value to -MF.
27138 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27140 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27141 (cortex_a53_fconst): Likewise.
27142 (cortex_a53_fpmul): Likewise.
27143 (cortex_a53_f_load_64): Likewise.
27144 (cortex_a53_f_load_many): Likewise.
27145 (cortex_a53_advsimd_alu): Likewise.
27146 (cortex_a53_advsimd_alu_q): Likewise.
27147 (cortex_a53_advsimd_mul): Likewise.
27148 (cortex_a53_advsimd_mul_q): Likewise.
27149 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27150 Add missing fmul, r2f_cvt and fconst cases.
27152 2017-06-14 Richard Biener <rguenther@suse.de>
27154 PR middle-end/81088
27155 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27157 (fold_binary_loc): When associating do not treat pre-existing
27158 TREE_OVERFLOW on literal constants as a reason to allow
27159 TREE_OVERFLOW on associated literal constants.
27161 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27163 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27164 (MASK_FEATURES): New macro.
27165 * config/sparc/sparc.c (sparc_option_override): Remove the special
27166 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27168 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27170 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27171 a division of 0 if non-call exceptions are enabled.
27173 2017-06-14 Andrew Pinski <apinski@cavium.com>
27174 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27177 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27178 Improve vector initialization code gen for only variable case.
27180 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27182 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27184 2017-06-14 Richard Biener <rguenther@suse.de>
27186 PR tree-optimization/81083
27187 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27190 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27192 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27193 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27194 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27195 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27196 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27197 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27199 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27201 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27202 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27204 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27206 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27208 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27210 * config/rs6000/t-rtems: Don't handle SPE.
27212 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27214 * config/rs6000/t-linux: Don't handle SPE.
27216 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27218 * config/rs6000/eabispe.h: Delete file.
27220 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27222 * config/rs6000/t-spe: Delete file.
27224 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27226 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27227 (rs6000_legitimate_offset_address_p): Return false for anything in
27228 V2SImode or V2SFmode.
27230 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27232 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27233 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27234 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27236 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27237 (rs6000_legitimate_offset_address_p): Ditto.
27238 (rs6000_emit_move): Ditto.
27239 (rs6000_init_builtins): Remove V4HI_type_node.
27241 2017-06-13 Martin Liska <mliska@suse.cz>
27244 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27245 (gate_asan): Likewise.
27246 * asan.h (asan_no_sanitize_address_p): Remove the function.
27247 (sanitize_flags_p): New function.
27248 * builtins.def: Fix coding style.
27249 * common.opt: Use renamed enum value.
27250 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27251 * doc/extend.texi: Document no_sanitize attribute.
27252 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27253 to SANITIZE_UNDEFINED_NONDEFAULT.
27254 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27255 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27256 Use sanitize_flags_p.
27257 * gimplify.c (gimplify_function_tree): Likewise.
27258 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27259 * opts.c (parse_no_sanitize_attribute): New function.
27260 (common_handle_option): Use renamed enum value.
27261 * opts.h (parse_no_sanitize_attribute): Declare.
27262 * tree.c (sanitize_flags_p): New function.
27263 * tree.h: Declared here.
27264 * tsan.c: Use sanitize_flags_p.
27265 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27266 (instrument_mem_ref): Likewise.
27267 (instrument_bool_enum_load): Likewise.
27268 (do_ubsan_in_current_function): Remove the function.
27269 (pass_ubsan::execute): Use sanitize_flags_p.
27270 * ubsan.h: Remove do_ubsan_in_current_function
27271 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27272 (dump_function_to_file): Use it here.
27274 2017-06-13 Martin Jambor <mjambor@suse.cz>
27276 PR tree-optimization/80803
27277 PR tree-optimization/81063
27278 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27279 (propagate_subaccesses_across_link): Enqueue subtree whenever
27280 necessary instead of relying on the caller.
27282 2017-06-13 Martin Jambor <mjambor@suse.cz>
27284 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27285 that have a first_link.
27286 (sort_and_splice_var_accesses): Do not check first_link before
27288 (subtree_mark_written_and_enqueue): Likewise.
27289 (propagate_all_subaccesses): Likewise and do not stop at first
27290 parent with a first_link.
27292 2017-06-13 Martin Jambor <mjambor@suse.cz>
27294 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27297 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27299 * match.pd: New pattern.
27301 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27303 * tree-vrp.c (is_masked_range_test): New function.
27304 (register_edge_assert_for): Determine ranges for
27307 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27309 PR tree-optimization/67328
27310 * fold-const.c (maskable_range_p): New function.
27311 (build_range_check): Generate bittests if possible.
27313 2017-06-13 Martin Liska <mliska@suse.cz>
27315 * gimple-pretty-print.c (dump_probability): Add new argument.
27316 (dump_edge_probability): Dump both probability and count.
27317 (dump_gimple_label): Likewise.
27318 (dump_gimple_bb_header): Likewise.
27320 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27323 * config/avr/avr-devices.c: Fix indentation.
27324 * config/avr/gen-avr-mmcu-specs.c: Dito.
27326 2017-06-13 Richard Biener <rguenther@suse.de>
27328 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27329 instead get vector type from stmt_info.
27330 (vectorizable_reduction): Adjust. Remove dead code.
27332 2017-06-13 Richard Biener <rguenther@suse.de>
27334 PR middle-end/81065
27335 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27336 case of C * (x * C2 + C3).
27337 (fold_addr_of_array_ref_difference): Properly fold index difference.
27339 2017-06-12 David S. Miller <davem@davemloft.net>
27342 * config/sparc/sparc.md (return expander): Emit frame blockage if
27343 function uses alloca.
27345 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27347 * combine.c (make_field_assignment): Check len rather than the mode
27348 precision when calling force_to_mode.
27350 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27352 Support multilibs and devices that see flash in RAM address range.
27355 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27356 (avr_mcu_t) <flash_pm_offset>: New field.
27357 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27358 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27359 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27360 (AVR_TINY_PM_OFFSET): Remove macro.
27361 * config/avr/avr.opt (-mshort-calls): New option.
27362 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27363 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27364 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27365 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27366 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27367 instead of avr_arch->have_jmp_call.
27368 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27369 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27370 avr_arch->flash_pm_offset to define.
27371 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27372 new field flash_pm_offset. Add entry for avrxmega3.
27373 (avr_texinfo): Add entry for avrxmega3.
27374 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27375 attiny212, attiny214,
27376 attiny412, attiny414, attiny416, attiny417,
27377 attiny814, attiny816, attiny817,
27378 attiny1614, attiny1616, attiny1617,
27379 attiny3214, attiny3216, attiny3217.
27380 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27381 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27382 (avr_print_operand_address) [AVR_TINY]: Same.
27383 (avr_asm_init_sections) <readonly_data_section>: Only patch
27384 callback if avr_arch->flash_pm_offset = 0.
27385 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27386 for rodata if avr_arch->flash_pm_offset != 0.
27387 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27388 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27389 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27390 (m_options): Append opt_rcall.
27391 (m_dirnames): Append dir_rcall.
27392 * config/avr/t-multilib: Regenerate.
27394 * configure.ac [target=avr]: Check whether avrxmega3 default
27395 linker description file works as needed.
27396 * configure: Regenerate.
27397 * doc/avr-mmcu.texi: Regenerate.
27398 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27399 <__AVR_ARCH__>: Document avrxmega3 and 103.
27400 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27401 <__AVR_SHORT_CALLS__>: Document it.
27402 <__AVR_PM_BASE_ADDRESS__>: Document it.
27403 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27404 (AVR Variable Attributes) <progmem>: Document this is
27405 not needed for avrxmega3.
27406 (AVR Named Address Spaces) <__flash>: Dito.
27408 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27410 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27412 2017-06-12 Doug Rupp <rupp@adacore.com>
27414 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27415 Append vxworks-stdint.h to the tm_file list.
27416 * config/vxworks-stdint.h: New file.
27418 2017-06-12 Martin Liska <mliska@suse.cz>
27420 PR tree-optimization/81041
27421 * tree-profile.c (gimple_gen_ic_func_profiler):
27422 Create an extra BB in profile-generate
27423 (gimple_gen_time_profiler): Likewise.
27425 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27427 PR tree-optimization/81003
27428 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27429 (update_range_test): Use it instead of force_gimple_operand_gsi.
27431 2017-06-12 Richard Biener <rguenther@suse.de>
27433 PR tree-optimization/81053
27434 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27435 with backedge value not defined in loop. Simplify def stmt
27438 2017-06-11 Tom de Vries <tom@codesourcery.com>
27441 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27443 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27444 nvptx_cannot_force_const_mem.
27446 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27448 * opts.c (finish_options): Move test for flag_split_stack after
27449 it has been initialized.
27451 2017-06-11 Jason Merrill <jason@redhat.com>
27453 * tree.h (id_equal): New.
27454 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27455 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27456 instead of strcmp of IDENTIFIER_POINTER.
27458 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27460 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27461 (mark_all_inlined_calls_cdtor): Fix formating.
27462 (inline_transform): Rescale profile before inlining.
27464 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27466 * cgraph.h (cgraph_edge::clone): Update prototype.
27467 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27468 (cgraph_node::create_clone): Update.
27469 (cgraph_node::create_version_clone): Update.
27470 * tree-inline.c (copy_bb): Update.
27471 (expand_call_inline): Update.
27473 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27475 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27476 factored out from ...
27477 (rs6000_emit_prologue): ... here.
27479 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27481 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27482 factored out from ...
27483 (rs6000_emit_prologue): ... here.
27485 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27487 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27489 (handle_missing_profiles): Fix computation of tp_first_run.
27490 (counts_to_freqs): Do not touch freqs when count is 0.
27492 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27494 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27497 2017-06-10 Tom de Vries <tom@codesourcery.com>
27499 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27500 attributes): Document signal effective target.
27502 2017-06-10 Tom de Vries <tom@codesourcery.com>
27504 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27505 Document effective target stack_size.
27507 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27509 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27510 to the edit_context if they can be auto-applied.
27512 2017-06-9 Ian Lance Taylor <iant@golang.org>
27514 * opts.c (finish_options): If -fsplit-stack, disable implicit
27515 -forder-blocks-and-partition.
27516 * doc/invoke.texi (Optimize Options): Document that when using
27517 -fsplit-stack -forder-blocks-and-partition is not implicitly
27520 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27522 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27523 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27524 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27525 * builtins.def (abort, trap, unreachable): Declare cold.
27526 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27527 * tree-core.h (ECF_COLD): New.
27528 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27529 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27531 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27533 * predict.c (unlikely_executed_stmt_p): Cleanup.
27535 2017-06-09 Richard Biener <rguenther@suse.de>
27537 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27538 model if the ref is always written to.
27540 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27542 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27544 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27546 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27549 2017-06-09 Tom de Vries <tom@codesourcery.com>
27552 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27553 "target cannot support label values" when encountering LABEL_REF.
27555 2017-06-09 Martin Liska <mliska@suse.cz>
27557 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27558 (gimple_gen_ic_func_profiler): Emit direct comparison
27559 of __gcov_indirect_call_callee with NULL.
27560 (gimple_gen_time_profiler): Change probability from
27561 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27563 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27565 * profile.c (edge_gcov_counts): Turn to pointer.
27566 (compute_branch_probabilities, compute_branch_probabilities): Update.
27567 (branch_prob): Do not clear edge_gcov_count.
27568 * profile.h (edge_gcov_counts): Turn to pointer.
27569 (edge_gcov_count): Update.
27571 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27573 * gimple.h (gimple_check_failed): Mark cold.
27575 2017-06-09 Richard Biener <rguenther@suse.de>
27577 PR tree-optimization/66623
27578 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27579 refactor check_reduction into two parts, properly computing
27580 whether we have to check reduction validity for outer loop
27583 2017-06-09 Richard Biener <rguenther@suse.de>
27585 PR tree-optimization/79483
27586 * graphite-scop-detection.c (order): New global.
27587 (get_order): Compute bb to order mapping that satisfies code
27588 generation constraints.
27589 (cmp_pbbs): New helper.
27590 (build_scops): Start domwalk at entry block, sort generated
27593 2017-06-09 Richard Biener <rguenther@suse.de>
27595 PR middle-end/81007
27596 * ipa-polymorphic-call.c
27597 (ipa_polymorphic_call_context::restrict_to_inner_class):
27598 Skip FIELD_DECLs with error_mark_node type.
27599 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27602 2017-06-09 Martin Liska <mliska@suse.cz>
27604 * predict.c (struct branch_predictor): New struct.
27605 (test_prediction_value_range): New test.
27606 (predict_c_tests): New function.
27607 * selftest-run-tests.c (selftest::run_tests): Run the function.
27608 * selftest.h: Declare new tests.
27610 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27613 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27614 gen_add3_insn did not fail.
27615 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27616 r0, construct that number in a temporary reg and add that reg to r0.
27617 If asked to put the result in r0 as well, fail.
27619 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27621 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27622 for early expansion of vec_eqv.
27624 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27626 PR middle-end/81005
27627 * ubsan.c (instrument_null): Avoid pointless code temporary.
27628 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27630 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27634 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27637 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27640 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27642 * predict.c (unlikely_executed_edge_p): Move ahead.
27643 (probably_never_executed_edge_p): Use it.
27645 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27647 PR middle-end/79988
27648 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27649 gimple_call_builtin_p call.
27651 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27653 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27654 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27655 rtl_check_failed_type2, rtl_check_failed_code1,
27656 rtl_check_failed_code2, rtl_check_failed_code_mode,
27657 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27658 rtvec_check_failed_bounds, rtl_check_failed_flag,
27659 _fatal_insn_not_found, _fatal_insn): Likewise.
27660 * tree.h (tree_contains_struct_check_failed,
27661 tree_check_failed, tree_not_check_failed,
27662 tree_class_check_failed, tree_range_check_failed,
27663 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27664 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27665 tree_operand_check_failed, omp_clause_check_failed,
27666 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27669 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27671 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27672 flag_branch_probabilities.
27673 * ipa-inline.c (edge_badness): Likewise.
27674 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27675 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27676 * predict.c (maybe_hot_frequency_p): Likewise.
27677 (probably_never_executed): Likewise.
27678 * sched-ebb.c (schedule_ebbs): Likewise.
27679 * sched-rgn.c (find_single_block_region): Likewise.
27680 * tracer.c (tail_duplicate): Likewise.
27682 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27684 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27685 longer requires x_flag_profile_use.
27687 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27689 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27690 instead of flag_reorder_blocks_and_partition.
27691 * dbxout.c (dbxout_function_end): Likewise.
27692 * dwarf2out.c (gen_subprogram_die): Likewise.
27693 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27694 * hw-doloop.c (reorg_loops): Likewise.
27695 * varasm.c (assemble_start_function,
27696 assemble_end_function): Likewise.
27697 (decide_function_section): Do not check for
27698 flag_reorder_blocks_and_partition.
27700 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27702 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27704 (chkp_get_hard_register_fake_addr_expr): Ditto.
27705 (chkp_build_addr_expr): Add check for hard reg case.
27706 (chkp_parse_array_and_component_ref): Ditto.
27707 (chkp_find_bounds_1): Ditto.
27708 (chkp_process_stmt): Don't generate bounds store for
27711 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27713 * predict.c (maybe_hot_bb_p): Do not check profile status.
27714 (maybe_hot_edge_p): Likewise.
27715 (probably_never_executed): Check for zero counts even if profile
27717 (unlikely_executed_edge_p): New function.
27718 (unlikely_executed_stmt_p): New function.
27719 (unlikely_executed_bb_p): New function.
27720 (set_even_probabilities): Use unlikely predicates.
27721 (combine_predictions_for_bb): Likewise.
27722 (predict_paths_for_bb): Likewise.
27723 (predict_paths_leading_to_edge): Likewise.
27724 (determine_unlikely_bbs): New function.
27725 (estimate_bb_frequencies): Use it.
27726 (compute_function_frequency): Use zero counts even if profile is
27728 * profile-count.h: Fix typo.
27730 2017-08-08 Julia Koval <julia.koval@intel.com>
27732 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27733 _mm512_mask_cvtsepi16_storeu_epi8,
27734 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27735 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27736 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27737 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27738 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27739 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27740 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27741 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27742 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27743 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27744 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27745 __builtin_ia32_pmovuswb256mem_mask,
27746 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27747 __builtin_ia32_pmovwb512mem_mask): New builtins.
27749 2017-08-08 Julia Koval <julia.koval@intel.com>
27751 PR target/73350,80862
27752 * config/i386/subst.md (round): Fix round pattern.
27753 * config/i386/i386.c (ix86_erase_embedded_rounding):
27754 Fix erasing rounding for the fixed pattern.
27756 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27758 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27760 2017-06-08 Martin Liska <mliska@suse.cz>
27762 PR gcov-profile/80911
27763 * gcov.c (block_info::block_info): New constructor.
27765 2017-06-07 Carl Love <cel@us.ibm.com>
27767 * config/rs6000/rs6000-c: The return type of the following
27768 built-in functions was implemented as int not long long. Fix sign
27769 of return value for the unsigned version of vec_mulo and vec_mule.
27770 vector unsigned long long vec_bperm (vector unsigned long long,
27771 vector unsigned char)
27772 vector signed long long vec_mule (vector signed int,
27774 vector unsigned long long vec_mule (vector unsigned int,
27775 vector unsigned int)
27776 vector signed long long vec_mulo (vector signed int,
27778 vector unsigned long long vec_mulo (vector unsigned int,
27779 vector unsigned int)
27780 * doc/extend.texi: Fix the documentation for the built-in
27783 2017-06-07 Carl Love <cel@us.ibm.com>
27786 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27789 2017-06-07 Carl Love <cel@us.ibm.com>
27791 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27792 support, Generate doublehv for signed int/float for BE case only.
27794 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27796 * doc/invoke.texi (mcx16): Rewrite.
27798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27800 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27801 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27802 *mov<mode>_softfloat, and an anonymous splitter): Use
27803 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27807 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27809 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27810 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27811 (rs6000_debug_reg_global): Adjust.
27812 (rs6000_init_hard_regno_mode_ok): Adjust.
27813 (rs6000_dbx_register_number): Adjust.
27814 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27815 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27816 Remove SPE_ACC and SPEFSCR.
27817 (REG_ALLOC_ORDER): Ditto.
27818 (FRAME_POINTER_REGNUM): Change to 111.
27819 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27820 (REG_CLASS_NAMES): Ditto.
27821 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27822 (REGISTER_NAMES): Ditto.
27823 (ADDITIONAL_REG_NAMES): Ditto.
27824 (rs6000_reg_names): Ditto.
27825 * config/rs6000/rs6000.md: Renumber some register number
27828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27830 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27832 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27833 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27835 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27836 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27837 Delete the SPE high registers.
27838 (REG_ALLOC_ORDER): Ditto.
27839 (enum reg_class): Remove SPE_HIGH_REGS.
27840 (REG_CLASS_NAMES): Ditto.
27841 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27842 (REGISTER_NAMES): Ditto.
27843 (rs6000_reg_names): Ditto.
27844 * doc/tm.texi.in: Remove SPE as example.
27845 * doc/tm.texi: Regenerate.
27847 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27849 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27850 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27851 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27852 * config/rs6000/rs6000.md (type): Remove "brinc".
27854 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27856 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27857 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27858 * config/rs6000/linuxspe.h: Delete file.
27859 * config/rs6000/rs6000.md: Don't include spe.md.
27860 * config/rs6000/spe.h: Delete file.
27861 * config/rs6000/spe.md: Delete file.
27862 * config/rs6000/t-rs6000: Remove spe.md.
27864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27866 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27867 (reg_or_none500mem_operand): Delete.
27868 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27869 instead of reg_or_none500mem_operand.
27871 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27874 handling of SPE flags.
27875 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27877 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27879 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27881 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27882 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27883 paired_divv2sf3): Similar.
27884 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27885 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27886 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27888 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27889 Rename the paired_* instruction patterns.
27890 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27892 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27893 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27894 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27895 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27896 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27897 PAIRED_VECTOR_MODE.
27898 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27899 (spe_func_has_64bit_regs_p): Delete.
27900 (spe_expand_predicate_builtin): Delete.
27901 (spe_expand_evsel_builtin): Delete.
27902 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27903 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27904 (invalid_e500_subreg): Delete.
27905 (rs6000_legitimize_address): Always force_reg op2 as well, for
27906 paired single memory accesses.
27907 (rs6000_member_type_forces_blk): Delete.
27908 (rs6000_spe_function_arg): Delete.
27909 (rs6000_expand_unop_builtin): Delete SPE handling.
27910 (rs6000_expand_binop_builtin): Ditto.
27911 (spe_expand_stv_builtin): Delete.
27912 (bdesc_2arg_spe): Delete.
27913 (spe_expand_builtin): Delete.
27914 (spe_expand_predicate_builtin): Delete.
27915 (spe_expand_evsel_builtin): Delete.
27916 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27917 (spe_init_builtins): Delete.
27918 (spe_func_has_64bit_regs_p): Delete.
27919 (savres_routine_name): Delete "info" parameter. Adjust callers.
27920 (rs6000_emit_stack_reset): Ditto.
27921 (rs6000_dwarf_register_span): Delete.
27922 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27923 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27924 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27926 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27928 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27929 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27930 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27931 mulv2sf3, divv2sf3): Delete expanders.
27933 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27935 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27937 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27939 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27940 * config/rs6000/rs6000.c: Ditto.
27942 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27944 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27945 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27946 comparison_operator.
27948 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27950 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27951 * config/rs6000/rs6000.opt: Ditto.
27952 * config/rs6000/t-rtems: Ditto.
27954 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27956 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27957 TARGET_E500_SINGLE by 0, simplify.
27958 * config/rs6000/rs6000.c: Ditto.
27959 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27960 (spe_build_register_parallel): Delete.
27961 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27962 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27963 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27964 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27965 (E500_CONVERT): Delete.
27966 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27968 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27970 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27971 * config/rs6000/dfp.md: Ditto.
27972 (negdd2, *negdd2_fpr): Merge.
27973 (absdd2, *absdd2_fpr): Merge.
27974 (negtd2, *negtd2_fpr): Merge.
27975 (abstd2, *abstd2_fpr): Merge.
27976 * config/rs6000/e500.h: Delete file.
27977 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27978 TARGET_FPRS by 1 and simplify.
27979 * config/rs6000/rs6000-c.c: Ditto.
27980 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27981 TARGET_DF_SPE by 0.
27982 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27984 * config/rs6000/rs6000.md: Ditto.
27985 (floatdidf2, *floatdidf2_fpr): Merge.
27986 (move_from_CR_gt_bit): Delete.
27987 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27988 (E500_CR_IOR_COMPARE): Delete.
27989 (All patterns that require !TARGET_FPRS): Delete.
27990 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27992 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27994 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27996 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27998 * graphds.c (add_edge): Intitialize edge's attached data.
27999 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28000 pointer parameter. Call pointed function on each edge during
28001 graph traversing. Skip traversing the edge when the function
28003 (graphds_dfs, graphds_scc): Ditto.
28004 (for_each_edge): New parameter. Pass the new parameter to callback
28006 * graphds.h (skip_edge_callback): New function pointer type.
28007 (graphds_dfs, graphds_scc): New function pointer parameter.
28008 (graphds_edge_callback, for_each_edge): New parameter.
28010 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28012 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28013 out code checking if runtime alias check is possible to below ...
28014 Call the new function.
28015 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28016 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28018 2017-06-07 Marek Polacek <polacek@redhat.com>
28021 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28022 TYPE_OVERFLOW_WRAPS check.
28024 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28026 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28027 if versioning is required.
28028 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28029 peeling with the check for versioning.
28031 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28033 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28034 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28035 Set true to new parameter if new ssa variable is defined.
28036 (vect_gen_vector_loop_niters): Refactor. Set range information
28037 for the new vector loop bound variable.
28038 (vect_do_peeling): Ditto.
28040 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28042 * tree-affine.c (ssa.h): Include header file.
28043 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28044 has wrapping overflow behavior.
28046 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28048 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28050 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28052 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28053 (tree_to_aff_combination): ... here.
28055 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28057 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28058 reg_pressure model function.
28059 (ivopts_global_cost_for_size): Delete.
28060 (determine_set_costs, iv_ca_recount_cost): Call new model function
28061 ivopts_estimate_reg_pressure.
28063 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28065 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28066 expensive than udiv. Remove floating point cases from mod.
28068 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28070 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28071 Increase idiv cost.
28073 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28075 * config/aarch64/aarch64.md
28076 (copysignsf3): Fix mask generation.
28078 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28080 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28082 (class dump_manager): Add register_dumps method.
28083 * dumpfile.c: Include langhooks.h.
28084 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28085 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28086 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28087 (dump_manager::dump_register): Start with 512 entries instead of 32.
28088 (dump_manager::register_dumps): New method.
28089 * toplev.c (general_init): Instead of invoking register_dumps
28090 langhook, invoke register_dumps method on the dump manager.
28091 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28094 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28096 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28097 Say that # requires an associated define_split to exist, and that
28098 the define_split must be suitable for use after register allocation.
28100 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28102 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28103 (compute_outgoing_frequencies): Also initialize zero counts.
28104 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28105 around loops; preserve more of profile when nothing changes.
28107 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28109 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28111 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28112 * config/arm/arm-cpu-cdata.h: Regenerate.
28113 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28114 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28115 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28116 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28117 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28119 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28121 * config/arm/t-rmprofile: Likewise.
28122 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28124 2017-06-06 David S. Miller <davem@davemloft.net>
28127 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28128 blockage if function uses alloca.
28130 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28132 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28133 New "uid" fields to hold pretty-print IDs of group and ref.
28134 Memory references are now identified as <group_id>:<ref_id>
28135 instead of using [random] addresses.
28136 (dump_mem_details): Simplify, no functional change.
28137 (dump_mem_ref): Simplify and make output more concise.
28138 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28139 (find_or_create_group): Initialize group uid.
28140 (record_ref): Initialize ref uid. Improve debug output.
28141 (prune_group_by_reuse, should_issue_prefetch_p,)
28142 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28143 (mark_nontemporal_store, determine_loop_nest_reuse):
28144 Improve debug output.
28146 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28148 * dbgcnt.def (prefetch): New debug counter.
28149 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28150 (schedule_prefetches): Stop issueing prefetches if debug counter
28153 2017-06-06 Tom de Vries <tom@codesourcery.com>
28155 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28156 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28158 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28160 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28161 Use aarch64_reg_or_zero predicate for operand 4.
28162 (aarch64_compare_and_swap<mode> define_insn_and_split):
28163 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28164 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28166 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28168 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28169 (arm_compute_save_core_reg_mask): This.
28170 (thumb1_compute_save_reg_mask): Rename into ...
28171 (thumb1_compute_save_core_reg_mask): This.
28172 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28173 (arm_compute_frame_layout): Likewise.
28175 2017-06-06 Richard Biener <rguenther@suse.de>
28177 PR tree-optimization/80974
28178 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28179 keep or clear leaders SSA info.
28181 2017-06-06 Tom de Vries <tom@codesourcery.com>
28183 * config/nvptx/nvptx.c (split_mode_p): New function.
28184 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28186 2017-06-06 Tom de Vries <tom@codesourcery.com>
28188 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28190 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28193 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28196 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28198 * shrink-wrap.c (handle_simple_exit): Update profile.
28199 (try_shrink_wrapping): Upate profile.
28201 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28203 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28204 (tree_guess_outgoing_edge_probabilities): New.
28205 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28206 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28208 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28210 * ipa-split.c (split_function): Initialize return bb profile.
28212 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28214 * profile.c (compute_branch_probabilities): Also initialize
28215 EXIT_BLOCK profile.
28217 2017-06-06 Richard Biener <rguenther@suse.de>
28219 PR tree-optimization/80928
28220 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28221 (vect_analyze_loop_operations): Properly guard analysis for
28223 (vect_transform_loop): Likewise.
28224 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28225 (vect_model_induction_cost): Do not cost for pure SLP.
28226 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28227 of induction in inner loop vectorization.
28228 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28229 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28230 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28232 (vect_analyze_slp_cost_1): Cost induction.
28233 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28234 (vect_get_slp_vect_defs): Likewise.
28235 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28236 (vect_transform_stmt): Handle SLP reductions.
28237 * tree-vectorizer.h (vectorizable_induction): Adjust.
28239 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28241 * config/rs6000/rs6000.c (make_resolver_func): Update
28242 init_lowered_empty_function call.
28244 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28246 * doc/invoke.texi: Document the -fprofile-abs-path option.
28247 * common.opt (fprofile-abs-path): New option.
28248 * gcov-io.h (gcov_write_filename): Declare.
28249 * gcov-io.c (gcov_write_filename): New function.
28250 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28251 * profile.c (output_location): Likewise.
28253 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28255 * shring-wrap.c: Revert accidental commit.
28257 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28259 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28261 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28263 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28265 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28266 profile in callgraph edge.
28267 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28268 (apply_scale): Likewise.
28269 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28270 Also scale profile when inlining function with zero profile.
28271 (initialize_cfun): Update exit block profile even when it is zero.
28272 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28273 when profile is read.
28275 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28277 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28278 (CLONE_*): New constants to define the processors we can generate
28279 code for with the target_clone attribute.
28280 (rs6000_clone_map): New array to identify which clone processors
28281 the current program is running on.
28282 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28283 target_clone attribute.
28284 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28285 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28286 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28287 (cpu_expand_builtin): Add support for target_clone attribute.
28288 (rs6000_valid_attribute_p): Allow "default" attribute.
28289 (get_decl_name): New debug function to simplify printing the
28290 current function name in debugging statements.
28291 (rs6000_clone_priority): New functions to support the target_clone
28292 attribute, and be able to generate code to switch between ISA 2.05
28293 through ISA 3.0 (power6 through power9).
28294 (rs6000_compare_version_priority): Likewise.
28295 (rs6000_get_function_versions_dispatcher): Likewise.
28296 (make_resolver_func): Likewise.
28297 (add_condition_to_bb): Likewise.
28298 (dispatch_function_versions): Likewise.
28299 (rs6000_generate_version_dispatcher_body): Likewise.
28300 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28301 (fusion_gpr_load_p): Fix a spacing issue.
28302 * doc/extend.texi (Common Function Attributes): Document that the
28303 PowerPC supports the target_clone attribute.
28305 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28307 * config/arm/arm.h: explain F symbol found in description of ARM
28308 register allocation in its legend.
28310 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28312 * config/mips/frame-header-opt.c: Include profile-count.h.
28313 * config/riscv/riscv.c: Include profile-count.h
28315 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28317 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28319 (sm_set_flag_if_changed): Add bbs field.
28320 (execute_sm_if_changed_flag_set): Pass BBS.
28321 (execute_sm): Update.
28323 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28325 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28328 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28330 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28331 (peephole2): New peephole2 to emit the above.
28332 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28334 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28336 * config/aarch64/aarch64.c (define_peephole2 above
28337 *sub_<shift>_<mode>): New peephole.
28339 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28341 * config/i386/i386.c (make_resolver_func): Update.
28342 * Makefile.in: Add profile-count.h and profile-count.o
28343 * auto-profile.c (afdo_indirect_call): Update to new API.
28344 (afdo_set_bb_count): Update.
28345 (afdo_propagate_edge): Update.
28346 (afdo_propagate_circuit): Update.
28347 (afdo_calculate_branch_prob): Update.
28348 (afdo_annotate_cfg): Update.
28349 * basic-block.h: Include profile-count.h
28350 (struct edge_def): Turn count to profile_count.
28351 (struct basic_block_def): Likewie.
28352 (REG_BR_PROB_BASE): Move to profile-count.h
28353 (RDIV): Move to profile-count.h
28354 * bb-reorder.c (max_entry_count): Turn to profile_count.
28355 (find_traces): Update.
28356 (rotate_loop):Update.
28357 (connect_traces):Update.
28358 (sanitize_hot_paths):Update.
28359 * bt-load.c (migrate_btr_defs): Update.
28360 * cfg.c (RDIV): Remove.
28361 (init_flow): Use alloc_block.
28362 (alloc_block): Uninitialize count.
28363 (unchecked_make_edge): Uninitialize count.
28364 (check_bb_profile): Update.
28365 (dump_edge_info): Update.
28366 (dump_bb_info): Update.
28367 (update_bb_profile_for_threading): Update.
28368 (scale_bbs_frequencies_int): Update.
28369 (scale_bbs_frequencies_gcov_type): Update.
28370 (scale_bbs_frequencies_profile_count): New.
28371 * cfg.h (update_bb_profile_for_threading): Update.
28372 (scale_bbs_frequencies_profile_count): Declare.
28373 * cfgbuild.c (compute_outgoing_frequencies): Update.
28374 (find_many_sub_basic_blocks): Update.
28375 * cfgcleanup.c (try_forward_edges): Update.
28376 (try_crossjump_to_edge): Update.
28377 * cfgexpand.c (expand_gimple_tailcall): Update.
28378 (construct_exit_block): Update.
28379 * cfghooks.c (verify_flow_info): Update.
28380 (dump_bb_for_graph): Update.
28381 (split_edge): Update.
28382 (make_forwarder_block): Update.
28383 (duplicate_block): Update.
28384 (account_profile_record): Update.
28385 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28386 (get_estimated_loop_iterations): Update.
28387 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28388 (single_likely_exit): Update.
28389 * cfgloopmanip.c (scale_loop_profile): Update.
28391 (set_zero_probability): Update.
28392 (lv_adjust_loop_entry_edge): Update.
28393 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28394 (purge_dead_edges): Update.
28395 (rtl_account_profile_record): Update.
28396 * cgraph.c (cgraph_node::create): Uninitialize count.
28397 (symbol_table::create_edge): Uninitialize count.
28398 (cgraph_update_edges_for_call_stmt_node): Update.
28399 (cgraph_edge::dump_edge_flags): Update.
28400 (cgraph_node::dump): Update.
28401 (cgraph_edge::maybe_hot_p): Update.
28402 * cgraph.h: Include profile-count.h
28403 (create_clone), create_edge, create_indirect_edge): Update.
28404 (cgraph_node): Turn count to profile_count.
28405 (cgraph_edge0: Likewise.
28406 (make_speculative, clone): Update.
28407 (create_edge): Update.
28408 (init_lowered_empty_function): Update.
28409 * cgraphclones.c (cgraph_edge::clone): Update.
28410 (duplicate_thunk_for_node): Update.
28411 (cgraph_node::create_clone): Update.
28412 * cgraphunit.c (cgraph_node::analyze): Update.
28413 (cgraph_node::expand_thunk): Update.
28414 * final.c (dump_basic_block_info): Update.
28415 * gimple-streamer-in.c (input_bb): Update.
28416 * gimple-streamer-out.c (output_bb): Update.
28417 * graphite.c (print_global_statistics): Update.
28418 (print_graphite_scop_statistics): Update.
28419 * hsa-brig.c: Include basic-block.h.
28420 * hsa-dump.c: Include basic-block.h.
28421 * hsa-gen.c (T sum_slice): Update.
28422 (convert_switch_statements):Update.
28423 * hsa-regalloc.c: Include basic-block.h.
28424 * ipa-chkp.c (chkp_produce_thunks): Update.
28425 * ipa-cp.c (struct caller_statistics): Update.
28426 (init_caller_stats): Update.
28427 (gather_caller_stats): Update.
28428 (ipcp_cloning_candidate_p): Update.
28429 (good_cloning_opportunity_p): Update.
28430 (get_info_about_necessary_edges): Update.
28431 (dump_profile_updates): Update.
28432 (update_profiling_info): Update.
28433 (update_specialized_profile): Update.
28434 (perhaps_add_new_callers): Update.
28435 (decide_about_value): Update.
28436 (ipa_cp_c_finalize): Update.
28437 * ipa-devirt.c (struct odr_type_warn_count): Update.
28438 (struct decl_warn_count): Update.
28439 (struct final_warning_record): Update.
28440 (possible_polymorphic_call_targets): Update.
28441 (ipa_devirt): Update.
28442 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28443 * ipa-icf.c (sem_function::merge): Update.
28444 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28445 * ipa-inline.c (compute_uninlined_call_time): Update.
28446 (compute_inlined_call_time): Update.
28447 (want_inline_small_function_p): Update.
28448 (want_inline_self_recursive_call_p): Update.
28449 (edge_badness): Update.
28450 (lookup_recursive_calls): Update.
28451 (recursive_inlining): Update.
28452 (inline_small_functions): Update.
28453 (dump_overall_stats): Update.
28454 (dump_inline_stats): Update.
28455 * ipa-profile.c (ipa_profile_generate_summary): Update.
28456 (ipa_propagate_frequency): Update.
28457 (ipa_profile): Update.
28458 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28459 * ipa-utils.c (ipa_merge_profiles): Update.
28460 * loop-doloop.c (doloop_modify): Update.
28461 * loop-unroll.c (report_unroll): Update.
28462 (unroll_loop_runtime_iterations): Update.
28463 * lto-cgraph.c (lto_output_edge): Update.
28464 (lto_output_node): Update.
28465 (input_node): Update.
28466 (input_edge): Update.
28467 (merge_profile_summaries): Update.
28468 * lto-streamer-in.c (input_cfg): Update.
28469 * lto-streamer-out.c (output_cfg): Update.
28470 * mcf.c (create_fixup_graph): Update.
28471 (adjust_cfg_counts): Update.
28472 (sum_edge_counts): Update.
28473 * modulo-sched.c (sms_schedule): Update.
28474 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28475 * predict.c (maybe_hot_count_p): Update.
28476 (probably_never_executed): Update.
28477 (dump_prediction): Update.
28478 (combine_predictions_for_bb): Update.
28479 (propagate_freq): Update.
28480 (handle_missing_profiles): Update.
28481 (counts_to_freqs): Update.
28482 (rebuild_frequencies): Update.
28483 (force_edge_cold): Update.
28484 * predict.h: Include profile-count.h
28485 (maybe_hot_count_p, counts_to_freqs): UPdate.
28486 * print-rtl-function.c: Do not include cfg.h
28487 * print-rtl.c: Include basic-block.h
28488 * profile-count.c: New file.
28489 * profile-count.h: New file.
28490 * profile.c (is_edge_inconsistent): Update.
28491 (correct_negative_edge_counts): Update.
28492 (is_inconsistent): Update.
28493 (set_bb_counts): Update.
28494 (read_profile_edge_counts): Update.
28495 (compute_frequency_overlap): Update.
28496 (compute_branch_probabilities): Update; Initialize and deinitialize
28498 (branch_prob): Update.
28499 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28500 (edge_gcov_count): New.
28501 (bb_gcov_count): New.
28502 * shrink-wrap.c (try_shrink_wrapping): Update.
28503 * tracer.c (better_p): Update.
28504 * trans-mem.c (expand_transaction): Update.
28505 (ipa_tm_insert_irr_call): Update.
28506 (ipa_tm_insert_gettmclone_call): Update.
28507 * tree-call-cdce.c: Update.
28508 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28509 (gimple_duplicate_sese_tail): Update.
28510 (gimple_account_profile_record): Update.
28511 (execute_fixup_cfg): Update.
28512 * tree-inline.c (copy_bb): Update.
28513 (copy_edges_for_bb): Update.
28514 (initialize_cfun): Update.
28515 (freqs_to_counts): Update.
28516 (copy_cfg_body): Update.
28517 (expand_call_inline): Update.
28518 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28519 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28520 (try_unroll_loop_completely): Update.
28521 (try_peel_loop): Update.
28522 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28523 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28524 * tree-ssa-loop-split.c (connect_loops): Update.
28525 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28526 * tree-ssa-reassoc.c (branch_fixup): Update.
28527 * tree-ssa-tail-merge.c (replace_block_by): Update.
28528 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28529 (compute_path_counts): Update.
28530 (update_profile): Update.
28531 (recompute_probabilities): Update.
28532 (update_joiner_offpath_counts): Update.
28533 (estimated_freqs_path): Update.
28534 (freqs_to_counts_path): Update.
28535 (clear_counts_path): Update.
28536 (ssa_fix_duplicate_block_edges): Update.
28537 (duplicate_thread_path): Update.
28538 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28539 (struct switch_conv_info): Update.
28540 * tree-tailcall.c (decrease_profile): Update.
28541 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28542 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28543 * value-prof.c (check_counter): Update.
28544 (gimple_divmod_fixed_value): Update.
28545 (gimple_mod_pow2): Update.
28546 (gimple_mod_subtract): Update.
28547 (gimple_ic_transform): Update.
28548 (gimple_stringop_fixed_value): Update.
28549 * value-prof.h (gimple_ic): Update.
28551 2017-06-02 Carl Love <cel@us.ibm.com>
28553 * config/rs6000/rs6000-c: Add support for built-in functions
28554 vector double vec_doublee (vector signed int);
28555 vector double vec_doublee (vector unsigned int);
28556 vector double vec_doublee (vector float);
28557 vector double vec_doubleh (vector signed int);
28558 vector double vec_doubleh (vector unsigned int);
28559 vector double vec_doubleh (vector float);
28560 vector double vec_doublel (vector signed int);
28561 vector double vec_doublel (vector unsigned int);
28562 vector double vec_doublel (vector float);
28563 vector double vec_doubleo (vector signed int);
28564 vector double vec_doubleo (vector unsigned int);
28565 vector double vec_doubleo (vector float);.
28566 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28567 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28569 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28570 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28571 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28573 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28574 vec_doublel, vec_doubleh.
28575 * doc/extend.texi: Update the built-in documentation file for the
28576 new built-in functions.
28578 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28581 * ipa-inline-analysis.c (free_growth_caches): Set
28582 edge_removal_hook_holder to NULL after removing it.
28584 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28586 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28587 comparision with zero.
28589 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28590 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28591 for early expansion of vec_min and vec_max builtins.
28592 (builtin_function_type): Add min/max unsigned variants to those
28593 identified as having unsigned arguments.
28595 2017-06-02 Olivier Hainque <hainque@adacore.com>
28597 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28599 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28601 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28602 Use VALL_F16 iterator rather than VALL.
28604 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28606 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28607 Emit CBNZ inside loop when doing a strong exchange and comparing
28608 against zero. Generate the CC flags after the loop.
28610 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28612 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28613 (dl_section_ref): New.
28614 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28615 On AIX, append an expression to subtract the size of the
28616 section length to dl_section_ref.
28618 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28621 for early expansion of vector absolute builtins.
28623 2017-06-02 Richard Biener <rguenther@suse.de>
28625 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28626 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28628 2017-06-02 Richard Biener <rguenther@suse.de>
28630 PR tree-optimization/80948
28631 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28632 stmt order as well.
28634 2017-06-02 Richard Biener <rguenther@suse.de>
28636 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28638 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28639 for inductions as relevant.
28641 2017-06-02 Richard Biener <rguenther@suse.de>
28643 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28644 (vectorizable_induction): ... this. Remove dead code.
28646 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28648 * builtins. (expand_builtin_alloca): Remove second parameter and
28649 infer its value from the first parameter instead.
28650 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28652 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28654 PR rtl-optimization/80903
28655 * loop-doloop.c (add_test): Unshare sequence.
28657 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28659 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28661 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28663 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28665 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28666 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28667 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28668 (xlogue_layout::get_stub_rtx): Make static.
28669 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28670 (xlogue_layout::compute_stub_managed_regs): Rename to...
28671 (xlogue_layout::count_stub_managed_regs): ...this.
28672 (xlogue_layout::is_stub_managed_reg): New function.
28673 (xlogue_layout::m_stub_names): Rename to...
28674 (xlogue_layout::s_stub_names): ...this, make static.
28675 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28676 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28677 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28678 xlogue_layout::s_stub_names): Instantiate statics.
28679 (stub_managed_regs): Remove.
28680 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28681 (disable_call_ms2sysv_xlogues): Rename to...
28682 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28683 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28685 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28686 change after reload_completed.
28687 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28689 (ix86_expand_prologue): Likewise.
28690 (ix86_expand_epilogue): Likewise.
28691 (ix86_expand_split_stack_prologue): Likewise.
28692 (ix86_compute_frame_layout): Remove frame parameter ...
28693 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28694 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28696 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28697 (ix86_frame): Move from here ...
28698 * config/i386/i386.h (ix86_frame): ... to here.
28699 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28700 complete ix86_frame data structure instead. Remove some_ld_name.
28702 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28704 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28705 symbols that hold a DECL_VALUE_EXPR.
28707 2017-06-01 Martin Jambor <mjambor@suse.cz>
28709 PR tree-optimization/80898
28710 * tree-sra.c (process_subtree_disqualification): Removed.
28711 (disqualify_candidate): Do not acll
28712 process_subtree_disqualification.
28713 (subtree_mark_written_and_enqueue): New function.
28714 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28715 RHS has been disqualified and re-queue LHS if necessary. Apart
28716 from that, ignore disqualified RHS.
28718 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28720 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28721 address fetch for z10 or later.
28723 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28725 * config/arc/arc.md (tst_movb): Add guard when splitting.
28727 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28729 * config/arc/arc.c (arc_can_eliminate): Test against
28730 arc_frame_pointer_needed.
28732 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28734 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28735 to prevent store reordering.
28736 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28737 (type): Add block type.
28738 (stack_tie): Define special instruction to be used in
28741 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28743 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28744 constraint. It is not valid for the pattern.
28745 (noncommutative_binary_comparison): Likewise.
28747 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28749 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28752 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28754 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28755 be used by the reg-alloc.
28757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28759 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28760 reg-alloc when having mul64 or mul32x16 instructions.
28761 (mulsidi3): Likewise.
28762 (umulsidi3): Likewise.
28763 (mulsi32x16): New pattern.
28764 (mulsi64): Likewise.
28765 (mulsidi64): Likewise.
28766 (umulsidi64): Likewise.
28767 (MUL32x16_REG): Define.
28768 (mul64_600): Use MUL32x16_REG.
28769 (mac64_600): Likewise.
28770 (umul64_600): Likewise.
28771 (umac64_600): Likewise.
28773 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28775 * config/arc/arc.md (mulsi3_700): Make it commutative.
28777 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28779 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28781 (*sign_extendsidi2_insn): Likewise for movstosw.
28783 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28785 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28786 the type of the input discriminant value. Convert the
28787 discriminant value of signedness vary.
28789 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28791 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28792 Add to -Wall section.
28794 2017-06-01 Richard Biener <rguenther@suse.de>
28796 PR middle-end/66313
28797 * fold-const.c (fold_plusminus_mult_expr): If the factored
28798 factor may be zero use a wrapping type for the inner operation.
28799 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28800 and handle moved defs.
28801 (process_assignment): Properly guard the unary op case. Return a
28802 tri-state indicating that moving the stmt before the call may allow
28803 to continue. Pass through to_move.
28804 (find_tail_calls): Handle moving unrelated defs before
28807 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28810 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28811 splitter result in the canonical way.
28813 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28815 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28816 also for 32bit target. Update insn attributes.
28817 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28819 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28821 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28822 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28823 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28824 (_mm_maskz_min_ss): New intrinsics.
28826 2017-05-31 Martin Liska <mliska@suse.cz>
28828 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28829 Change comment style to one we normally use.
28830 (vectorizable_reduction): Likewise.
28831 (vectorizable_induction): Likewise.
28832 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28833 (vectorizable_call): Likewise.
28834 (vectorizable_simd_clone_call): Likewise.
28835 (vectorizable_conversion): Likewise.
28836 (vectorizable_assignment): Likewise.
28837 (vectorizable_shift): Likewise.
28838 (vectorizable_operation): Likewise.
28839 (vectorizable_store): Likewise.
28840 (vectorizable_load): Likewise.
28841 * tree-vectorizer.h: Likewise.
28843 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28845 * passes.c (emergency_dump_function): New.
28846 * tree-pass.h (emergency_dump_function): Declare.
28847 * plugin.c (plugins_internal_error_function): Remove.
28848 * plugin.h (plugins_internal_error_function): Remove declaration.
28849 * toplev.c (internal_error_function): New static function. Use it...
28850 (general_init): ...here.
28852 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28854 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28855 (arc_rtx_costs): Add costs for new patterns.
28856 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28857 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28859 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28861 * tree-ssa-strlen.c (get_next_strinfo): New function.
28862 (get_stridx_plus_constant): Use it.
28863 (zero_length_string): Likewise.
28864 (adjust_related_strinfos): Likewise.
28865 (adjust_last_stmt): Likewise.
28867 2017-05-31 Richard Biener <rguenther@suse.de>
28870 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28871 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28873 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28875 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28876 loop_vinfo argument and use of dependence distance vectors.
28877 Check instead whether the two references differ only in their
28878 initial value and assume that they have the same alignment if the
28879 difference is a multiple of the vector alignment.
28880 (vect_analyze_data_refs_alignment): Update call accordingly.
28882 2017-05-31 Martin Liska <mliska@suse.cz>
28885 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28887 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28889 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28890 (create_intersect_range_checks): Move from ...
28891 * tree-data-ref.c (create_intersect_range_checks_index)
28892 (create_intersect_range_checks): ... to here.
28893 (create_runtime_alias_checks): New function factored from ...
28894 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28895 here. Call above function.
28896 * tree-data-ref.h (create_runtime_alias_checks): New function.
28898 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28900 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28901 segment length for dr_b and compute it in wide_int.
28903 2017-05-31 Richard Biener <rguenther@suse.de>
28905 PR tree-optimization/80906
28906 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28907 and pass through iv_map.
28908 (copy_bb_and_scalar_dependences): Adjust.
28909 (translate_pending_phi_nodes): Likewise.
28910 (copy_loop_close_phi_args): Handle code-generating IVs instead
28913 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28915 * diagnostic-color.c (color_dict): Add "type-diff".
28916 (parse_gcc_colors): Update comment.
28917 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28918 -fdiagnostics-show-template-tree and -fno-elide-type.
28919 (GCC_COLORS): Add type-diff to example.
28921 (-fdiagnostics-show-template-tree): New.
28922 (-fno-elide-type): New.
28923 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28924 the pp_format_decoder callback. Call any m_format_postprocessor's
28926 (pretty_printer::pretty_printer): Initialize
28927 m_format_postprocessor.
28928 (pretty_printer::~pretty_printer): Delete any
28929 m_format_postprocessor.
28930 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28931 (class format_postprocessor): New class.
28932 (struct pretty_printer::format_decoder): Document the new parameters.
28933 (struct pretty_printer::m_format_postprocessor): New field.
28934 * tree-diagnostic.c (default_tree_printer): Update for new
28935 bool and const char ** params.
28936 * tree-diagnostic.h (default_tree_printer): Likewise.
28938 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28940 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28941 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28942 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28943 rs6000_gen_cell_microcode code.
28944 (rs6000_final_prescan_insn): Delete.
28945 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28946 "warn-cell-microcode" entries.
28947 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28948 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28949 throughout. Change cc_reg_not_micro_cr0_operand to
28950 cc_reg_not_cr0_operand throughout.
28951 (*extendhi<mode>2_noload): Delete.
28952 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28953 (mwarn-cell-microcode): Delete.
28954 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28955 -mgen-cell-microcode and -mwarn-cell-microcode.
28957 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28960 * config/i386/constraints.md (Yd): New constraint.
28962 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28963 and (?Yd, r) alternatives. Update insn attributes.
28964 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28965 and (?*Yd, r) alternatives. Update insn attributes.
28966 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28968 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28970 * gimplify.c (gimplify_modify_expr): Don't create a
28971 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28974 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28976 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28978 2017-05-30 Richard Biener <rguenther@suse.de>
28980 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28981 and reduc_def fields.
28982 (STMT_VINFO_REDUC_TYPE): New define.
28983 (STMT_VINFO_REDUC_DEF): Likewise.
28984 (vect_force_simple_reduction): Adjust prototype.
28985 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28986 (vect_is_simple_reduction): Remove check_reduction argument.
28987 (vect_force_simple_reduction): Adjust and set
28988 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28989 (vectorizable_reduction): Do not re-do reduction analysis
28990 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28991 * tree-parloops.c (gather_scalar_reductions): Adjust.
28993 2017-05-30 Richard Biener <rguenther@suse.de>
28995 PR middle-end/80901
28996 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28999 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29001 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29002 Introduce unknown_misalignment parameter and remove vf.
29003 (vect_peeling_hash_get_lowest_cost):
29004 Pass unknown_misalignment parameter.
29005 (vect_enhance_data_refs_alignment):
29006 Fix unsupportable data ref treatment.
29008 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29010 * tree-vect-data-refs.c (vect_get_data_access_cost):
29011 Workaround for SLP handling.
29012 (vect_enhance_data_refs_alignment):
29013 Compute costs for doing no peeling at all, compare to the best
29014 peeling costs so far and avoid peeling if cheaper.
29016 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29018 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29019 Return peeling info and set costs to zero for unlimited cost
29021 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29022 with unknown misalignment. Compute and costs for unknown
29023 misalignment, compare them to the costs for known misalignment
29024 and choose the cheapest for peeling.
29026 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29028 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29029 (vect_get_peeling_costs_all_drs): Create function.
29030 (vect_peeling_hash_get_lowest_cost):
29031 Use vect_get_peeling_costs_all_drs.
29032 (vect_peeling_supportable): Create function.
29033 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29035 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29037 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29038 DR_HAS_NEGATIVE_STEP.
29039 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29040 (vect_enhance_data_refs_alignment): Use.
29041 (vect_duplicate_ssa_name_ptr_info): Use.
29042 * tree-vectorizer.h (dr_misalignment): Use.
29043 (known_alignment_for_access_p): Use.
29045 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29048 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29050 (has_section_name): New function.
29052 2017-05-30 Martin Liska <mliska@suse.cz>
29055 * auto-profile.c (get_function_decl_from_block): Fix
29058 2017-05-30 Richard Biener <rguenther@suse.de>
29060 PR middle-end/80876
29061 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29063 2017-05-30 Martin Liska <mliska@suse.cz>
29065 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29066 * dumpfile.h (struct dump_file_info): Remove ctors.
29068 2017-05-30 Martin Liska <mliska@suse.cz>
29070 * predict.def: Fix GNU coding style.
29072 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29074 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29075 Mark 'to' argument with ATTRIBUTE_UNUSED.
29077 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29079 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29080 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29081 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29084 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29086 * doc/install.texi (Options specification): Restore entry of
29087 --enable-sjlj-exceptions.
29089 2017-05-27 Michael Eager <eager@eagercon.com>
29092 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29094 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29096 * config/microblaze/microblaze.h
29097 (FIXED_REGISTERS): Update in macro.
29098 (CALL_USED_REGISTERS): Update in macro.
29100 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29102 * doc/install.texi: Add links to macOS binary distributions.
29104 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29108 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29110 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29112 2017-05-26 Martin Liska <mliska@suse.cz>
29114 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29116 2017-05-26 Martin Liska <mliska@suse.cz>
29118 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29119 always leading ';; '.
29120 (dump_bb_info): Likewise.
29121 (brief_dump_cfg): Likewise.
29122 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29123 * dumpfile.c: Remove usage of TDF_VERBOSE.
29124 * dumpfile.h (enum dump_kind): Likewise.
29125 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29126 * print-tree.c (debug_verbose): Remove.
29127 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29128 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29129 * tree-diagnostic.c (default_tree_printer): Replace
29130 TDF_DIAGNOSTIC with TDF_SLIM.
29132 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29134 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29135 in parameter loop, rather than loop_vinfo.
29136 (create_intersect_range_checks): Ditto.
29137 (vect_create_cond_for_alias_checks): Update call to above functions.
29139 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29141 PR tree-optimization/80815
29142 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29143 for merging runtime alias checks. Handle negative DR_STEPs.
29145 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29147 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29149 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29150 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29151 out code pruning runtime alias checks.
29152 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29153 factored out from above.
29154 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29156 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29158 (prune_runtime_alias_test_list): New decalaration.
29160 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29162 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29163 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29164 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29165 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29166 (operator==, comp_dr_with_seg_len_pair): Ditto.
29167 (vect_prune_runtime_alias_test_list): Ditto.
29169 2017-05-26 Martin Liska <mliska@suse.cz>
29172 * params.def: Bound partial-inlining-entry-probability param.
29174 2017-05-26 Marek Polacek <polacek@redhat.com>
29177 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29180 2017-05-26 Richard Biener <rguenther@suse.de>
29182 PR tree-optimization/80842
29183 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29186 2017-05-26 Richard Biener <rguenther@suse.de>
29188 PR tree-optimization/80844
29189 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29191 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29193 * doc/md.texi (Machine Constraints): Update x86 family
29194 machine constraints section to match 'config/i386/constraints.md'.
29196 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29198 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29200 2017-05-25 Nathan Sidwell <nathan@acm.org>
29202 * doc/invoke.texi (--enable-languages): Update documentation.
29204 2017-05-25 Martin Liska <mliska@suse.cz>
29206 * dumpfile.c: Add TDF_FOLDING.
29207 * dumpfile.h (enum dump_kind): Likewise.
29208 * genmatch.c (dt_simplify::gen_1): Use it.
29210 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29212 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29214 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29216 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29217 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29219 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29221 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29222 * match.pd (X == C): Rewrite it here.
29223 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29224 with_certain_nonzero_bits2): New predicates.
29225 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29227 2017-05-24 Nathan Sidwell <nathan@acm.org>
29229 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29232 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29235 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29237 * config/powerpcspe: New port. Files are copied from the rs6000
29238 port, with "rs6000" in filenames replaced by "powerpcspe".
29240 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29242 PR rtl-optimization/80754
29243 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29245 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29247 * config/sparc/sparc.md (length): Return the correct value for -mflat
29248 sibcalls to match output_sibcall.
29250 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29254 * config/rs6000/rs6000.c (struct machine_function): Add new field
29256 (rs6000_get_separate_components): Init that field, use it.
29257 (rs6000_components_for_bb): Use the field.
29259 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29261 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29263 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29265 PR middle-end/80823
29266 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29268 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29271 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29272 address against address_operand predicate.
29273 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29275 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29277 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29278 parameters passed indirectly.
29280 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29282 * config/i386/i386.md (*movdi_internal): Remove SSE4
29283 alternative 18 (?r, *v). Update insn attributes.
29284 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29285 Update insn attributes.
29286 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29287 Update insn attributes.
29288 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29289 alternative 1 (r, v). Remove isa attribute.
29290 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29291 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29292 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29294 2017-05-23 Tom de Vries <tom@codesourcery.com>
29296 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29299 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29301 * cgraphunit.c (symbol_table::process_new_functions): Update.
29302 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29303 (inline_generate_summary): Rename to ...
29304 (ipa_fn_summary_generate): ... this one.
29305 (inline_read_summary): Rename to ...
29306 (ipa_fn_summary_read): ... this one.
29307 (inline_write_summary): Rename to ...
29308 (ipa_fn_summary_write): ... this one.
29309 (inline_free_summary): Rename to ...
29310 (ipa_free_fn_summary): ... this one.
29311 (pass_data_local_fn_summary, pass_local_fn_summary,
29312 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29313 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29314 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29315 make_pass_ipa_fn_summary): New.
29316 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29317 inline_write_summary, inline_free_summary): Remove.
29318 (ipa_free_fn_summary) : New.
29319 * ipa-inline.c (ipa_inline): Update.
29320 (pass_ipa_inline): Do not generate summaries.
29321 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29323 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29324 and add pass_ipa_fn_summary.
29325 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29327 (make_pass_inline_parameters): Remove.
29329 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29331 * omp-low.c (struct omp_context): Remove "default_kind" member.
29334 * omp-offload.c (execute_oacc_device_lower): Remove the
29335 parallelism dimensions function attributes for unparallelized
29336 OpenACC kernels constructs.
29338 2017-05-23 Martin Liska <mliska@suse.cz>
29340 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29342 (cgraph_edge::make_speculative): Likewise.
29343 (cgraph_edge::resolve_speculation): Likewise.
29344 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29345 (cgraph_node::dump): Likewise.
29346 * cgraph.h: Likewise.
29347 * cgraphunit.c (analyze_functions): Likewise.
29348 (symbol_table::compile): Likewise.
29349 * ipa-cp.c (print_all_lattices): Likewise.
29350 (determine_versionability): Likewise.
29351 (initialize_node_lattices): Likewise.
29352 (ipcp_verify_propagated_values): Likewise.
29353 (estimate_local_effects): Likewise.
29354 (update_profiling_info): Likewise.
29355 (create_specialized_node): Likewise.
29356 (perhaps_add_new_callers): Likewise.
29357 (decide_about_value): Likewise.
29358 (decide_whether_version_node): Likewise.
29359 (identify_dead_nodes): Likewise.
29360 (ipcp_store_bits_results): Likewise.
29361 * ipa-devirt.c (dump_targets): Likewise.
29362 (ipa_devirt): Likewise.
29363 * ipa-icf.c (sem_item::dump): Likewise.
29364 (sem_function::equals): Likewise.
29365 (sem_variable::equals): Likewise.
29366 (sem_item_optimizer::read_section): Likewise.
29367 (sem_item_optimizer::execute): Likewise.
29368 (congruence_class::dump): Likewise.
29369 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29370 (dump_inline_summary): Likewise.
29371 (estimate_node_size_and_time): Likewise.
29372 (inline_analyze_function): Likewise.
29373 * ipa-inline-transform.c (inline_call): Likewise.
29374 * ipa-inline.c (report_inline_failed_reason): Likewise.
29375 (want_early_inline_function_p): Likewise.
29376 (edge_badness): Likewise.
29377 (update_edge_key): Likewise.
29378 (inline_small_functions): Likewise.
29379 * ipa-profile.c (ipa_profile): Likewise.
29380 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29381 (ipa_make_edge_direct_to_target): Likewise.
29382 (remove_described_reference): Likewise.
29383 (ipa_impossible_devirt_target): Likewise.
29384 (propagate_controlled_uses): Likewise.
29385 (ipa_print_node_params): Likewise.
29386 (ipcp_transform_function): Likewise.
29387 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29388 (propagate_pure_const): Likewise.
29389 * ipa-reference.c (generate_summary): Likewise.
29390 (read_write_all_from_decl): Likewise.
29391 (propagate): Likewise.
29392 (ipa_reference_read_optimization_summary): Likewise.
29393 * ipa-utils.c (ipa_merge_profiles): Likewise.
29394 * ipa.c (walk_polymorphic_call_targets): Likewise.
29395 (symbol_table::remove_unreachable_nodes): Likewise.
29396 (ipa_single_use): Likewise.
29397 * passes.c (execute_todo): Likewise.
29398 * predict.c (drop_profile): Likewise.
29399 * symtab.c (symtab_node::get_dump_name): New function.
29400 (symtab_node::dump_name): Likewise.
29401 (symtab_node::dump_asm_name): Likewise.
29402 (symtab_node::dump_references): Likewise.
29403 (symtab_node::dump_referring): Likewise.
29404 (symtab_node::dump_base): Likewise.
29405 (symtab_node::debug_symtab): Likewise.
29406 * tree-sra.c (convert_callers_for_node): Likewise.
29407 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29408 * value-prof.c (init_node_map): Likewise.
29410 2017-05-23 Martin Liska <mliska@suse.cz>
29412 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29413 and symtab_node::debug_symtab to symbol_table::debug.
29414 * cgraphunit.c (analyze_functions): Use the renamed function.
29415 (symbol_table::compile): Likewise.
29416 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29417 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29418 * passes.c (execute_todo): Likewise.
29419 * symtab.c (symbol_table::dump): New function.
29420 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29422 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29424 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29425 that nonconst implies exec.
29427 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29429 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29430 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29431 (inline_edge_summary_vec): Turn into ...
29432 (ipa_call_summaries): ... this one.
29433 (redirect_to_unreachable, edge_set_predicate,
29434 evaluate_properties_for_edge, inline_summary_alloc,
29435 reset_ipa_call_summary, reset_inline_summary,
29436 inline_summary_t::duplicate): Update.
29437 (inline_edge_duplication_hook): Turn to ...
29438 (ipa_call_summary_t::duplicate): ... this one.
29439 (inline_edge_removal_hook): Turn to ...
29440 (ipa_call_summary_t::remove): ... this one.
29441 (dump_inline_edge_summary): Turn to ...
29442 (dump_ipa_call_summary): ... this one.
29443 (estimate_function_body_sizes): Update.
29444 (inline_update_callee_summaries): Update.
29445 (remap_edge_change_prob): Update.
29446 (remap_edge_summaries): Update.
29447 (inline_merge_summary): Update.
29448 (do_estimate_edge_time): Update.
29449 (inline_generate_summary): Update.
29450 (inline_read_section): Update.
29451 (inline_read_summary): Update.
29452 (inline_free_summary): Update.
29453 * ipa-inline.c (can_inline_edge_p): Update.
29454 (compute_inlined_call_time): Update.
29455 (want_inline_small_function_p): Update.
29456 (edge_badness): Update.
29457 (early_inliner): Update.
29458 * ipa-inline.h (inline_edge_summary): Turn to ...
29459 (ipa_call_summary): ... this one.
29460 (ipa_call_summary_t): New class.
29461 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29462 (ipa_call_summaries): New.
29463 (inline_edge_summary): Remove.
29464 (estimate_edge_growth): Update.
29465 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29466 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29467 * ipa-split.c (execute_split_functions): Update.
29468 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29470 2017-05-23 Tom de Vries <tom@codesourcery.com>
29472 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29473 attributes): Document rdrand effective target.
29475 2017-05-23 Tom de Vries <tom@codesourcery.com>
29477 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29478 attributes): Sort alphabetically.
29480 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29482 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29484 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29487 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29488 V2DF/V2DI splat into two separate patterns, one that handles
29489 registers, and the other that only handles memory. Drop support
29490 for splatting from a GPR on ISA 2.07 and then splitting the
29491 splat into direct move and splat.
29492 (vsx_splat_<mode>_reg): Likewise.
29493 (vsx_splat_<mode>_mem): Likewise.
29495 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29497 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29499 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29501 PR middle-end/80809
29502 * omp-low.c (finish_taskreg_remap): New function.
29503 (finish_taskreg_scan): If unit size of ctx->record_type
29504 is non-constant, unshare the size expression and replace
29505 decls in it with possible outer var refs.
29507 PR middle-end/80809
29508 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29509 GOVD_SHARED rather than GOVD_PRIVATE with it.
29510 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29511 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29513 PR middle-end/80853
29514 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29515 as last argument to build_outer_var_ref for pointer bases of array
29516 section reductions.
29518 2017-05-19 Martin Sebor <msebor@redhat.com>
29520 * print-tree.c (print_node): Print DECL_READ_P flag.
29522 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29524 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29525 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29526 * cgraph.c: Likewise.
29527 * cgraphunit.c: Likewise.
29528 * gengtype.c: Likewise.
29529 * ipa-cp.c: Likewise.
29530 * ipa-devirt.c: Likewise.
29531 * ipa-icf.c: Likewise.
29532 * ipa-predicate.c: Likewise.
29533 * ipa-profile.c: Likewise.
29534 * ipa-prop.c: Likewise.
29535 * ipa-split.c: Likewise.
29537 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29538 edge_predicate_pool, dump_inline_hints,
29539 inline_summary::account_size_time, redirect_to_unreachable,
29540 edge_set_predicate, set_hint_predicate,
29541 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29542 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29543 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29544 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29545 ipa_call_summary_t::remove, initialize_growth_caches,
29546 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29547 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29548 mark_modified, unmodified_parm_1, unmodified_parm,
29549 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29550 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29551 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29552 will_be_nonconstant_predicate, record_modified_bb_info,
29553 get_minimal_bb, record_modified, param_change_prob,
29554 phi_result_unknown_predicate, predicate_for_phi_result,
29555 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29556 estimate_function_body_sizes, compute_inline_parameters,
29557 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29558 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29559 inline_update_callee_summaries, remap_edge_change_prob,
29560 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29561 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29562 inline_analyze_function, inline_summary_t::insert,
29563 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29564 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29565 inline_free_summary): Move to ipa-fnsummary.h
29566 (predicate_t): Remove.
29567 * ipa-fnsummary.c: New file.
29568 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29569 (enum inline_hints_vals, inline_hints, agg_position_info,
29570 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29571 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29572 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29573 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29574 inline_read_summary, inline_write_summary, inline_free_summary,
29575 inline_analyze_function, initialize_inline_failed,
29576 inline_merge_summary, inline_update_overall_summary,
29577 compute_inline_parameters): Move to ipa-fnsummary.h
29578 * ipa-fnsummary.h: New file.
29579 * ipa-inline-transform.h: Include ipa-inline.h.
29580 * ipa-inline.c: LIkewise.
29582 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29584 * ipa-inline.c (edge_badness): Use inlined_time instead of
29585 inline_summaries->get.
29587 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29589 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29591 2017-05-22 Nathan Sidwell <nathan@acm.org>
29593 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29594 (fdump-lang): Document 'raw' option.
29595 * dumpfile.h (TDI_tu): Delete.
29596 * dumpfile.c (dump_files): Remove translation-unit.
29597 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29599 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29601 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29602 command option from $(AWK) call.
29603 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29604 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29605 [FORMAT]: Remove handling of variable.
29606 * config/avr/t-multilib: Regenerate.
29608 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29610 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29612 (dump_inline_summary): Do not print self_time.
29613 (estimate_function_body_sizes): Do not set self_time.
29614 (compute_inline_parameters): Likewise.
29615 (inline_read_section, inline_write_summary): Do not stream self_time.
29616 * ipa-inline.h (inline_summary): Drop self_time.
29618 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29620 * ipa-inline-analysis.c (account_size_time): Rename to ...
29621 (inline_summary::account_size_time): ... this one.
29622 (reset_ipa_call_summary): Turn to ...
29623 (ipa_call_summary::reset): ... this one.
29624 (reset_inline_summary): Turn to ...
29625 (inline_summary::reset): ... this one.
29626 (inline_summary_t::remove): Update.
29627 (inline_summary_t::duplicate): Update.
29628 (ipa_call_summary_t::remove): Update.
29629 (dump_inline_summary): Update.
29630 (estimate_function_body_sizes): Update.
29631 (compute_inline_parameters): Update.
29632 (estimate_node_size_and_time): Update.
29633 (inline_merge_summary): Update.
29634 (inline_update_overall_summary): Update.
29635 (inline_read_section): Update.
29636 (inline_write_summary): Update.
29637 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29638 add account_size_time and reset member functions.
29639 (ipa_call_summary): Add reset function.
29640 * ipa-predicate.h (predicate::operator &): Constify.
29642 2017-05-22 Richard Biener <rguenther@suse.de>
29644 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29646 2017-05-19 Jason Merrill <jason@redhat.com>
29648 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29650 2017-05-19 Marek Polacek <polacek@redhat.com>
29653 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29654 TYPE_OVERFLOW_WRAPS checks.
29656 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29658 * tree-core.h (enum omp_clause_default_kind): Add
29659 "OMP_CLAUSE_DEFAULT_PRESENT".
29660 * tree-pretty-print.c (dump_omp_clause): Handle it.
29661 * gimplify.c (enum gimplify_omp_var_data): Add
29662 "GOVD_MAP_FORCE_PRESENT".
29663 (gimplify_adjust_omp_clauses_1): Map it to
29664 "GOMP_MAP_FORCE_PRESENT".
29665 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29667 * gimplify.c (oacc_default_clause): Clarify.
29669 2017-05-19 Nathan Sidwell <nathan@acm.org>
29671 LANG_HOOK_REGISTER_DUMPS
29672 * toplev.c (general_init): Call register dump lang hook.
29673 * doc/invoke.texi: Document -fdump-lang option family.
29674 * dumpfile.c (dump_files): Remove class dump here.
29675 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29676 * dumpfile.h (tree_dump_index): Remove TDI_class.
29677 * langhooks-def.h (lhd_register_dumps): Declare.
29678 (LANG_HOOKS_REGISTER_DUMPS): Define.
29679 (LANG_HOOKS_INITIALIZER): Add it.
29680 * langhooks.c (lhd_register_dumps): Define.
29681 * langhooks.h (struct lang_hooks): Add register_dumps.
29683 2017-05-19 Nathan Sidwell <nathan@acm.org>
29685 * context.h (context::set_passes): New.
29686 * context.c (context::context): Do not create pass manager.
29687 * toplev.c (general_init): Create pass manager here.
29689 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29691 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29692 use this splitter if two add or or instructions would also work for
29693 the constant we want to generate.
29695 2017-05-19 Richard Biener <rguenther@suse.de>
29698 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29699 predicate evaluation.
29701 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29703 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29705 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29708 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29710 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29711 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29712 (inline_edge_summary_vec): Turn into ...
29713 (ipa_call_summaries): ... this one.
29714 (redirect_to_unreachable, edge_set_predicate,
29715 evaluate_properties_for_edge, inline_summary_alloc,
29716 reset_ipa_call_summary, reset_inline_summary,
29717 inline_summary_t::duplicate): Update.
29718 (inline_edge_duplication_hook): Turn to ...
29719 (ipa_call_summary_t::duplicate): ... this one.
29720 (inline_edge_removal_hook): Turn to ...
29721 (ipa_call_summary_t::remove): ... this one.
29722 (dump_inline_edge_summary): Turn to ...
29723 (dump_ipa_call_summary): ... this one.
29724 (estimate_function_body_sizes): Update.
29725 (inline_update_callee_summaries): Update.
29726 (remap_edge_change_prob): Update.
29727 (remap_edge_summaries): Update.
29728 (inline_merge_summary): Update.
29729 (do_estimate_edge_time): Update.
29730 (inline_generate_summary): Update.
29731 (inline_read_section): Update.
29732 (inline_read_summary): Update.
29733 (inline_free_summary): Update.
29734 * ipa-inline.c (can_inline_edge_p): Update.
29735 (compute_inlined_call_time): Update.
29736 (want_inline_small_function_p): Update.
29737 (edge_badness): Update.
29738 (early_inliner): Update.
29739 * ipa-inline.h (inline_edge_summary): Turn to ...
29740 (ipa_call_summary): ... this one.
29741 (ipa_call_summary_t): New class.
29742 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29743 (ipa_call_summaries): New.
29744 (inline_edge_summary): Remove.
29745 (estimate_edge_growth): Update.
29746 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29747 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29748 * ipa-split.c (execute_split_functions): Update.
29749 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29751 2017-05-19 Richard Biener <rguenther@suse.de>
29753 PR middle-end/80764
29754 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29756 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29758 * config/rs6000/rs6000.c (struct machine_function): Add field
29759 fpr_is_wrapped_separately.
29760 (rs6000_get_separate_components): Use 64 components. Handle the
29761 new FPR components.
29762 (rs6000_components_for_bb): Handle the FPR components.
29763 (rs6000_emit_prologue_components): Handle the FPR components.
29764 (rs6000_emit_epilogue_components): Handle the FPR components.
29765 (rs6000_set_handled_components): Handle the FPR components.
29766 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29767 that are already separately shrink-wrapped.
29768 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29769 that are already separately shrink-wrapped.
29771 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29774 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29777 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29778 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29779 cases where the register allocator uses a move and an offsettable
29780 memory operation to/from a FPR register on ISA 2.06/2.07.
29781 (define_peephole2 for Altivec d-form store): Likewise.
29783 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29786 * config/i386/mmx.md (*mov<mode>_internal): Enable
29787 alternatives 11, 12, 13 and 14 also for 32bit targets.
29788 Remove alternatives 15, 16, 17 and 18.
29789 * config/i386/sse.md (vec_concatv2di): Change
29790 alternative (!x, *y) to (x, ?!*Yn).
29792 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29794 * dumpfile.h (enum dump_kind): Remove stray comma.
29796 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29798 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29799 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29800 predicate::num_conditions
29801 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29802 (CHANGED): turn into predicate::changed.
29803 (agg_position_info): Move to ipa-predicate.h
29804 (add_condition, predicate::add_clause, predicate::operator &=,
29805 predicate::or_with, predicate::evaluate, predicate::probability,
29806 dump_condition, dump_clause, predicate::dump,
29807 predicate::remap_after_duplication, predicate::remap_after_inlining,
29808 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29809 (evaluate_conditions_for_known_args): Update.
29810 (set_cond_stmt_execution_predicate): Update.
29811 * ipa-inline.h: Include ipa-predicate.h
29812 (condition, inline_param_summary, conditions, agg_position_info,
29813 predicate): Move to ipa-predicate.h
29814 * ipa-predicate.c: New file.
29815 * ipa-predicate.h: New file.
29817 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29819 * final.c (leaf_function_p): Check we are not in a sequence.
29821 2017-05-18 Martin Liska <mliska@suse.cz>
29823 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29824 * dumpfile.c (dump_register): Use new enum dump_kind.
29825 (get_dump_file_name): Likewise.
29826 (dump_enable_all): Likewise.
29827 (dump_switch_p_1): Likewise.
29828 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29829 * dumpfile.h (enum dump_kind): New enum type.
29830 (struct dump_file_info): Create constructor and
29831 format fields and comments.
29832 * passes.c (pass_manager::register_one_dump_file):
29834 * statistics.c (statistics_early_init): Likewise.
29835 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29836 TDF_TREE with TDF_SLIM.
29837 (gather_memory_references_ref): Likewise.
29839 2017-05-18 Martin Liska <mliska@suse.cz>
29841 * vec.h (struct vnull): Use it.
29843 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29845 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29846 (true_predicate, false_predicate, true_predicate_p,
29847 false_predicate_p): Remove.
29848 (single_cond_predicate, not_inlined_predicate): Turn to member function
29850 (add_condition): Update.
29851 (add_clause): Turn to...
29852 (predicate::add_clause): ... this one; update; allow passing NULL
29854 (and_predicates): Turn to ...
29855 (predicate::operator &=): ... this one.
29856 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29857 (or_predicates): Turn to ...
29858 (predicate::or_with): ... this one.
29859 (evaluate_predicate): Turn to ...
29860 (predicate::evaluate): ... this one.
29861 (predicate_probability): Turn to ...
29862 (predicate::probability): ... this one.
29863 (dump_condition): Update.
29864 (dump_predicate): Turn to ...
29865 (predicate::dump): ... this one.
29866 (account_size_time): Update.
29867 (edge_set_predicate): Update.
29868 (set_hint_predicate): UPdate.
29869 (evaluate_conditions_for_known_args): Update.
29870 (evaluate_properties_for_edge): Update.
29871 (remap_predicate_after_duplication): Turn to...
29872 (predicate::remap_after_duplication): ... this one.
29873 (remap_hint_predicate_after_duplication): Update.
29874 (inline_summary_t::duplicate): UPdate.
29875 (dump_inline_edge_summary): Update.
29876 (dump_inline_summary): Update.
29877 (set_cond_stmt_execution_predicate): Update.
29878 (set_switch_stmt_execution_predicate): Update.
29879 (compute_bb_predicates): Update.
29880 (will_be_nonconstant_expr_predicate): Update.
29881 (will_be_nonconstant_predicate): Update.
29882 (phi_result_unknown_predicate): Update.
29883 (predicate_for_phi_result): Update.
29884 (array_index_predicate): Update.
29885 (estimate_function_body_sizes): Update.
29886 (estimate_node_size_and_time): Update.
29887 (estimate_ipcp_clone_size_and_time): Update.
29888 (remap_predicate): Rename to ...
29889 (predicate::remap_after_inlining): ... this one.
29890 (remap_hint_predicate): Update.
29891 (inline_merge_summary): Update.
29892 (inline_update_overall_summary): Update.
29893 (estimate_size_after_inlining): Update.
29894 (read_predicate): Rename to ...
29895 (predicate::stream_in): ... this one.
29896 (read_inline_edge_summary): Update.
29897 (write_predicate): Rename to ...
29898 (predicate::stream_out): ... this one.
29899 (write_inline_edge_summary): Update.
29900 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29901 (clause_t): Turn to uint32_t
29902 (predicate): Turn to class; implement constructor and operators
29904 (size_time_entry): Update.
29905 (inline_summary): Update.
29906 (inline_edge_summary): Update.
29908 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29910 * fold-const.c (fold_binary_loc): Move transformation...
29911 * match.pd (C - X CMP X): ... here.
29913 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29915 * config/sparc/sparc.c (sparc_option_override): Set function
29916 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29917 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29919 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29921 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29923 2017-05-18 Marek Polacek <polacek@redhat.com>
29926 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29927 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29928 gimple_assign_load_p.
29930 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29932 PR middle-end/80692
29933 * real.c (do_compare): Give decimal_do_compare preference over
29934 comparing just the signs.
29936 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29938 * doc/md.texi (Canonicalization of Instructions): Describe the
29939 canonical form of instructions that inherently set a condition
29942 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29944 PR middle-end/80775
29945 * tree-cfg.c: Move deletion of unreachable case statements to after
29946 the merging of consecutive case labels.
29948 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29950 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29951 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29952 restoring of callee-saved registers.
29954 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29956 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29957 * config/visium/visium.c (single_set_and_flags): Likewise.
29958 * config/visium/visium.md (Substitutions): Likewise.
29960 2017-05-17 Martin Liska <mliska@suse.cz>
29962 * cfg.c: Introduce dump_flags_t type and
29963 use it instead of int type.
29965 * cfghooks.c: Likewise.
29966 * cfghooks.h (struct cfg_hooks): Likewise.
29967 * cfgrtl.c: Likewise.
29968 * cfgrtl.h: Likewise.
29969 * cgraph.c (cgraph_node::get_body): Likewise.
29970 * coretypes.h: Likewise.
29971 * domwalk.c: Likewise.
29972 * domwalk.h: Likewise.
29973 * dumpfile.c (struct dump_option_value_info): Likewise.
29974 (dump_enable_all): Likewise.
29975 (dump_switch_p_1): Likewise.
29976 (opt_info_switch_p): Likewise.
29977 * dumpfile.h (enum tree_dump_index): Likewise.
29978 (struct dump_file_info): Likewise.
29979 * genemit.c: Likewise.
29980 * generic-match-head.c: Likewise.
29981 * gengtype.c (open_base_files): Likewise.
29982 * gimple-pretty-print.c: Likewise.
29983 * gimple-pretty-print.h: Likewise.
29984 * graph.c (print_graph_cfg): Likewise.
29985 * graphite-scop-detection.c (dot_all_sese): Likewise.
29986 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29987 * loop-unroll.c (report_unroll): Likewise.
29988 * passes.c (pass_manager::register_one_dump_file): Likewise.
29989 * print-tree.c: Likewise.
29990 * statistics.c: Likewise.
29991 * tree-cfg.c: Likewise.
29992 * tree-cfg.h: Likewise.
29993 * tree-dfa.c: Likewise.
29994 * tree-dfa.h: Likewise.
29995 * tree-dump.c (dump_function): Likewise.
29996 * tree-dump.h (struct dump_info): Likewise.
29997 * tree-pretty-print.c: Likewise.
29998 * tree-pretty-print.h: Likewise.
29999 * tree-ssa-live.c: Likewise.
30000 * tree-ssa-live.h: Likewise.
30001 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30002 * tree-vect-loop.c: Likewise.
30003 * tree-vect-slp.c: Likewise.
30005 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30006 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30008 PR tree-optimization/80457
30009 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30010 of all arguments to a statement as scalar_to_vec operations.
30011 (vectorizable_call): Adjust call to vect_model_simple_cost for
30013 (vectorizable_conversion): Likewise.
30014 (vectorizable_assignment): Likewise.
30015 (vectorizable_shift): Likewise.
30016 (vectorizable_operation): Likewise.
30017 (vectorizable_comparison): Likewise.
30018 (vect_is_simple_cond): Record the def types for operands.
30019 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30020 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30021 for statement argument count.
30023 2017-05-16 Carl Love <cel@us.ibm.com>
30025 * config/rs6000/rs6000-c: Add support for built-in functions
30026 vector unsigned long long vec_bperm (vector unsigned long long,
30027 vector unsigned char)
30028 vector signed long long vec_mule (vector signed int,
30030 vector unsigned long long vec_mule (vector unsigned int,
30031 vector unsigned int)
30032 vector signed long long vec_mulo (vector signed int,
30034 vector unsigned long long vec_mulo (vector unsigned int,
30035 vector unsigned int)
30036 vector signed char vec_sldw (vector signed char,
30037 vector signed char,
30039 vector unsigned char vec_sldw (vector unsigned char,
30040 vector unsigned char,
30042 vector signed short vec_sldw (vector signed short,
30043 vector signed short,
30045 vector unsigned short vec_sldw (vector unsigned short,
30046 vector unsigned short,
30048 vector signed int vec_sldw (vector signed int,
30051 vector unsigned int vec_sldw (vector unsigned int,
30052 vector unsigned int,
30054 vector signed long long vec_sldw (vector signed long long,
30055 vector signed long long,
30057 vector unsigned long long vec_sldw (vector unsigned long long,
30058 vector unsigned long long,
30060 * config/rs6000/rs6000-c: Add support for built-in functions
30061 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30062 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30063 * doc/extend.texi: Update the built-in documentation for the
30064 new built-in functions.
30066 2017-05-16 Marek Polacek <polacek@redhat.com>
30070 * tree.c (save_expr): Don't fold the expression.
30072 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30074 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30075 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30076 and (?*y,m). Update insn attributes.
30078 2017-05-16 Martin Liska <mliska@suse.cz>
30080 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30081 flags argument of print_gimple_stmt, print_gimple_expr,
30082 print_generic_stmt and print_generic_expr.
30083 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30084 * coretypes.h: Likewise.
30085 * except.c (dump_eh_tree): Likewise.
30086 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30087 * gimple-pretty-print.h: Likewise.
30088 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30089 (backprop::push_to_worklist): Likewise.
30090 (backprop::pop_from_worklist): Likewise.
30091 (backprop::process_use): Likewise.
30092 (backprop::intersect_uses): Likewise.
30093 (note_replacement): Likewise.
30094 * gimple-ssa-store-merging.c
30095 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30096 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30097 (pass_store_merging::execute): Likewise.
30098 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30099 (ssa_base_cand_dump_callback): Likewise.
30100 (dump_incr_vec): Likewise.
30101 (replace_refs): Likewise.
30102 (replace_mult_candidate): Likewise.
30103 (create_add_on_incoming_edge): Likewise.
30104 (create_phi_basis): Likewise.
30105 (insert_initializers): Likewise.
30106 (all_phi_incrs_profitable): Likewise.
30107 (introduce_cast_before_cand): Likewise.
30108 (replace_one_candidate): Likewise.
30109 * gimplify.c (gimplify_expr): Likewise.
30110 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30111 (set_rename): Likewise.
30112 (rename_uses): Likewise.
30113 (copy_loop_phi_nodes): Likewise.
30114 (add_close_phis_to_merge_points): Likewise.
30115 (copy_loop_close_phi_args): Likewise.
30116 (copy_cond_phi_args): Likewise.
30117 (graphite_copy_stmts_from_block): Likewise.
30118 (translate_pending_phi_nodes): Likewise.
30119 * graphite-poly.c (print_pdr): Likewise.
30120 (dump_gbb_cases): Likewise.
30121 (dump_gbb_conditions): Likewise.
30122 (print_scop_params): Likewise.
30123 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30124 (build_cross_bb_scalars_use): Likewise.
30125 (gather_bbs::before_dom_children): Likewise.
30126 * hsa-dump.c (dump_hsa_immed): Likewise.
30127 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30128 (get_replacement_map): Likewise.
30129 * ipa-inline-analysis.c (dump_condition): Likewise.
30130 (estimate_function_body_sizes): Likewise.
30131 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30132 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30133 * ipa-prop.c (ipa_dump_param): Likewise.
30134 (ipa_print_node_jump_functions_for_edge): Likewise.
30135 (ipa_modify_call_arguments): Likewise.
30136 (ipa_modify_expr): Likewise.
30137 (ipa_dump_param_adjustments): Likewise.
30138 (ipa_dump_agg_replacement_values): Likewise.
30139 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30140 * ipa-pure-const.c (check_stmt): Likewise.
30141 (pass_nothrow::execute): Likewise.
30142 * ipa-split.c (execute_split_functions): Likewise.
30143 * omp-offload.c (dump_oacc_loop_part): Likewise.
30144 (dump_oacc_loop): Likewise.
30145 * trans-mem.c (tm_log_emit): Likewise.
30146 (tm_memopt_accumulate_memops): Likewise.
30147 (dump_tm_memopt_set): Likewise.
30148 (dump_tm_memopt_transform): Likewise.
30149 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30150 (print_loop): Likewise.
30151 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30152 (chkp_gather_checks_info): Likewise.
30153 (chkp_get_check_result): Likewise.
30154 (chkp_remove_check_if_pass): Likewise.
30155 (chkp_use_outer_bounds_if_possible): Likewise.
30156 (chkp_reduce_bounds_lifetime): Likewise.
30157 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30158 (chkp_mark_completed_bounds): Likewise.
30159 (chkp_register_incomplete_bounds): Likewise.
30160 (chkp_mark_invalid_bounds): Likewise.
30161 (chkp_maybe_copy_and_register_bounds): Likewise.
30162 (chkp_build_returned_bound): Likewise.
30163 (chkp_get_bound_for_parm): Likewise.
30164 (chkp_build_bndldx): Likewise.
30165 (chkp_get_bounds_by_definition): Likewise.
30166 (chkp_generate_extern_var_bounds): Likewise.
30167 (chkp_get_bounds_for_decl_addr): Likewise.
30168 * tree-chrec.c (chrec_apply): Likewise.
30169 * tree-data-ref.c (dump_data_reference): Likewise.
30170 (dump_subscript): Likewise.
30171 (dump_data_dependence_relation): Likewise.
30172 (analyze_overlapping_iterations): Likewise.
30173 * tree-inline.c (expand_call_inline): Likewise.
30174 (tree_function_versioning): Likewise.
30175 * tree-into-ssa.c (dump_defs_stack): Likewise.
30176 (dump_currdefs): Likewise.
30177 (dump_names_replaced_by): Likewise.
30178 (dump_update_ssa): Likewise.
30179 (update_ssa): Likewise.
30180 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30181 * tree-parloops.c (build_new_reduction): Likewise.
30182 (try_create_reduction_list): Likewise.
30183 (ref_conflicts_with_region): Likewise.
30184 (oacc_entry_exit_ok_1): Likewise.
30185 (oacc_entry_exit_single_gang): Likewise.
30186 * tree-pretty-print.h: Likewise.
30187 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30188 (get_scalar_evolution): Likewise.
30189 (add_to_evolution): Likewise.
30190 (get_loop_exit_condition): Likewise.
30191 (analyze_evolution_in_loop): Likewise.
30192 (analyze_initial_condition): Likewise.
30193 (analyze_scalar_evolution): Likewise.
30194 (instantiate_scev): Likewise.
30195 (number_of_latch_executions): Likewise.
30196 (gather_chrec_stats): Likewise.
30197 (final_value_replacement_loop): Likewise.
30198 (scev_const_prop): Likewise.
30199 * tree-sra.c (dump_access): Likewise.
30200 (disqualify_candidate): Likewise.
30201 (create_access): Likewise.
30202 (reject): Likewise.
30203 (maybe_add_sra_candidate): Likewise.
30204 (create_access_replacement): Likewise.
30205 (analyze_access_subtree): Likewise.
30206 (analyze_all_variable_accesses): Likewise.
30207 (sra_modify_assign): Likewise.
30208 (initialize_constant_pool_replacements): Likewise.
30209 (find_param_candidates): Likewise.
30210 (decide_one_param_reduction): Likewise.
30211 (replace_removed_params_ssa_names): Likewise.
30212 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30213 * tree-ssa-copy.c (dump_copy_of): Likewise.
30214 (copy_prop_visit_cond_stmt): Likewise.
30215 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30216 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30217 (record_equivalences_from_stmt): Likewise.
30218 * tree-ssa-dse.c (compute_trims): Likewise.
30219 (delete_dead_call): Likewise.
30220 (delete_dead_assignment): Likewise.
30221 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30222 (forward_propagate_into_cond): Likewise.
30223 (pass_forwprop::execute): Likewise.
30224 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30225 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30227 (move_computations_worker): Likewise.
30228 (execute_sm): Likewise.
30229 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30230 (remove_exits_and_undefined_stmts): Likewise.
30231 (remove_redundant_iv_tests): Likewise.
30232 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30233 (adjust_iv_update_pos): Likewise.
30234 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30235 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30236 (value_replacement): Likewise.
30237 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30238 * tree-ssa-pre.c (print_pre_expr): Likewise.
30239 (get_representative_for): Likewise.
30240 (create_expression_by_pieces): Likewise.
30241 (insert_into_preds_of_block): Likewise.
30242 (eliminate_insert): Likewise.
30243 (eliminate_dom_walker::before_dom_children): Likewise.
30244 (eliminate): Likewise.
30245 (remove_dead_inserted_code): Likewise.
30246 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30247 * tree-ssa-reassoc.c (get_rank): Likewise.
30248 (eliminate_duplicate_pair): Likewise.
30249 (eliminate_plus_minus_pair): Likewise.
30250 (eliminate_not_pairs): Likewise.
30251 (undistribute_ops_list): Likewise.
30252 (eliminate_redundant_comparison): Likewise.
30253 (update_range_test): Likewise.
30254 (optimize_range_tests_var_bound): Likewise.
30255 (optimize_vec_cond_expr): Likewise.
30256 (rewrite_expr_tree): Likewise.
30257 (rewrite_expr_tree_parallel): Likewise.
30258 (linearize_expr): Likewise.
30259 (break_up_subtract): Likewise.
30260 (linearize_expr_tree): Likewise.
30261 (attempt_builtin_powi): Likewise.
30262 (attempt_builtin_copysign): Likewise.
30263 (transform_stmt_to_copy): Likewise.
30264 (transform_stmt_to_multiply): Likewise.
30265 (dump_ops_vector): Likewise.
30266 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30267 (print_scc): Likewise.
30268 (set_ssa_val_to): Likewise.
30269 (visit_reference_op_store): Likewise.
30270 (visit_use): Likewise.
30271 (sccvn_dom_walker::before_dom_children): Likewise.
30272 (run_scc_vn): Likewise.
30273 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30275 (expr_hash_elt::print): Likewise.
30276 (const_and_copies::pop_to_marker): Likewise.
30277 (const_and_copies::record_const_or_copy_raw): Likewise.
30278 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30279 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30280 (dump_predicates): Likewise.
30281 (find_uninit_use): Likewise.
30282 (warn_uninitialized_phi): Likewise.
30283 (pass_late_warn_uninitialized::execute): Likewise.
30284 * tree-ssa.c (verify_vssa): Likewise.
30285 (verify_ssa): Likewise.
30286 (maybe_optimize_var): Likewise.
30287 * tree-vrp.c (dump_value_range): Likewise.
30288 (dump_all_value_ranges): Likewise.
30289 (dump_asserts_for): Likewise.
30290 (register_edge_assert_for_2): Likewise.
30291 (vrp_visit_cond_stmt): Likewise.
30292 (vrp_visit_switch_stmt): Likewise.
30293 (vrp_visit_stmt): Likewise.
30294 (vrp_visit_phi_node): Likewise.
30295 (simplify_cond_using_ranges_1): Likewise.
30296 (fold_predicate_in): Likewise.
30297 (evrp_dom_walker::before_dom_children): Likewise.
30298 (evrp_dom_walker::push_value_range): Likewise.
30299 (evrp_dom_walker::pop_value_range): Likewise.
30300 (execute_early_vrp): Likewise.
30302 2017-05-16 Richard Biener <rguenther@suse.de>
30304 * dwarf2out.c (loc_list_from_tree_1): Do not create
30305 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30307 2017-05-16 Richard Biener <rguenther@suse.de>
30309 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30311 (note_variable_value_in_expr): If we resolved the decl ref
30312 do not push to the stack.
30314 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30316 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30317 operations in fast-math mode.
30318 (vaddq_f16): Likewise.
30319 (vmul_f16): Likewise.
30320 (vmulq_f16): Likewise.
30321 (vsub_f16): Likewise.
30322 (vsubq_f16): Likewise.
30323 * config/arm/neon.md (add<mode>3): New.
30325 (fma:<VH:mode>3): New. Also remove outdated comment.
30328 2017-05-16 Martin Liska <mliska@suse.cz>
30332 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30333 (odr_types_equivalent_p): Likewise.
30335 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30337 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30339 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30342 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30343 non-interunit SSE move alternatives with '?'.
30344 (zero-extendsidi peephole2): New peephole to skip intermediate
30345 general register in SSE zero-extend sequence.
30347 2017-05-15 Jeff Law <law@redhat.com>
30349 * reorg.c (relax_delay_slots): Create a new variable to hold
30350 the temporary target rather than clobbering TARGET_LABEL.
30352 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30353 missing argument to extract_bit_field call.
30354 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30356 2017-05-15 Martin Liska <mliska@suse.cz>
30359 * gcc.c (process_command): Do not allow empty argument of -o option.
30361 2017-05-15 Renlin Li <renlin.li@arm.com>
30363 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30364 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30365 * config/aarch64/constraints.md (Usf): Add long call check.
30366 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30367 (call_value): Likewise.
30368 (sibcall): Likewise.
30369 (sibcall_value): Likewise.
30371 (call_value_insn): New.
30372 (sibcall_insn): Update rtx pattern.
30373 (sibcall_value_insn): Likewise.
30374 (call_internal): Remove.
30375 (call_value_internal): Likewise.
30376 (sibcall_internal): Likewise.
30377 (sibcall_value_internal): Likewise.
30378 (call_reg): Likewise.
30379 (call_symbol): Likewise.
30380 (call_value_reg): Likewise.
30381 (call_value_symbol): Likewise.
30383 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30386 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30388 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30390 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30391 compatible with CCGOCmode and with CCZmode.
30393 2017-05-14 Martin Sebor <msebor@redhat.com>
30395 PR middle-end/77671
30396 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30397 (gimple_fold_builtin_snprintf): Same.
30398 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30399 (gimple_fold_builtin_snprintf): Same.
30400 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30401 of character types.
30402 (is_call_safe): New function.
30403 (try_substitute_return_value): Call it.
30404 (try_simplify_call): New function.
30405 (pass_sprintf_length::handle_gimple_call): Call it.
30407 2017-05-14 Martin Sebor <msebor@redhat.com>
30409 PR middle-end/80669
30410 * builtins.c (expand_builtin_stpncpy): Simplify.
30412 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30414 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30415 * config/i386/i386.h
30416 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30417 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30418 (struct machine_function): Add new members call_ms2sysv,
30419 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30420 (struct machine_frame_state): New fields sp_realigned and
30421 sp_realigned_offset.
30422 * config/i386/i386.c
30423 (enum xlogue_stub): New enum.
30424 (enum xlogue_stub_sets): New enum.
30425 (class xlogue_layout): New class.
30426 (struct ix86_frame): New fields stack_realign_allocate_offset,
30427 stack_realign_offset and outlined_save_offset. Modify comments to
30428 detail stack layout when using out-of-line stubs.
30429 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30430 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30431 -mcall-ms2sysv-xlogues.
30432 (stub_managed_regs): New static variable.
30433 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30434 registers managed by out-of-line stub.
30435 (disable_call_ms2sysv_xlogues): New function.
30436 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30437 m->call_ms2sysv when appropriate and compute frame layout for
30439 (sp_valid_at, fp_valid_at): New inline functions.
30440 (choose_basereg): New function.
30441 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30443 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30444 Use align parameter of choose_baseaddr to generated aligned SSE movs
30446 (pro_epilogue_adjust_stack): Modify to track
30447 machine_frame_state::sp_realigned.
30448 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30449 (ix86_nsaved_sseregs): Likewise.
30450 (ix86_emit_save_regs): Likewise.
30451 (ix86_emit_save_regs_using_mov): Likewise.
30452 (ix86_emit_save_sse_regs_using_mov): Likewise.
30453 (get_scratch_register_on_entry): Likewise.
30454 (gen_frame_set): New function.
30455 (gen_frame_load): Likewise.
30456 (gen_frame_store): Likewise.
30457 (emit_outlined_ms2sysv_save): Likewise.
30458 (emit_outlined_ms2sysv_restore): Likewise.
30459 (ix86_expand_prologue): Modify stack re-alignment code and call
30460 emit_outlined_ms2sysv_save when appropriate.
30461 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30462 parameter rtx_insn *insn, which allows the function to be used to only
30463 generate the notes.
30464 (ix86_expand_epilogue): Modify validity checks of frame and stack
30465 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30466 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30467 * config/i386/predicates.md
30468 (save_multiple): New predicate.
30469 (restore_multiple): Likewise.
30470 * config/i386/sse.md
30471 (save_multiple<mode>): New pattern.
30472 (save_multiple_realign<mode>): Likewise.
30473 (restore_multiple<mode>): Likewise.
30474 (restore_multiple_and_return<mode>): Likewise.
30475 (restore_multiple_leave_return<mode>): Likewise.
30476 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30478 2017-05-14 Julia Koval <julia.koval@intel.com>
30480 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30481 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30482 (__builtin_ia32_xsetbv): New builtins.
30483 * config/i386/i386.c (ix86_expand_special_args_builtin):
30485 (ix86_expand_builtin): Special expand for new intrinsics.
30486 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30487 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30488 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30492 * cfganal.c (inverted_post_order_compute): Change argument type
30494 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30495 * df-core.c (rest_of_handle_df_initialize): Adjust.
30496 (rest_of_handle_df_finish): Likewise.
30497 (df_analyze_1): Likewise.
30498 (df_analyze): Likewise.
30499 (loop_inverted_post_order_compute): Change argument to be a vec *.
30500 (df_analyze_loop): Adjust.
30501 (df_get_n_blocks): Likewise.
30502 (df_get_postorder): Likewise.
30503 * df.h (struct df_d): Change field to be a vec.
30504 * lcm.c (compute_laterin): Adjust.
30505 (compute_available): Likewise.
30506 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30507 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30508 * tree-ssa-pre.c (compute_antic): Likewise.
30510 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30512 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30513 (depth_first_search::depth_first_search): Change structure init
30514 function to this constructor.
30515 (depth_first_search::add_bb): Rename function to this member.
30516 (depth_first_search::execute): Likewise.
30517 (flow_dfs_compute_reverse_finish): Adjust.
30519 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30521 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30522 (longest_simple_path): Likewise.
30523 * shrink-wrap.c (spread_components): Likewise.
30524 (disqualify_problematic_components): Likewise.
30525 (emit_common_heads_for_components): Likewise.
30526 (emit_common_tails_for_components): Likewise.
30527 (insert_prologue_epilogue_for_components): Likewise.
30529 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30531 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30536 * df-core.c (df_set_blocks): Start using auto_bitmap.
30537 (df_compact_blocks): Likewise.
30538 * df-problems.c (df_rd_confluence_n): Likewise.
30539 * df-scan.c (df_insn_rescan_all): Likewise.
30540 (df_process_deferred_rescans): Likewise.
30541 (df_update_entry_block_defs): Likewise.
30542 (df_update_exit_block_uses): Likewise.
30543 (df_entry_block_bitmap_verify): Likewise.
30544 (df_exit_block_bitmap_verify): Likewise.
30545 (df_scan_verify): Likewise.
30546 * lra-constraints.c (lra_constraints): Likewise.
30547 (undo_optional_reloads): Likewise.
30548 (lra_undo_inheritance): Likewise.
30549 * lra-remat.c (calculate_gen_cands): Likewise.
30550 (do_remat): Likewise.
30551 * lra-spills.c (assign_spill_hard_regs): Likewise.
30552 (spill_pseudos): Likewise.
30553 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30554 (bitmap_set_subtract_values): Likewise.
30556 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30558 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30559 management with auto_bitmap.
30560 (fix_inter_tick): Likewise.
30561 (fix_recovery_deps): Likewise.
30562 * ira.c (add_store_equivs): Likewise.
30563 (find_moveable_pseudos): Likewise.
30564 (split_live_ranges_for_shrink_wrap): Likewise.
30565 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30566 (rtx_reuse_manager::seen_def_p): Likewise.
30567 (rtx_reuse_manager::set_seen_def): Likewise.
30568 * print-rtl.h (class rtx_reuse_manager): Likewise.
30570 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30572 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30574 (migrate_btr_def): Likewise.
30575 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30576 * df-core.c (loop_post_order_compute): Likewise.
30577 (loop_inverted_post_order_compute): Likewise.
30578 * hsa-common.h: Likewise.
30579 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30580 * init-regs.c (initialize_uninitialized_regs): Likewise.
30581 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30582 (inline_small_functions): Likewise.
30583 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30584 * ira.c (combine_and_move_insns): Likewise.
30585 (build_insn_chain): Likewise.
30586 * loop-invariant.c (find_invariants): Likewise.
30587 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30588 * predict.c (tree_predict_by_opcode): Likewise.
30589 (predict_paths_leading_to): Likewise.
30590 (predict_paths_leading_to_edge): Likewise.
30591 (estimate_loops_at_level): Likewise.
30592 (estimate_loops): Likewise.
30593 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30594 (spread_components): Likewise.
30595 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30596 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30597 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30598 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30599 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30600 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30601 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30602 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30603 (mark_threaded_blocks): Likewise.
30604 (thread_through_all_blocks): Likewise.
30605 * tree-ssa.c (verify_ssa): Likewise.
30606 (execute_update_addresses_taken): Likewise.
30607 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30609 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30611 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30613 (post_order_compute): Likewise.
30614 (inverted_post_order_compute): Likewise.
30615 (pre_and_rev_post_order_compute_fn): Likewise.
30617 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30619 * genrecog.c (int_set::int_set): Explicitly construct our
30620 auto_vec base class.
30621 * vec.h (auto_vec::auto_vec): New constructor.
30623 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30625 * bitmap.h (class auto_bitmap): New constructor taking
30626 bitmap_obstack * argument.
30628 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30630 * bitmap.h (class auto_bitmap): Change type of m_bits to
30631 bitmap_head, and adjust ctor / dtor and member operators.
30633 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30635 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30636 when returned register mode doesn't match original mode.
30638 2017-05-12 Jeff Law <law@redhat.com>
30639 Jakub Jelinek <jakub@redhat.com>
30641 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30642 we look for cc setter after the compare-elim changes.
30643 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30644 within the vector to match what compare-elim now expects.
30645 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30646 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30648 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30649 after the compare-elim changes.
30650 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30651 the vector to match what compare-elim now expects.
30652 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30653 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30654 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30655 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30656 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30658 * config/visium/visium.c (single_set_and_flags): Fix where
30659 we look for cc setter after the compare-elim changes.
30660 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30661 with the vector to match what compare-elim now expects.
30662 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30663 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30664 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30665 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30666 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30667 (neg<mode>2_insn_set_overflow): Likewise.
30669 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30671 PR middle-end/79794
30672 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30673 maybe_expand_insn call, set ops[0].target. If still set after call,
30674 set alt_rtl. Add extra arg to recursive calls.
30675 (extract_bit_field): Add alt_rtl argument. Pass to
30677 * expmed.h (extract_bit_field): Fix prototype.
30678 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30679 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30680 to extract_bit_field_calls.
30681 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30682 Pass alt_rtl to extract_bit_field calls.
30683 * calls.c (store_unaligned_arguments_into_psuedos)
30684 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30685 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30687 * optabs.h (struct expand_operand): Add target bitfield.
30689 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30691 * compare-elim.c (try_eliminate_compare): Canonicalize
30692 operation with embedded compare to
30693 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30694 (set (reg) (operation)].
30696 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30698 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30701 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30702 cost of adding a carry flag for ADC instruction.
30703 [case MINUS]: Ignore the cost of subtracting a carry flag
30704 for SBB instruction.
30706 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30708 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30710 * config/rs6000/bmiintrin.h: New file.
30711 * config/rs6000/bmi2intrin.h: New file.
30712 * config/rs6000/x86intrin.h: New file.
30714 2017-05-12 Jeff Law <law@redhat.com>
30716 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30719 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30721 PR middle-end/80707
30722 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30724 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30727 early expansion of vector divide builtins.
30728 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30729 builtins identified as having unsigned arguments.
30731 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30733 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30734 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30735 expansion of vector logical operations (and, andc, or, xor,
30738 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30740 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30741 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30743 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30745 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30746 early GIMPLE expansion of vector multiplies.
30748 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30750 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30751 TARGET_HAVE_MOVT conditional.
30752 (movt splitter): Likewise.
30754 2017-05-12 Richard Biener <rguenther@suse.de>
30756 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30757 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30758 Fold all stmts not inplace.
30760 2017-05-12 Richard Biener <rguenther@suse.de>
30762 PR tree-optimization/80713
30763 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30764 inserted_exprs bit for not removed stmts.
30766 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30768 PR middle-end/69921
30769 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30770 parallelized" attribute for parallelized OpenACC kernels.
30771 * omp-offload.c (execute_oacc_device_lower): Use it.
30773 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30774 Set "oacc kernels" attribute.
30775 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30776 parameter. Adjust all users.
30777 (oacc_fn_attrib_kernels_p): Remove function.
30778 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30779 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30780 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30781 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30782 assert "oacc kernels" attribute is set.
30784 2017-05-11 Carl Love <cel@us.ibm.com>
30786 * config/rs6000/rs6000-c: Add support for built-in functions
30787 vector unsigned char vec_popcnt (vector signed char)
30788 vector unsigned char vec_popcnt (vector unsigned char)
30789 vector unsigned short vec_popcnt (vector signed short)
30790 vector unsigned short vec_popcnt (vector unsigned short)
30791 vector unsigned int vec_popcnt (vector signed int)
30792 vector unsigned int vec_popcnt (vector unsigned int)
30793 vector unsigned long long vec_popcnt (vector signed long long)
30794 vector unsigned long long vec_popcnt (vector unsigned long long)
30795 vector signed long long vec_slo (vector signed long long,
30796 vector signed char)
30797 vector signed long long vec_slo (vector signed long long,
30798 vector unsigned char)
30799 vector unsigned long long vec_slo (vector unsigned long long,
30800 vector signed char)
30801 vector unsigned long long vec_slo (vector unsigned long long,
30802 vector unsigned char)
30803 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30804 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30805 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30806 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30807 * doc/extend.texi: Update the built-in documentation file for the
30808 new built-in functions.
30810 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30812 * attribs.h (sorted_attr_string): Move machine independent
30813 functions for target clone support from the i386 port to common
30814 code. Rename ix86_function_versions to common_function_versions.
30815 Rename make_name to make_unique_name.
30816 (common_function_versions): Likewise.
30817 (make_unique_name): Likewise.
30818 (make_dispatcher_decl): Likewise.
30819 (is_function_default_version): Likewise.
30820 * attribs.c (attr_strcmp): Likewise.
30821 (sorted_attr_string): Likewise.
30822 (common_function_versions): Likewise.
30823 (make_unique_name): Likewise.
30824 (make_dispatcher_decl): Likewise.
30825 (is_function_default_version): Likewise.
30826 * config/i386/i386.c (attr_strcmp): Likewise.
30827 (sorted_attr_string): Likewise.
30828 (ix86_function_versions): Likewise.
30829 (make_name): Likewise.
30830 (make_dispatcher_decl): Likewise.
30831 (is_function_default_version): Likewise.
30832 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30834 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30837 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30838 Account for direct move costs for vec_construct of integer
30841 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30844 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30845 (UNSPEC_STX_ATOMIC): Ditto.
30846 (loaddi_via_sse): New insn.
30847 (storedi_via_sse): Ditto.
30848 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30849 Update corresponding peephole2 patterns.
30850 (atomic_storedi_fpu): Ditto.
30852 2017-05-11 Julia Koval <julia.koval@intel.com>
30854 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30855 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30857 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30858 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30859 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30861 2017-05-11 Nathan Sidwell <nathan@acm.org>
30863 * graphite-poly.c: Include dumpfile.h.
30865 * dumpfle.h (dump_function): Declare here ...
30866 * tree-dump.h (dump_function): ... not here.
30867 * dumpfile.c: #include tree-cfg.h.
30868 (dump_function): Move here from ...
30869 * tree-dump.c (dump_function): ... here.
30870 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30871 * graphite-poly.c: Don't include tree-dump.h.
30872 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30873 * print-tree.c: Likewise.
30874 * stor-layout.c: Likewise.
30875 * tree-nested.c: Likewise.
30877 * dumpfile.c (dump_start): Use TDF_FLAGS.
30878 (dump_enable_all): Fix TDF_KIND check thinko.
30880 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30882 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30883 array entries to represent two legal parameterizations of the
30884 overloaded __builtin_cmpb function, as represented by the
30885 P6_OV_BUILTIN_CMPB constant.
30886 (altivec_resolve_overloaded_builtin): Add special case handling
30887 for the __builtin_cmpb function, as represented by the
30888 P6_OV_BUILTIN_CMPB constant.
30889 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30890 (BU_P6_64BIT_2): New macro.
30891 (BU_P6_OVERLOAD_2): New macro
30892 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30893 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30894 (CMPB): Add overload support to represent both 32-bit and 64-bit
30895 compare-bytes function.
30896 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30897 support for TARGET_CMPB.
30898 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30899 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30900 documentation of the __builtin_cmpb overloaded built-in function.
30902 2017-05-11 Richard Biener <rguenther@suse.de>
30904 PR tree-optimization/80705
30905 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30906 bases are not vectorizable.
30908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30910 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30911 when counting register pressure.
30913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30915 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30916 (iv_ca_more_deps): Renamed to ...
30917 (iv_ca_compare_deps): ... this.
30918 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30920 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30922 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30924 (determine_group_iv_costs): ... here.
30925 (find_inv_vars_cb): Record inv var if it's not recorded before.
30927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30929 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30930 (get_shiftadd_cost): Ditto.
30932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30934 * tree-ssa-address.c: Include header file.
30935 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30937 (add_to_parts): Refactor.
30938 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30939 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30944 PR tree-optimization/53090
30945 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30947 (extract_cond_operands): Detect condition with IV on both sides
30948 and return COMP_IV_EXPR_2.
30949 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30950 (rewrite_use_compare): Simplify by removing call to function
30951 extract_cond_operands.
30953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30955 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30956 (extract_cond_operands): Detect condition comparing against non-
30957 invariant bound and return appropriate enum value.
30958 (find_interesting_uses_cond): Update use of extract_cond_operands.
30959 Handle its return value accordingly.
30960 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30962 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30964 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30965 nonlinear iv_use computation in loop invariant sensitive way.
30967 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30969 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30970 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30974 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30975 (dump_cand): Support iv_cand.inv_exprs.
30976 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30978 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30981 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30983 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30985 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30986 as local function. Include necessary header files.
30987 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30989 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30991 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30995 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30996 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30997 RSHIFT_EXPR and BIT_NOT_EXPR.
30999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31001 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31002 (adjust_setup_cost): New parameter supporting round up adjustment.
31003 (struct address_cost_data): Delete.
31004 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31005 (split_address_cost, ptr_difference_cost): Delete.
31006 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31007 (struct ainc_cost_data): New struct.
31008 (get_address_cost_ainc): New function.
31009 (get_address_cost, get_computation_cost): Reimplement.
31010 (determine_group_iv_cost_address): Record inv_expr for all uses of
31012 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31013 (iv_ca_has_deps): Reimplemented to ...
31014 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31016 (iv_ca_extend): Call iv_ca_more_deps.
31018 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31020 * tree-ssa-address.c (struct mem_address): Move to header file.
31021 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31022 * tree-ssa-address.h (struct mem_address): Move from C file.
31023 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31027 * tree-affine.h (aff_combination_type): New interface.
31028 (aff_combination_zero_p): Remove static.
31029 (aff_combination_const_p): New interface.
31030 (aff_combination_singleton_var_p): New interfaces.
31032 2017-05-11 Richard Biener <rguenther@suse.de>
31034 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31035 Skip unreachable blocks and destinations.
31036 (eliminate): Move stmt removal and fixup ...
31037 (fini_eliminate): ... here. Skip inserted exprs.
31038 (pass_pre::execute): Move fini_pre after fini_eliminate.
31039 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31040 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31041 PRE to get rid of dead code that has invalid SSA form and
31042 split critical edges again.
31044 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31046 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31048 2017-05-11 Richard Biener <rguenther@suse.de>
31050 * passes.c (execute_function_todo): Verify loops if they are
31051 said to be up-to-date.
31052 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31053 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31055 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31058 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31059 handle calling assemble_external ourself.
31062 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31063 modes with zero size. Enhance comment.
31065 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31067 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31068 built-ins for vec_xl and vec_xst with short and char pointer
31071 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31073 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31074 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31075 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31076 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31077 (_mm_maskz_min_round_ss): New intrinsics.
31078 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31079 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31080 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31081 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31082 (__builtin_ia32_minss_mask_round): New builtins.
31083 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31084 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31085 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31087 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31088 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31090 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31093 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31095 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31096 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31097 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31098 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31099 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31100 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31101 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31102 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31103 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31104 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31105 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31106 (__builtin_ia32_mulss_mask_round): New builtins.
31107 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31108 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31109 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31111 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31112 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31114 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31117 2017-05-10 Julia Koval <julia.koval@intel.com>
31119 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31120 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31121 (_mm256_setr_m128i): New intrinsics.
31123 2017-05-10 Julia Koval <julia.koval@intel.com>
31125 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31126 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31127 (_mm_maskz_rcp14_ss): New intrinsics.
31128 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31129 (__builtin_ia32_rcp14ss_mask): New builtins.
31130 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31132 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31134 PR tree-optimization/51513
31135 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31136 (assert_unreachable_fallthru_edge_p): Use it.
31137 (group_case_labels_stmt): Likewise.
31138 * tree-cfg.h: Prototype it.
31139 * stmt.c: Include cfghooks.h and tree-cfg.h.
31140 (emit_case_dispatch_table) <gap_label>: New local variable.
31141 Use it to fill dispatch table gaps.
31142 Test for default_label before updating probabilities.
31143 (expand_case) <default_label>: Remove unneeded initialization.
31144 Test for unreachable default case statement and remove its edge.
31145 Set default_label accordingly.
31146 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31148 2017-05-10 Carl Love <cel@us.ibm.com>
31150 * config/rs6000/rs6000-c: Add support for built-in functions
31151 vector signed char vec_neg (vector signed char)
31152 vector signed short int vec_neg (vector short int)
31153 vector signed int vec_neg (vector signed int)
31154 vector signed long long vec_neg (vector signed long long)
31155 vector float vec_neg (vector float)
31156 vector double vec_neg (vector double)
31157 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31159 * config/rs6000/altivec.h: Add define for vec_neg
31160 * doc/extend.texi: Update the built-in documentation for the
31161 new built-in functions.
31163 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31165 PR tree-optimization/77644
31166 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31168 2017-05-10 Nathan Sidwell <nathan@acm.org>
31170 * dumpfile.h (TDI_lang_all): New.
31171 (TDF_KIND): New. Renumber others
31172 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31174 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31176 (get_dump_file_name): Adjust suffix generation.
31177 (dump_enable_all): Use TDF_KIND.
31178 * doc/invoke.texi (-fdump-lang-all): Document.
31180 * dumpfile.h: Tabify.
31182 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31185 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31186 Move member access before delete.
31188 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31190 * tree-inline.c (expand_call_inline): Split block at stmt
31193 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31196 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31197 are now unused after splitting mov{sf,sd}_hardfloat.
31198 (f32_lr2): Likewise.
31199 (f32_lm): Likewise.
31200 (f32_lm2): Likewise.
31201 (f32_li): Likewise.
31202 (f32_li2): Likewise.
31203 (f32_lv): Likewise.
31204 (f32_sr): Likewise.
31205 (f32_sr2): Likewise.
31206 (f32_sm): Likewise.
31207 (f32_sm2): Likewise.
31208 (f32_si): Likewise.
31209 (f32_si2): Likewise.
31210 (f32_sv): Likewise.
31211 (f32_dm): Likewise.
31212 (f32_vsx): Likewise.
31213 (f32_av): Likewise.
31214 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31215 For movsf, order stores so the VSX stores occur before the GPR
31216 store which encourages the register allocator to use a traditional
31217 FPR instead of a GPR. For movsd, order the stores so that the GPR
31218 store comes before the VSX stores to allow the power6 to work.
31219 This is due to the power6 not having a 32-bit integer store
31220 instruction from a FPR.
31221 (movsf_hardfloat): Likewise.
31222 (movsd_hardfloat): Likewise.
31224 2017-05-09 Martin Sebor <msebor@redhat.com>
31226 PR translation/80280
31227 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31230 PR translation/80280
31231 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31232 data member added in r247778.
31233 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31235 2017-05-09 Nathan Sidwell <nathan@acm.org>
31237 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31239 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31242 2017-05-09 Marek Polacek <polacek@redhat.com>
31244 * doc/invoke.texi: Fix typo.
31246 2017-05-09 Richard Biener <rguenther@suse.de>
31248 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31249 (vrp_val_is_min): Likewise.
31250 (set_value_range_to_value): Likewise.
31251 (set_value_range_to_nonnegative): Likewise.
31252 (gimple_assign_nonzero_p): Likewise.
31253 (gimple_stmt_nonzero_p): Likewise.
31254 (vrp_int_const_binop): Likewise. Remove unreachable case.
31255 (adjust_range_with_scev): Adjust comments.
31256 (compare_range_with_value): Likewise.
31257 (extract_range_from_phi_node): Likewise.
31258 (test_for_singularity): Likewise.
31260 2017-05-09 Richard Biener <rguenther@suse.de>
31262 * tree-vrp.c (get_single_symbol): Add assert that we don't
31263 get overflowed constants as invariant part.
31264 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31265 checks. Use wi::cmp instead of recursing for integer constants.
31266 (compare_values): Just ignore whether we assumed undefined
31267 overflow instead of failing the compare.
31268 (extract_range_for_var_from_comparison_expr): Add comment before the
31269 TREE_NO_WARNING sets.
31270 (test_for_singularity): Likewise.
31271 (extract_range_from_comparison): Do not disable optimization
31272 when we assumed undefined overflow.
31273 (extract_range_basic): Remove init of unused var.
31275 2017-05-09 Richard Biener <rguenther@suse.de>
31277 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31278 (extract_range_from_multiplicative_op_1): Adjust.
31279 (extract_range_from_binary_expr_1): Use int_const_binop.
31281 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31284 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31285 rs6000_store_data_bypass_p in seven define_bypass directives and
31286 in several comments.
31287 * config/rs6000/rs6000-protos.h: Add prototype for
31288 rs6000_store_data_bypass_p function.
31289 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31290 function implements slightly different (rs6000-specific) semantics
31291 than store_data_bypass_p, returning false rather than aborting
31292 with assertion error when arguments do not satisfy the
31293 requirements of store data bypass.
31294 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31295 rs6000_store_data_bypass_p.
31297 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31299 * config/xtensa/xtensa-protos.h
31300 (xtensa_initial_elimination_offset): New declaration.
31301 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31302 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31303 macro definition, add case for FRAME_POINTER_REGNUM when
31304 FRAME_GROWS_DOWNWARD.
31305 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31306 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31307 xtensa_initial_elimination_offset.
31309 2017-05-08 Nathan Sidwell <nathan@acm.org>
31311 * doc/invoke.texi: Alphabetize -fdump options.
31313 2017-05-08 Martin Sebor <msebor@redhat.com>
31315 PR translation/80280
31316 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31318 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31320 * target.def (compute_frame_layout): New optional target hook.
31321 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31322 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31323 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31325 * reload1.c (verify_initial_elim_offsets): Likewise.
31326 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31327 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31328 (arm_get_frame_offsets): Split up into this ...
31329 (arm_compute_frame_layout): ... and this function.
31331 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31333 * config/aarch64/constraints.md (Usa): New constraint.
31334 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31336 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31338 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31339 with_multilib_list after it has been checked.
31341 2017-05-08 Richard Biener <rguenther@suse.de>
31343 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31344 (bitmap_set_subtract_values): Likewise.
31346 2017-05-08 Richard Biener <rguenther@suse.de>
31348 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31349 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31351 (gimple_stmt_nonzero_warnv_p): Rename to ...
31352 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31354 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31355 (extract_range_basic): Adjust, do not disable propagation on
31356 strict overflow sensitive simplification.
31357 (vrp_visit_cond_stmt): Likewise.
31359 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31361 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31362 body size unconditionally.
31364 2017-05-07 Jeff Law <law@redhat.com>
31367 2017-05-06 Jeff Law <law@redhat.com>
31368 PR tree-optimization/78496
31369 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31372 PR tree-optimization/78496
31373 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31374 (simplify_stmt_using_ranges): Call it.
31375 (vrp_dom_walker::before_dom_children): Extract equivalences
31376 from an ASSERT_EXPR with an equality comparison against a
31379 2017-05-06 Jeff Law <law@redhat.com>
31381 PR tree-optimization/78496
31382 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31385 PR tree-optimization/78496
31386 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31387 (simplify_stmt_using_ranges): Call it.
31388 (vrp_dom_walker::before_dom_children): Extract equivalences
31389 from an ASSERT_EXPR with an equality comparison against a
31392 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31394 * lra-constraints.c (lra_copy_reg_equiv): New function.
31395 (split_reg): Use it to copy equivalence information from the
31396 original register to the spill register.
31398 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31400 PR rtl-optimization/75964
31401 * simplify-rtx.c (simplify_const_relational_operation): Remove
31402 invalid handling of comparisons of integer ABS.
31404 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31406 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31407 initialize to zero.
31408 (init_regs): Remove declaration.
31409 (function_arg_advance_32): Initialize error_p as boolean variable.
31411 2017-05-05 Nathan Sidwell <nathan@acm.org>
31413 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31414 lines. Use for (;;).
31416 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31418 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31419 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31420 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31421 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31422 VF=2 that require versioning.
31424 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31426 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31429 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31431 * diagnostic.h (diagnostic_override_option_index): Convert from
31432 macro to inline function.
31434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31436 * diagnostic.c (last_module_changed_p): New function.
31437 (set_last_module): New function.
31438 (diagnostic_report_current_module): Convert macro usage to
31439 the above functions.
31440 * diagnostic.h (diagnostic_context::last_module): Strengthen
31441 from const line_map * to const line_map_ordinary *.
31442 (diagnostic_last_module_changed): Delete macro.
31443 (diagnostic_set_last_module): Delete macro.
31445 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31447 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31448 with diagnostic_report_diagnostic.
31449 (diagnostic_n_impl_richloc): Likewise.
31450 * diagnostic.h (report_diagnostic): Delete macro.
31451 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31452 with diagnostic_report_diagnostic.
31453 * substring-locations.c (format_warning_va): Likewise.
31455 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31457 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31458 save/restor of format_spec. Move option-printing code to...
31459 (print_option_information): ...this new function, and
31460 reimplement by simply printing to the pretty_printer,
31461 rather than appending to the format string.
31463 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31465 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31466 handling logic into...
31467 (update_effective_level_from_pragmas): ...this new function.
31469 2017-05-04 Andrew Waterman <andrew@sifive.com>
31471 * config/riscv/riscv.opt (mstrict-align): New option.
31472 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31473 (SLOW_UNALIGNED_ACCESS): Define.
31474 (riscv_slow_unaligned_access): Declare.
31475 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31477 (riscv_slow_unaligned_access): New variable.
31478 (rocket_tune_info): Set slow_unaligned_access to true.
31479 (optimize_size_tune_info): Set slow_unaligned_access to false.
31480 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31481 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31482 (riscv_option_override): Set riscv_slow_unaligned_access.
31483 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31485 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31487 * config/riscv/riscv.md: Unify indentation.
31489 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31494 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31496 (extendsi<mode>2): Add support for doing sign extension via
31497 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31498 don't have ISA 3.0 instructions.
31499 (extendsi<mode>2 splitter): Likewise.
31500 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31501 generate the normal insns since SImode can now go in vector
31502 registers. Disallow the special UNSPECs needed for previous
31503 machines to hide SImode being used. Add new insns
31504 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31505 (fix_trunc<mode>si2_stfiwx): Likewise.
31506 (fix_trunc<mode>si2_internal): Likewise.
31507 (fixuns_trunc<mode>si2): Likewise.
31508 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31509 (fctiw<u>z_<mode>_smallint): Likewise.
31510 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31511 of floating point to 32-bit integer from doing a direct move to
31512 the GPR registers to do a store.
31513 (fctiwz_<mode>): Break long line.
31515 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31517 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31518 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31519 (addr_list, addr_offset_valid_p): New.
31520 (split_address_groups): Check offset validity with above function.
31521 (gt-tree-ssa-loop-ivopts.h): Include header file.
31523 2017-05-05 Nathan Sidwell <nathan@acm.org>
31525 * config.gcc (arm*-*-*): Add missing 'fi'.
31527 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31529 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31530 included in -fopt-info does not matter.
31531 * doc/optinfo.texi (-fopt-info): Fix description of default
31532 behavour. Explicitly say order of options included in -fopt-info
31535 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31537 * config.gcc: Allow combinations of aprofile and rmprofile values for
31538 --with-multilib-list.
31539 * config/arm/t-multilib: New file.
31540 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31541 variables. Remove setting of ISA and floating-point ABI in
31542 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31543 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31544 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31545 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31547 * config/arm/t-rmprofile: Likewise except for the matches changes.
31548 * doc/install.texi (--with-multilib-list): Document the combination of
31549 aprofile and rmprofile values and warn about pitfalls in doing that.
31551 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31553 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31554 (movdi_aarch64): Likewise.
31556 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31558 PR tree-optimization/80632
31559 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31561 (build_arrays): Initialize it for virtual phis.
31562 (fix_phi_nodes): Use it for virtual phis.
31564 PR tree-optimization/80558
31565 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31566 [x, y] op z into [x op, y op z] for op & or | if conditions
31569 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31570 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31573 * config/arm/arm.md (use_literal_pool): Remove.
31574 (64-bit immediate split): No longer takes cost into consideration
31575 if arm_disable_literal_pool is enabled.
31576 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31577 used when arm_disable_literal_pool is enabled.
31578 (arm_max_const_double_inline_cost): Remove use of
31579 arm_disable_literal_pool.
31580 (push_minipool_fix): Add assert.
31581 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31582 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31583 (no_literal_pool_sf_immediate): New.
31585 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31587 PR tree-optimization/80613
31588 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31589 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31591 2017-05-05 Richard Biener <rguenther@suse.de>
31593 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31595 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31597 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31598 of this flag from insn conditions due to removal from r247495.
31600 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31602 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31604 (arm_early_store_addr_dep_ptr): Likewise.
31605 * config/arm/aarch-common-protos.h
31606 (arm_early_load_addr_dep_ptr): Add prototype.
31607 (arm_early_store_addr_dep_ptr): Likewise.
31608 * config/arm/cortex-a53.md: Add new bypasses.
31610 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31612 * tree.c (next_type_uid): Change type to unsigned.
31613 (type_hash_canon): Decrement back next_type_uid if
31614 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31615 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31618 2017-05-04 Martin Sebor <msebor@redhat.com>
31620 * builtins.c: Fix a trivial typo in a comment.
31622 PR middle-end/79234
31623 * builtins.c (check_sizes): Adjust to handle reading past the end.
31624 Avoid printing excessive upper bound of ranges. Use %E to print
31625 tree nodes instead of converting them to %wu.
31626 (expand_builtin_memchr): New function.
31627 (compute_dest_size): Rename...
31628 (compute_objsize): ...to this.
31629 (expand_builtin_memcpy): Adjust.
31630 (expand_builtin_mempcpy): Adjust.
31631 (expand_builtin_strcat): Adjust.
31632 (expand_builtin_strcpy): Adjust.
31633 (check_strncat_sizes): Adjust.
31634 (expand_builtin_strncat): Adjust.
31635 (expand_builtin_strncpy): Adjust and simplify.
31636 (expand_builtin_memset): Adjust.
31637 (expand_builtin_bzero): Adjust.
31638 (expand_builtin_memcmp): Adjust.
31639 (expand_builtin): Handle memcmp.
31640 (maybe_emit_chk_warning): Check strncat just once.
31642 2017-05-04 Martin Sebor <msebor@redhat.com>
31644 PR preprocessor/79214
31645 PR middle-end/79222
31646 PR middle-end/79223
31647 * builtins.c (check_sizes): Add inlining context and issue
31648 warnings even when -Wno-system-headers is set.
31649 (check_strncat_sizes): Same.
31650 (expand_builtin_strncat): Same.
31651 (expand_builtin_memmove): New function.
31652 (expand_builtin_stpncpy): Same.
31653 (expand_builtin): Handle memmove and stpncpy.
31655 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31657 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31658 which is not used any more.
31660 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31662 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31664 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31666 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31667 (cortexa53_tunings): Likewise.
31668 (cortexa57_tunings): Likewise.
31669 (cortexa72_tunings): Likewise.
31670 (cortexa73_tunings): Likewise.
31672 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31674 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31675 Set loop alignment to 8.
31677 2017-05-04 Martin Sebor <msebor@redhat.com>
31679 PR translation/80280
31680 * builtins.c (expand_builtin_object_size): Add missing quoting to
31681 %D and like directives.
31682 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31683 (hsa_type_for_tree_type): Same.
31684 (verify_function_arguments): Same.
31685 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31686 * varasm.c (get_section): Same.
31689 2017-05-04 Martin Sebor <msebor@redhat.com>
31691 PR translation/80280
31692 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31694 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31696 * config/aarch64/aarch64.c (generic_addrcost_table):
31697 Change HI/TI mode setting.
31699 2017-05-04 Martin Jambor <mjambor@suse.cz>
31701 PR tree-optimization/80622
31702 * tree-sra.c (comes_initialized_p): New function.
31703 (build_accesses_from_assign): Only set write lazily when
31704 comes_initialized_p is false.
31705 (analyze_access_subtree): Use comes_initialized_p.
31706 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31707 instead of testing for PARM_DECL.
31709 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31711 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31712 constraint on operand 0 to allow more general addressing modes.
31713 Adjust output template.
31714 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31716 * config/aarch64/aarch64-protos.h
31717 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31718 * config/aarch64/constraints.md (Dp): New address constraint.
31719 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31722 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31724 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31725 update use of estimate_ipcp_clone_size_and_time.
31726 (estimate_local_effects): Update use of
31727 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31728 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31729 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31730 Return nonspecialized time.
31732 2017-05-04 Richard Biener <rguenther@suse.de>
31734 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31735 for the last VUSE which def dominates the PHI. Directly call
31737 (get_continuation_for_phi_1): Remove.
31739 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31741 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31742 to explain the use of truncating division. Cap the number of
31743 iterations to the maximum given by nb_iterations_upper_bound,
31746 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31748 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31749 * configure: Regenerate.
31750 * config.in: Regenerate.
31751 * config/i386/driver-mingw32.c: new file.
31752 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31753 * config.host: Link driver-mingw32.o on MinGW host.
31754 * doc/install.texi: Document new --enable-mingw-wildcard configure
31757 2017-05-04 Marek Polacek <polacek@redhat.com>
31759 PR tree-optimization/80612
31760 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31762 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31763 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31765 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31766 (movt splitter): Likewise.
31767 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31768 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31769 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31770 block for Thumb-1 with MOVT.
31771 (thumb2_legitimate_address_p): Move code block ...
31772 (can_avoid_literal_pool_for_label_p): ... into this new function.
31773 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31775 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31776 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31777 "M-profile targets with the MOVT instruction".
31779 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31781 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31782 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31783 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31785 2017-05-04 Martin Liska <mliska@suse.cz>
31787 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31788 variable cond_code.
31790 2017-05-04 Richard Biener <rguenther@suse.de>
31792 * tree.c (array_at_struct_end_p): Handle arrays at struct
31793 end with flexarrays more conservatively. Refactor and treat
31794 arrays of arrays or aggregates more strict. Fix
31795 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31796 * tree.c (array_at_struct_end_p): Adjust prototype.
31797 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31798 * gimple-fold.c (get_range_strlen): Likewise.
31799 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31801 2017-05-04 Richard Biener <rguenther@suse.de>
31803 PR tree-optimization/31130
31804 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31806 (supports_overflow_infinity): Likewise.
31807 (is_negative_overflow_infinity): Likewise.
31808 (is_positive_overflow_infinity): Likewise.
31809 (is_overflow_infinity): Likewise.
31810 (stmt_overflow_infinity): Likewise.
31811 (overflow_infinity_range_p): Likewise.
31812 (usable_range_p): Remove as always returning true.
31813 (make_overflow_infinity): Remove.
31814 (negative_overflow_infinity): Likewise.
31815 (positive_overflow_infinity): Likewise.
31816 (avoid_overflow_infinity): Likewise.
31817 (set_value_range): Adjust accordingly.
31818 (set_value_range_to_nonnegative): Likewise, remove now unused
31819 overflow_infinity arg.
31820 (vrp_operand_equal_p): Adjust.
31821 (update_value_range): Likewise.
31822 (range_int_cst_singleton_p): Likewise.
31823 (operand_less_p): Likewise.
31824 (compare_values_warnv): Likewise.
31825 (extract_range_for_var_from_comparison_expr): Likewise.
31826 (vrp_int_const_binop): Likewise.
31827 (zero_nonzero_bits_from_vr): Likewise.
31828 (extract_range_from_multiplicative_op_1): Likewise.
31829 (extract_range_from_binary_expr_1): Likewise.
31830 (extract_range_from_unary_expr): Likewise.
31831 (extract_range_from_comparison): Likewise.
31832 (extract_range_basic): Likewise.
31833 (adjust_range_with_scev): Likewise.
31834 (compare_ranges): Likewise.
31835 (compare_range_with_value): Likewise.
31836 (dump_value_range): Likewise.
31837 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31839 (simplify_cond_using_ranges): Adjust.
31841 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31843 * brig-builtins.def: Added a builtin for class_f64.
31844 * builtin-types.def: Added a builtin type needed by class_f64.
31846 2017-05-03 Jason Merrill <jason@redhat.com>
31848 * timevar.def: Add TV_CONSTEXPR.
31850 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31852 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31854 2017-05-03 Martin Jambor <mjambor@suse.cz>
31856 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31857 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31858 * ipa-cp.c (ipcp_propagate_stage): Do not call
31859 ipa_update_after_lto_read.
31860 * ipa-inline.c (ipa_inline): Likewise.
31862 2017-05-03 Martin Jambor <mjambor@suse.cz>
31864 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31865 tag. Added a default constructor and a destructor.
31866 (ipa_edge_args_sum_t): New class;
31867 (ipa_edge_args_sum): Declare.
31868 (ipa_edge_args_vector): Remove declaration.
31869 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31870 (ipa_free_edge_args_substructures): Remove declaration.
31871 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31872 (ipa_edge_args_info_available_for_edge_p): Likewise.
31873 * ipa-prop.c (ipa_edge_args_vector): Removed.
31874 (edge_removal_hook_holder): Likewise.
31875 (edge_duplication_hook_holder): Likewise.
31876 (ipa_edge_args_sum): New variable.
31877 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31878 ipa_edge_args_vector.
31879 (ipa_free_edge_args_substructures): Likewise.
31880 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31881 ipa_edge_args_vector.
31882 (ipa_edge_removal_hook): Turned into method
31883 ipa_edge_args_sum_t::remove.
31884 (ipa_edge_duplication_hook): Turned into method
31885 ipa_edge_args_sum_t::duplicate.
31886 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31887 registering edge hooks.
31888 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31889 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31890 ipa_edge_args_sum instead of ipa_edge_args_vector.
31891 * ipa-profile.c (ipa_profile): Likewise.
31893 2017-05-03 Martin Jambor <mjambor@suse.cz>
31895 * symbol-summary.h (function_summary): New method exists.
31896 (function_summary::symtab_removal): Deallocate through release.
31897 (call_summary): New class.
31898 (gt_ggc_mx): New overload.
31899 (gt_pch_nx): Likewise.
31900 (gt_pch_nx): Likewise.
31902 2017-05-03 Jeff Law <law@redhat.com>
31904 PR tree-optimization/78496
31905 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31906 from simplify_cond_using_ranges. Split off code to walk
31907 backwards through casts into ...
31908 (simplify_cond_using_ranges_2): New function.
31909 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31910 (execute_vrp): After identifying jump threads, call
31911 simplify_cond_using_ranges_2.
31913 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31916 * ipa-inline.h (inline_summary): Add ctor.
31917 (create_ggc): Do not use ggc_cleared_alloc.
31919 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31920 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31922 * gcc.c (handle_braces): Support escaping in switch matching
31924 * doc/invoke.texi (Spec Files): Document it.
31925 Remove superfluous @code markup in items.
31927 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31929 * diagnostic-show-locus.c (struct column_range): New struct.
31930 (get_affected_columns): New function.
31931 (get_printed_columns): New function.
31932 (struct correction): New struct.
31933 (correction::ensure_capacity): New function.
31934 (correction::ensure_terminated): New function.
31935 (struct line_corrections): New struct.
31936 (line_corrections::~line_corrections): New dtor.
31937 (line_corrections::add_hint): New function.
31938 (layout::print_trailing_fixits): Reimplement in terms of the new
31940 (selftest::test_overlapped_fixit_printing): New function.
31941 (selftest::diagnostic_show_locus_c_tests): Call it.
31943 2017-05-03 Nathan Sidwell <nathan@acm.org>
31945 Canonicalize canonical type hashing
31946 * tree.h (type_hash_canon_hash): Declare.
31947 * tree.c (type_hash_list, attribute_hash_list): Move into
31948 type_hash_canon_hash.
31949 (build_type_attribute_qual_variant): Break out hash code calc into
31950 type_hash_canon_hash.
31951 (type_hash_canon_hash): New. Generic type hash computation.
31952 (build_range_type_1, build_array_type_1, build_function_type,
31953 build_method_type_directly, build_offset_type, build_complex_type,
31954 make_vector_type): Call it.
31956 2017-05-03 Richard Biener <rguenther@suse.de>
31958 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31959 When all DRs have unknown misaligned do not always peel
31960 when there is a store but apply the same costing model as if
31961 there were only loads.
31963 2017-05-03 Richard Biener <rguenther@suse.de>
31966 PR tree-optimization/80492
31967 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31968 compare_base_decls returning dont-know properly.
31970 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31972 * config/arm/iterators.md (CCSI): New mode iterator.
31973 (arch): New mode attribute.
31974 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31975 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31976 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31977 code iterator for success result mode.
31978 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31979 the corresponding new insn generators.
31981 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31984 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31985 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31987 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31989 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31990 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31991 (DDR_A): Wrap DDR argument in brackets.
31992 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31993 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31994 (DDR_REVERSED_P): Likewise.
31996 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31998 PR tree-optimization/79472
31999 * tree-switch-conversion.c (struct switch_conv_info): Add
32000 contiguous_range and default_case_nonstandard fields.
32001 (collect_switch_conv_info): Compute contiguous_range and
32002 default_case_nonstandard fields, don't clear final_bb if
32003 contiguous_range and only the default case doesn't have the required
32005 (check_all_empty_except_final): Set default_case_nonstandard instead
32006 of failing if contiguous_range and the default case doesn't have empty
32008 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32009 and only the default case doesn't have the required constants. Skip
32011 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32012 if default_case_nonstandard.
32013 (build_constructors): Build constant 1 just once. Assert that default
32014 values aren't inserted in between cases if contiguous_range. Skip
32016 (build_arrays): Skip virtual phis.
32017 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32018 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32019 Handle virtual phis.
32020 (gen_inbound_check): Handle default_case_nonstandard case.
32021 (process_switch): Adjust check_final_bb caller. Call
32022 gather_default_values with the first non-default case instead of
32023 default case if default_case_nonstandard.
32025 2017-05-02 Nathan Sidwell <nathan@acm.org>
32027 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32028 check. Fix formatting.
32030 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32032 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32033 errors when comparing specialized and unspecialized times.
32035 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32037 * diagnostic-show-locus.c
32038 (layout::should_print_annotation_line_p): Make private.
32039 (layout::print_annotation_line): Make private.
32040 (layout::annotation_line_showed_range_p): Make private.
32041 (layout::show_ruler): Make private.
32042 (layout::print_source_line): Make private. Pass in line and
32043 line_width, rather than calling location_get_source_line. Drop
32045 (layout::print_leading_fixits): New method.
32046 (layout::print_any_fixits): Rename to...
32047 (layout::print_trailing_fixits): ...this, and make private.
32048 Don't print newline fixits.
32049 (diagnostic_show_locus): Move logic for printing one row into...
32050 (layout::print_line): ...this new function. Move the
32051 location_get_source_line call and error-handling from
32052 print_source_line to here. Call print_leading_fixits, and rename
32053 print_any_fixits to print_trailing_fixits.
32054 (selftest::test_fixit_insert_containing_newline): Update now that
32055 newlines are partially supported.
32056 (selftest::test_fixit_insert_containing_newline_2): New test.
32057 (selftest::test_fixit_replace_containing_newline): Update comments.
32058 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32059 * edit-context.c (class added_line): New class.
32060 (class edited_line): Describe newline handling in comment.
32061 (edited_line::actually_edited_p): New method.
32062 (edited_line::print_content): Delete redundant decl.
32063 (edited_line::m_predecessors): New field.
32064 (edited_file::print_content): Call edited_line::print_content.
32065 (edited_file::print_diff): Update to support newlines.
32066 (edited_file::print_diff_hunk): Likewise.
32067 (edited_file::print_run_of_changed_lines): New function.
32068 (edited_file::print_diff_line): Convert to...
32069 (print_diff_line): ...this.
32070 (edited_file::get_effective_line_count): New function.
32071 (edited_line::edited_line): Initialize new field m_predecessors.
32072 (edited_line::~edited_line): Clean up m_predecessors.
32073 (edited_line::apply_fixit): Handle newlines.
32074 (edited_line::get_effective_line_count): New function.
32075 (edited_line::print_content): New function.
32076 (edited_line::print_diff_lines): New function.
32077 (selftest::test_applying_fixits_insert_containing_newline): New
32079 (selftest::test_applying_fixits_replace_containing_newline): New
32081 (selftest::insert_line): New function.
32082 (selftest::test_applying_fixits_multiple_lines): Add example of
32084 (selftest::edit_context_c_tests): Call the new tests.
32086 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32088 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32089 parameter cand. Update dump information.
32090 (get_computation_cost): Update uses.
32092 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32094 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32095 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32096 (get_computation_at, rewrite_use_address): Update use of
32097 get_computation_aff.
32099 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32101 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32102 (get_computation): Delete.
32103 (get_computation_cost): Implement like get_computation_cost_at.
32104 Use get_computation_at.
32105 (get_computation_cost_at): Delete.
32106 (rewrite_use_nonlinear_expr): Use get_computation_at.
32107 (rewrite_use_compare, remove_unused_ivs): Ditto.
32109 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32111 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32113 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32115 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32116 (ivopts_global_cost_for_size): Rename parameter and update uses.
32117 (iv_ca_recount_cost): Update uses.
32118 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32119 candidates seperately in n_invs and n_cands.
32120 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32122 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32124 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32125 (find_inv_vars_cb): New.
32126 (find_depends): Renamed to ...
32127 (find_inv_vars): ... this.
32128 (add_candidate_1, force_var_cost): Call find_inv_vars.
32129 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32131 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32133 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32134 inv_vars. Add inv_exprs.
32135 (struct iv_cand): Rename depends_on to inv_vars.
32136 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32137 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32138 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32139 (dump_cand): Dump inv_vars.
32140 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32141 (record_invariant, find_depends, add_candidate_1): Ditto.
32142 (set_group_iv_cost, force_var_cost): Ditto.
32143 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32144 (get_computation_cost_at, get_computation_cost): Ditto.
32145 (determine_group_iv_cost_generic): Ditto.
32146 (determine_group_iv_cost_address): Ditto.
32147 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32148 (determine_group_iv_costs): Ditto.
32149 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32150 (iv_ca_set_remove_invariants): Renamed to ...
32151 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32152 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32153 (iv_ca_set_add_invariants): Renamed to ...
32154 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32155 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32156 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32157 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32158 (create_new_ivs): Remove useless dump.
32160 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32162 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32164 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32165 (iv_ca_set_no_cp, create_new_iv): Ditto.
32167 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32169 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32171 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32173 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32174 function tree_check2.
32176 2017-05-02 Martin Liska <mliska@suse.cz>
32178 * doc/gcov.texi: Add missing preposition.
32179 * gcov.c (function_info::function_info): Properly fill up
32180 all member variables.
32182 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32184 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32186 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32188 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32190 2017-05-02 Martin Liska <mliska@suse.cz>
32193 * lto-streamer-in.c (lto_read_tree_1): Remove
32194 LTO_STREAMER_DEBUG.
32195 * lto-streamer.c (struct tree_hash_entry): Likewise.
32196 (struct tree_entry_hasher): Likewise.
32197 (tree_entry_hasher::hash): Likewise.
32198 (tree_entry_hasher::equal): Likewise.
32199 (lto_streamer_init): Likewise.
32200 (lto_orig_address_map): Likewise.
32201 (lto_orig_address_get): Likewise.
32202 (lto_orig_address_remove): Likewise.
32203 * lto-streamer.h: Likewise.
32204 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32205 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32207 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32209 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32210 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32211 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32212 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32213 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32214 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32215 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32216 (mm_maskz_sub_ss): New intrinsics.
32217 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32218 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32219 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32220 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32221 (__builtin_ia32_subss_mask_round): New builtins.
32222 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32223 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32224 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32226 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32227 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32229 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32232 2017-05-02 Martin Jambor <mjambor@suse.cz>
32234 PR tree-optimization/78687
32235 * tree-sra.c (access): New field parent.
32236 (process_subtree_disqualification): New function.
32237 (disqualify_candidate): Call it.
32238 (build_accesses_from_assign): Reset write flag if creating an
32240 (build_access_subtree): Fill in parent field and also prpagate
32241 down grp_write flag.
32242 (create_artificial_child_access): New parameter set_grp_write, set
32243 grp_write to its value.
32244 (propagate_subaccesses_across_link): Also propagate grp_write flag
32246 (propagate_all_subaccesses): Push the closest parent back to work
32247 queue if add_access_to_work_queue returned true.
32249 2017-05-02 Richard Biener <rguenther@suse.de>
32251 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32252 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32253 -fstrict-overflow documentation.
32254 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32255 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32256 flag_strict_overflow.
32257 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32258 * lto-opts.c (lto_write_options): Do not stream it.
32259 * lto-wrapper.c (merge_and_complain): Do not handle it.
32260 * opts.c (default_options_table): Do not set -fstrict-overflow.
32261 (finish_options): Likewise do not clear it when sanitizing.
32262 * simplify-rtx.c (simplify_const_relational_operation): Do not
32263 test flag_strict_overflow.
32265 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32267 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32268 using enabled attribute.
32269 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32270 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32271 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32272 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32273 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32274 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32275 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32276 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32277 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32278 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32280 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32282 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32284 2017-05-02 Richard Biener <rguenther@suse.de>
32286 PR tree-optimization/80591
32288 2017-04-10 Richard Biener <rguenther@suse.de>
32290 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32293 2017-05-02 Richard Biener <rguenther@suse.de>
32295 PR tree-optimization/80549
32296 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32297 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32298 headers if they do not have a preheader.
32300 2017-05-02 Martin Liska <mliska@suse.cz>
32303 * common.opt: Fix typo.
32304 * doc/invoke.texi: Likewise.
32306 2017-05-01 Jan Beulich <jbeulich@suse.com>
32308 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32309 swapping, add (x,x,m,x,n) alternative.
32311 2017-05-01 Nathan Sidwell <nathan@acm.org>
32313 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32314 unnecessary unadjusted_alignment check.
32316 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32319 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32321 * gimplify.c (gimplify_cilk_detach): New function.
32322 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32323 * tree-core.h: Document EXPR_CILK_SPAWN.
32324 * tree.h (EXPR_CILK_SPAWN): Define.
32326 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32328 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32329 to use new fixit_hint representation, using the "replace" logic.
32330 (get_line_span_for_fixit_hint): Likewise.
32331 (layout::print_any_fixits): Likewise.
32332 (selftest::test_one_liner_many_fixits): Rename to...
32333 (selftest::test_one_liner_many_fixits_1): ...this, and update
32334 comment and expected output to reflect that the multiple fix-it
32335 hints are now consolidated into one insertion.
32336 (selftest::test_one_liner_many_fixits_2): New test.
32337 (selftest::test_diagnostic_show_locus_one_liner): Update for
32339 (selftest::test_fixit_consolidation): Update for fix-it API
32341 * diagnostic.c (print_parseable_fixits): Likewise.
32342 * edit-context.c (edited_line::m_line_events): Convert from
32343 auto_vec <line_event *> to auto_vec <line_event>.
32344 (class line_event): Convert from abstract base class to a concrete
32345 class, taking over the role of replace_event.
32346 (class insert_event): Delete.
32347 (class replace_event): Rename to class line_event. Convert to
32349 (edit_context::add_fixits): Reimplement.
32350 (edit_context::apply_insert): Delete.
32351 (edit_context::apply_replace): Rename to...
32352 (edit_context::apply_fixit): ...this. Convert to half-open range.
32353 (edited_file::apply_insert): Delete.
32354 (edited_file::apply_replace): Rename to...
32355 (edited_file::apply_fixit): ...this.
32356 (edited_line::~edited_line): Drop deletion of events.
32357 (edited_line::apply_insert): Delete.
32358 (edited_line::apply_replace): Rename to...
32359 (edited_line::apply_fixit): ...this. Convert to half-open range.
32360 Update for change to type of m_line_events.
32361 * edit-context.h (edit_context::apply_insert): Delete.
32362 (edit_context::apply_replace): Rename to...
32363 (edit_context::apply_fixit): ...this.
32365 2017-05-01 Martin Sebor <msebor@redhat.com>
32367 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32370 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32373 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32374 __get_cpuid_max returns 0.
32375 (__get_cpuid_count): Ditto.
32377 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32379 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32380 replacement expression is another instance of one of its arguments.
32382 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32385 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32386 check for stack push/pop autoinc.
32387 * config/i386/i386.c (ix86_agi_dependent): Return false
32388 if the only reason why modified_in_p returned true is that
32389 addr is SP based and set_insn is a push or pop.
32391 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32393 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32396 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32399 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32400 (account_size_time): Use two predicates - exec_pred and
32402 (evaluate_conditions_for_known_args): Compute both clause and
32404 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32405 (inline_summary_t::duplicate): Update.
32406 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32408 (compute_inline_parameters): Likewise.
32409 (estimate_edge_size_and_time): Update caluclation of time.
32410 (estimate_node_size_and_time): Compute both time and nonspecialized
32412 (estimate_ipcp_clone_size_and_time): Update.
32413 (inline_merge_summary): Update.
32414 (do_estimate_edge_time): Update.
32415 (do_estimate_edge_size): Update.
32416 (do_estimate_edge_hints): Update.
32417 (inline_read_section, inline_write_summary): Stream both new predicates.
32418 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32420 (compute_inlined_call_time): Cleanup.
32421 (big_speedup_p): Update.
32422 (edge_badness): Update.
32423 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32424 (size_time_entry): Replace predicate by exec_predicate and
32425 nonconst_predicate.
32426 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32427 (estimate_edge_time): Return also nonspec_time.
32428 (reset_edge_growth_cache): Update.
32430 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32432 PR rtl-optimization/80491
32433 * ifcvt.c (noce_process_if_block): When looking for x setter
32434 with missing else_bb, don't check only the insn right before
32435 cond_earliest, but look for the last insn that x is modified in
32436 within the same bb.
32438 PR rtl-optimization/80491
32439 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32441 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32443 PR tree-optimization/80487
32444 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32446 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32448 PR tree-optimization/79697
32449 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32450 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32451 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32453 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32454 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32456 2017-04-28 Martin Sebor <msebor@redhat.com>
32458 PR tree-optimization/80523
32459 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32460 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32462 (maybe_warn, format_directive, parse_directive): Use new functions.
32463 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32465 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32467 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32469 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32471 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32472 target_header_dir): Set correctly.
32473 * configure: Regenerated.
32474 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32475 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32476 instead of SYSTEM_HEADER_DIR.
32478 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32480 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32481 (estimate_local_effects): Likewise.
32482 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32483 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32484 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32485 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32486 do_estimate_edge_time, estimate_edge_time): Likewise.
32487 * ipa-inline-analysis.c (estimate_node_size_and_time,
32488 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32489 (estimate_time_after_inlining): Remove.
32491 2017-04-28 Martin Liska <mliska@suse.cz>
32493 * doc/gcov.texi: Enhance documentation of gcov.
32495 2017-04-28 Martin Liska <mliska@suse.cz>
32497 * doc/gcov.texi: Sort options in alphabetic order.
32498 * doc/gcov-dump.texi: Likewise.
32499 * doc/gcov-tool.texi: Likewise.
32500 * gcov.c (print_usage): Likewise.
32501 * gcov-dump.c (print_usage): Likewise.
32502 * gcov-tool.c (print_merge_usage_message): Likewise.
32503 (print_rewrite_usage_message): Likewise.
32504 (print_overlap_usage_message): Likewise.
32506 2017-04-28 Martin Liska <mliska@suse.cz>
32508 PR gcov-profile/53915
32509 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32511 2017-04-28 Martin Liska <mliska@suse.cz>
32513 PR gcov-profile/79891
32514 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32515 is marked by compiler as living on a line.
32516 (get_cycles_count): Remove usage of the union.
32517 (output_intermediate_file): Likewise.
32518 (find_source): Fix GNU coding style.
32519 (accumulate_line_counts): Remove old non-all block mode.
32520 (output_lines): Remove usage of the union.
32521 * profile.c (output_location): Include all BBs, even if
32522 belonging to a same line (and file) as a previous BB.
32524 2017-04-28 Martin Liska <mliska@suse.cz>
32526 * gcov.c (process_args): Handle new argument 'w'.
32527 (read_graph_file): Assign ID to BBs.
32528 (output_branch_count): Display BB # if verbose flag is set.
32529 (output_lines): Likewise for arcs.
32530 (print_usage): Add '--verbose' option help.
32531 * doc/gcov.texi: Document --verbose (-w) option.
32533 2017-04-28 Martin Liska <mliska@suse.cz>
32535 * gcov.c (struct block_location_info): New struct.
32536 (process_file): Fill up the new structure.
32537 (read_graph_file): Replace usage of encoding by the newly added
32539 (add_line_counts): Likewise.
32540 (accumulate_line_counts): Remove usage of the union.
32541 (function_info::function_info): New function.
32542 (function_info::~function_info): Likewise.
32543 (process_file): Call delete instead of release_function.
32544 (release_function): Release the function.
32545 (release_structures): Call delete instead of release_function.
32546 (solve_flow_graph): Replace usage of num_blocks.
32547 (find_exception_blocks): Likewise.
32548 (output_lines): Fix GNU coding style.
32550 2017-04-28 Martin Liska <mliska@suse.cz>
32553 * coverage.c (coverage_remove_note_file): New function.
32554 * coverage.h: Declare the function.
32555 * toplev.c (finalize): Clean if an error has been seen.
32557 2017-04-28 Martin Liska <mliska@suse.cz>
32559 PR gcov-profile/80031
32560 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32561 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32562 * gcov.c (read_graph_file): Read just number of blocks.
32563 * profile.c (branch_prob): Do not stream 0 flags per a basic
32566 2017-04-28 Martin Liska <mliska@suse.cz>
32568 * gcov-dump.c (tag_*): Add new argument to declarations.
32569 (dump_gcov_file): Likewise.
32570 (tag_blocks): Add and use new argument depth.
32571 (tag_arcs): Likewise.
32572 (tag_lines): Likewise.
32573 (tag_counters): Likewise.
32574 (tag_summary): Likewise.
32575 (dump_working_sets): Use depth to do a proper indentation.
32577 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32580 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32581 * symtab.c (symtab_node::debug_symtab): Move definition here.
32583 2017-04-28 Richard Biener <rguenther@suse.de>
32585 * lto-streamer.h (LTO_major_version): Bump to 7.
32587 2017-04-28 Richard Biener <rguenther@suse.de>
32589 * tree-vrp.c (assert_info): New struct.
32590 (add_assert_info): New helper.
32591 (register_edge_assert_for_2): Refactor to add asserts to a vector
32593 (register_edge_assert_for_1): Likewise.
32594 (register_edge_assert_for): Likewise.
32595 (finish_register_edge_assert_for): New helper actually registering
32596 asserts where live on edge.
32597 (find_conditional_asserts): Adjust.
32598 (find_switch_asserts): Likewise.
32599 (evrp_dom_walker::try_find_new_range): Generalize.
32600 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32602 2017-04-27 Marek Polacek <polacek@redhat.com>
32605 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32606 arg10 and arg11 to itype.
32608 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32610 * doc/extend.texi (Object Size Checking): Improve grammar.
32612 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32615 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32616 that the logic for permitting reciprocal estimates matches that
32619 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32622 * tree.c (type_cache_hasher::equal): Only compare
32623 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32624 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32625 non-aggregate element types.
32626 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32627 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32629 2017-04-27 Richard Biener <rguenther@suse.de>
32631 PR middle-end/80533
32632 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32633 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32634 keeping a reference to a trailing array.
32636 2017-04-27 Richard Biener <rguenther@suse.de>
32638 PR middle-end/80539
32639 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32640 being in loop-closed SSA form conservatively.
32641 (chrec_fold_multiply_poly_poly): Likewise.
32643 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32645 PR middle-end/79665
32646 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32647 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32649 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32652 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32653 (aarch64_function_arg_alignment): Return unsigned int again, but still
32654 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32655 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32656 Don't emit -Wpsabi note.
32657 (aarch64_function_arg_boundary): Likewise.
32658 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32661 2017-04-26 Nathan Sidwell <nathan@acm.org>
32663 * tree.h (crc32_unsigned_n): Declare.
32664 (crc32_unsigned, crc32_unsigned): Make inline.
32665 * tree.c (crc32_unsigned_bits): Replace with ...
32666 (crc32_unsigned_n): ... this.
32667 (crc32_unsigned, crc32_byte): Remove.
32668 (crc32_string): Remove unnecessary braces.
32670 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32672 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32673 * ipa-inline-analysis.c (MAX_TIME): Remove.
32674 (account_size_time): Use sreal for time.
32675 (dump_inline_summary): Update.
32676 (estimate_function_body_sizes): Update.
32677 (estimate_edge_size_and_time): Update.
32678 (estimate_calls_size_and_time): Update.
32679 (estimate_node_size_and_time): Update.
32680 (inline_merge_summary): Update.
32681 (inline_update_overall_summary): Update.
32682 (estimate_time_after_inlining): Update.
32683 (inline_read_section): Update.
32684 (inline_write_summary): Update.
32685 * ipa-inline.c (compute_uninlined_call_time): Update.
32686 (compute_inlined_call_time): Update.
32687 (recursive_inlining): Update.
32688 (inline_small_functions): Update.
32689 (dump_overall_stats): Update.
32690 * ipa-inline.h: Include sreal.h.
32691 (size_time_entry): Turn time to sreal.
32692 (inline_summary): Turn self_time nad time to sreal.
32694 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32696 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32698 (sreal::stream_out, sreal::stream_in): New.
32699 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32701 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32703 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32706 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32709 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32711 (dimode_scalar_chain::compute_convert_gain): Ditto.
32712 (dimode_scalar_chain::make_vector_copies): Ditto.
32713 (dimode_scalar_chain::convert_reg): Ditto.
32714 (dimode_scalar_chain::convert_insn): Ditto.
32715 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32716 (VI248_AVX512BW_1): New mode iterator.
32717 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32718 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32721 2017-04-25 Martin Sebor <msebor@redhat.com>
32723 PR tree-optimization/80497
32724 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32725 constants are representable in HOST_WIDE_INT.
32726 (parse_directive): Ditto.
32728 2017-04-25 Martin Sebor <msebor@redhat.com>
32731 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32732 (new_zero_array): Adjust signature.
32733 (dom_info::dom_init): Used unsigned rather that size_t.
32734 (dom_info::dom_info): Same.
32736 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32737 Jakub Jelinek <jakub@redhat.com>
32740 * config/arm/arm.c: Include gimple.h.
32741 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32742 returns negative, increment ncrn only if it returned positive.
32743 (arm_needs_doubleword_align): Return int instead of bool,
32744 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32745 members, but if there is any such non-FIELD_DECL
32746 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32747 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32748 returns negative, increment nregs only if it returned positive.
32749 (arm_setup_incoming_varargs): Likewise.
32750 (arm_function_arg_boundary): Emit -Wpsabi note if
32751 arm_needs_doubleword_align returns negative, return
32752 DOUBLEWORD_ALIGNMENT only if it returned positive.
32754 2017-04-25 Marek Polacek <polacek@redhat.com>
32757 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32758 first argument to type.
32760 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32763 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32764 type checks to test for compatibility instead of equality.
32766 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32767 Jakub Jelinek <jakub@redhat.com>
32770 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32772 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32773 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32774 the alignment computation, but return their maximum in warn_alignment.
32775 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32776 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32778 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32779 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32782 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32784 * config/arc/simdext.md (dmpyh): Fix typo.
32786 2017-04-25 Richard Biener <rguenther@suse.de>
32788 PR tree-optimization/80492
32789 * alias.c (compare_base_decls): Handle registers with asm
32790 specification conservatively.
32791 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32792 compare_base_decls returning dont-know properly.
32794 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32796 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32797 (legitimate_offset_address_p): New function.
32798 (arc_legitimate_address_p): Use above function.
32800 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32802 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32804 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32806 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32807 ACCH registers whenever they are available.
32809 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32811 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32812 double regs fix when not used.
32814 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32816 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32818 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32819 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32821 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32823 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32824 when emitting short ADD instructions.
32826 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32828 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32830 (cmpsi_cc_c_insn): Likewise.
32831 (cbranchsi4_scratch): Compute proper instruction length using
32832 compact_hreg_operand.
32833 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32835 2017-04-25 Richard Biener <rguenther@suse.de>
32837 PR middle-end/80509
32838 * passes.c (pass_manager::pass_manager): Initialize
32839 m_name_to_pass_map.
32841 2017-04-25 Richard Biener <rguenther@suse.de>
32843 PR tree-optimization/79201
32844 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32846 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32849 * config/s390/vector.md: Split MEM->GPR vector moves for
32850 non-s_operand addresses.
32852 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32855 * config/s390/predicates.md (reload_const_wide_int_operand): New
32857 * config/s390/s390.md ("movti"): Remove d/P alternative.
32858 ("movti_bigconst"): New pattern definition.
32860 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32863 * s390-protos.h (s390_expand_cs_hqi): Removed.
32864 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32865 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32866 modes as well as CCZ1mode and CCZmode.
32867 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32868 signature of s390_emit_compare_and_swap.
32869 (s390_expand_cs_hqi): Likewise, make static.
32870 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32871 compare-and-swap, in some cases.
32872 (s390_expand_cs): Wrapper function.
32873 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32875 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32876 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32877 patterns for small and large integers. Forbid symref memory operands.
32878 Move expander to s390.c. Require cc register.
32879 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32880 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32881 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32882 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32883 symref memory operands. Remove CC mode and call s390_match_ccmode
32885 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32887 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32889 * config/s390/s390.md (define_peephole2): New peephole to help
32890 combining the load-and-test pattern with volatile memory.
32892 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32894 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32895 with CCZmode for TARGET_Z196.
32897 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32899 PR rtl-optimization/80501
32900 * combine.c (make_compound_operation_int): Set subreg_code to SET
32901 even for AND with mask of the sign bit of mode.
32903 PR rtl-optimization/80500
32904 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32905 sum's initial value.
32907 2017-04-25 Julian Brown <julian@codesourcery.com>
32908 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32910 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32912 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32914 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32916 2017-04-25 Julian Brown <julian@codesourcery.com>
32917 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32919 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32920 (thunderx2t99_sha): New Reservation.
32922 2017-04-25 Julian Brown <julian@codesourcery.com>
32923 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32925 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32926 type for 1-element load.
32928 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32930 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32932 2017-04-24 Martin Jambor <mjambor@suse.cz>
32934 PR tree-optimization/80293
32935 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32936 char arrays not totally scalarizable if it is false.
32937 (analyze_all_variable_accesses): Pass correct value in the new
32938 parameter. Add a statistics counter.
32940 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32942 PR middle-end/79931
32943 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32945 2017-04-24 Richard Biener <rguenther@suse.de>
32947 PR tree-optimization/80494
32948 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32949 out for complex types.
32951 2017-04-24 Richard Biener <rguenther@suse.de>
32953 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32954 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32955 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32956 (DFS): Adjust and never fail.
32957 (sccvn_dom_walker::fail): Remove.
32958 (sccvn_dom_walker::before_dom_children): Adjust.
32959 (run_scc_vn): Likewise and never fail.
32960 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32961 (pass_fre::execute): Likewise.
32963 2017-04-24 Richard Biener <rguenther@suse.de>
32965 PR tree-optimization/79725
32966 * tree-ssa-sink.c (statement_sink_location): Return whether
32967 failure reason was zero uses. Move that check later.
32968 (sink_code_in_bb): Deal with zero uses by removing the stmt
32971 2017-04-24 Richard Biener <rguenther@suse.de>
32974 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32975 pointer-based references.
32977 2017-04-24 Richard Biener <rguenther@suse.de>
32980 * pass_manager.h (pass_manager::operator new): Remove.
32981 (pass_manager::operator delete): Likewise.
32982 * passes.c (pass_manager::operator new): Remove.
32983 (pass_manager::operator delete): Likewise.
32984 (pass_manager::pass_manager): Zero individual pass members.
32986 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32989 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32990 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32991 Check "XEXP (src, 1)" operand here.
32992 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32993 Check "XEXP (src, 1)" operand here.
32994 (dimode_scalar_chain::make_vector_copies): Detect count register
32995 of a shift instruction. Zero extend count register from QImode
32996 to DImode to satisfy vector shift pattern count operand predicate.
32997 Substitute vector shift count operand with a DImode copy.
32998 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33001 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33003 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33004 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33005 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33006 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33007 (UNSPEC_NOREX_MEM): Remove definition.
33009 2017-04-21 Richard Biener <rguenther@suse.de>
33011 PR tree-optimization/79547
33012 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33013 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33014 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33015 without any constraints.
33017 2017-04-21 Richard Biener <rguenther@suse.de>
33019 PR tree-optimization/78847
33020 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33022 2017-04-21 Richard Biener <rguenther@suse.de>
33024 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33025 (build_distinct_type_copy): Likewise.
33026 (build_variant_type_copy): Likewise.
33027 * tree.c (build_qualified_type): Pass down mem-stat info.
33028 (build_distinct_type_copy): Likewise.
33029 (build_variant_type_copy): Likewise.
33031 2017-04-21 Richard Biener <rguenther@suse.de>
33033 PR tree-optimization/80237
33034 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33036 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33037 for a simplified result.
33039 2016-04-21 Richard Biener <rguenther@suse.de>
33041 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33042 sth as strict as a simple_iv but a chrec without symbols and an
33043 operand defined in the loop we are peeling (and not some subloop).
33044 (propagate_constants_for_unrolling): Propagate all constants.
33046 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33049 * config/i386/i386.c (print_reg): Remove assert for disalowed
33050 regno values, call output_operand_lossage instead.
33052 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33055 * config/i386/constraints.md (Yc): New register constraint.
33056 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33057 Use Yc constraint for alternative 2 of operand 0. Remove
33058 preferred_for_speed attribute.
33060 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33062 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33063 lastprivate clauses in SIMT case.
33065 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33067 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33069 2017-04-20 Richard Biener <rguenther@suse.de>
33071 PR tree-optimization/57796
33072 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33073 as N scalar stores.
33074 (vect_model_load_cost): Cost gathers as N scalar loads.
33076 2017-04-20 Richard Biener <rguenther@suse.de>
33078 * ggc-page.c (ggc_allocated_p): Rename to ...
33079 (safe_lookup_page_table_entry): ... this and return the lookup
33081 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33083 2017-04-20 Richard Biener <rguenther@suse.de>
33085 PR tree-optimization/80453
33086 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33087 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33088 from the conditions.
33089 (vn_phi_eq): Pass them down.
33090 (vn_phi_lookup): Record them.
33091 (vn_phi_insert): Likewise.
33093 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33095 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33096 uninitialized variable warning to avoid buffer overrun.
33098 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33101 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33102 is suppressed for '{ 0 }' in C.
33104 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33106 * BASE-VER: Set to 8.0.0.
33108 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33110 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33111 priority .init_array and .fini_array section with SECTION_NOTYPE
33114 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33116 PR middle-end/80423
33117 * tree.h (build_array_type): Add typeless_storage default argument.
33118 * tree.c (type_cache_hasher::equal): Also compare
33119 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33120 (build_array_type): Add typeless_storage argument, set
33121 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33123 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33124 (build_array_type): Likewise. Add typeless_storage argument.
33126 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33127 Jakub Jelinek <jakub@redhat.com>
33129 PR tree-optimization/80426
33130 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33131 operation on symbolic operands, also compute the overflow for the
33132 invariant part when the operation degenerates into a negation.
33134 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33137 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33138 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33141 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33143 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33146 * config/avr/avr.c (tree.h): Include it.
33147 (cgraph.h): Include it.
33148 (avr_encode_section_info): Don't warn for uninitialized progmem
33149 variable if it's just an alias.
33151 2017-04-19 Richard Biener <rguenther@suse.de>
33154 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33155 when needed by AutoPGO.
33157 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33160 * doc/lto.texi: Remove an extra 'that'.
33162 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33164 PR rtl-optimization/80429
33165 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33166 are only used in debug insns.
33168 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33169 Vladimir Makarov <vmakarov@redhat.com>
33171 * config/sparc/predicates.md (input_operand): Add comment. Return
33172 true for any memory operand when LRA is in progress.
33173 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33175 2017-04-18 Jeff Law <law@redhat.com>
33178 * mips.md ({return,simple_return}_internal): Do not overwrite
33181 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33183 PR tree-optimization/80443
33184 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33185 instead of adding 1, subtract -1 and similarly instead of subtracting
33188 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33190 PR rtl-optimization/80357
33191 * haifa-sched.c (tmp_bitmap): New variable.
33192 (model_recompute): Handle duplicate use records.
33193 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33194 (free_global_sched_pressure_data): Free it.
33196 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33199 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33200 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33201 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33202 instead of SYSTEM_HEADER_DIR.
33204 2017-04-18 Jeff Law <law@redhat.com>
33206 PR middle-end/80422
33207 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33208 predecessors after walking up the insn chain.
33210 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33213 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33214 sizetype type into debug info.
33216 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33219 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33220 unneeded test for TARGET_UPPER_REGS_SF.
33221 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33223 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33226 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33227 instead of gsi_after_labels.
33229 2017-04-18 Jeff Law <law@redhat.com>
33231 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33235 2017-04-13 Jeff Law <law@redhat.com>
33236 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33237 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33239 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33242 * config/avr/avr.c (intl.h): Include it.
33243 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33245 2017-04-18 Martin Liska <mliska@suse.cz>
33247 PR gcov-profile/78783
33248 * gcov-tool.c (gcov_output_files): Validate that destination
33249 file is either removed by the tool or by a user.
33251 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33252 Guy Benyei <guybe@mellanox.com>
33254 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33255 block, and do not negate it, the stored id is already negative.
33257 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33259 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33261 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33264 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33265 masks of options that should be turned off if the VSX vector
33266 options are turned off.
33267 (OTHER_P8_VECTOR_MASKS): Likewise.
33268 (OTHER_VSX_VECTOR_MASKS): Likewise.
33269 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33270 rs6000_disable_incompatible_switches to validate no type switches
33272 (rs6000_incompatible_switch): New function to disallow turning on
33273 other vector options if -mno-vsx, -mno-power8-vector, or
33274 -mno-power9-vector are specified.
33276 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33278 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33280 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33282 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33283 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33284 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33285 (ARG_POINTER_CFA_OFFSET): Likewise.
33287 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33289 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33290 conditions to take advantage of various optimizations.
33292 2017-04-13 Jeff Law <law@redhat.com>
33294 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33295 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33296 (zero_extendsidi2_dext): Likewise.
33298 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33301 * fold-const.c (fold_ternary_loc): Revert
33302 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33305 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33307 PR rtl-optimization/80343
33308 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33309 new scratch pseudo.
33311 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33314 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33315 to ubsan_encode_value.
33317 2017-04-13 Jeff Law <law@redhat.com>
33319 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33320 appearing in DEBUG_INSNs.
33322 2017-04-13 Martin Liska <mliska@suse.cz>
33324 PR gcov-profile/80413
33325 * gcov-io.c (gcov_write_string): Copy to buffer just when
33326 allocated size is greater than zero.
33328 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33331 * dwarf2out.c (decls_for_scope): Ignore declarations of
33332 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33334 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33337 * ipa-visibility.c (non_local_p): Fix typos.
33338 (localize_node): When localizing symbol in same comdat group,
33339 dissolve the group only when we know external symbols are going
33341 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33343 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33345 PR tree-optimization/79390
33346 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33347 order does not result in usable sequence, retry with reversed operand
33353 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33354 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33355 op0 instead of fold_convert_loc (loc, type, arg0).
33357 2017-04-12 Jeff Law <law@redhat.com>
33359 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33360 has a delay slot in the generated code.
33362 * config/cris/cris.md (cris_preferred_reload_class): Return
33363 GENNONACR_REGS rather than GENERAL_REGS.
33365 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33368 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33369 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33370 signedness of the result type.
33372 2017-04-12 Richard Biener <rguenther@suse.de>
33373 Jeff Law <law@redhat.com>
33375 PR tree-optimization/80359
33376 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33377 trim stores to TARGET_MEM_REFs.
33379 2017-04-12 Richard Biener <rguenther@suse.de>
33381 PR tree-optimization/79390
33382 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33383 threading case even more.
33385 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33388 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33389 for quad_address_p for TImode, instead of just not indexed_address.
33391 2017-04-12 Richard Biener <rguenther@suse.de>
33392 Bernd Edlinger <bernd.edlinger@hotmail.de>
33394 PR middle-end/79671
33395 * alias.c (component_uses_parent_alias_set_from): Handle
33396 TYPE_TYPELESS_STORAGE.
33397 (get_alias_set): Likewise.
33398 * tree-core.h (tree_type_common): Add typeless_storage flag.
33399 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33400 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33401 for types containing members with TYPE_TYPELESS_STORAGE.
33402 (place_field): Likewise.
33403 (layout_type): Likewise for ARRAY_TYPE.
33404 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33405 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33406 TYPE_TYPELESS_STORAGE.
33407 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33409 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33412 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33413 first argument to type.
33415 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33419 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33420 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33421 (rs6000_expand_binop_builtin): Likewise.
33422 (rs6000_expand_ternop_builtin): Likewise; also add missing
33423 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33424 vshasigma built-ins.
33425 * doc/extend.texi: Document that vec_xxpermdi's third argument
33426 must be a constant.
33428 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33430 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33431 Use shift_const cost parameter when calculating gain of STV shifts.
33433 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33435 PR rtl-optimization/70478
33436 * lra-constraints.c (process_alt_operands): Check memory for
33437 disfavoring memory insn operand.
33439 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33441 PR middle-end/80100
33442 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33443 left shift in unsigned HOST_WIDE_INT type.
33445 PR rtl-optimization/80385
33446 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33447 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33450 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33451 if they have any depend clauses.
33453 2017-04-11 Martin Liska <mliska@suse.cz>
33456 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33457 * ipa-split.c (split_function): Create a local comdat symbol
33458 if caller is in a comdat group.
33460 2017-04-11 Martin Liska <mliska@suse.cz>
33463 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33466 2017-04-11 Martin Sebor <msebor@redhat.com>
33468 PR middle-end/80364
33469 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33470 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33472 (directive::set_width, directive::set_precision, format_character):
33474 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33477 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33480 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33481 conflict, set target->arch_name instead of target->cpu_name.
33483 2017-04-11 Richard Biener <rguenther@suse.de>
33485 PR tree-optimization/80374
33486 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33487 build_zero_cst, remove fold_convertible_p check again.
33489 2017-04-11 Martin Liska <mliska@suse.cz>
33492 * ubsan.c (instrument_object_size): Do not instrument register
33495 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33498 * config/i386/i386-builtin-types.def
33499 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33500 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33501 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33502 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33503 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33504 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33505 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33506 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33507 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33508 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33509 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33510 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33511 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33512 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33513 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33514 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33515 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33516 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33517 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33518 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33519 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33520 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33521 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33522 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33523 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33524 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33525 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33526 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33527 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33528 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33529 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33530 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33531 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33532 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33533 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33534 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33535 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33536 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33537 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33538 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33539 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33540 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33541 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33542 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33543 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33544 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33546 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33547 flag to second_arg_count, handle 4 argument function type _COUNT
33548 aliases, handle second_arg_count on second argument rather than last.
33550 2017-04-10 Jeff Law <law@redhat.com>
33552 PR tree-optimization/80374
33553 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33554 record anything if we can not convert integer_zero_node to the
33557 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33560 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33561 Enhance special handling given to the TARGET_P9_MINMAX option in
33562 relation to certain other options.
33564 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33566 PR tree-optimization/80153
33567 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33568 remove POINTER_PLUS_EXPR's base part directly, rather than through
33571 2017-04-10 Richard Biener <rguenther@suse.de>
33572 Bin Cheng <bin.cheng@arm.com>
33574 PR tree-optimization/80153
33575 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33576 the first element of pointer type aff_tree. Build result expr in
33578 (add_elt_to_tree): Convert to type unconditionally. Remove other
33579 fold_convert calls.
33580 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33581 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33583 2017-04-10 Richard Biener <rguenther@suse.de>
33585 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33588 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33590 PR rtl-optimization/70478
33591 * lra-constraints.c (curr_small_class_check): New.
33592 (update_and_check_small_class_inputs): New.
33593 (process_alt_operands): Update curr_small_class_check. Disfavor
33594 alternative insn memory operands. Check available regs for small
33597 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33600 * config/mips/mips.opt (-mvirt): Update description.
33601 * doc/invoke.texi (-mvirt): Likewise.
33603 2017-04-10 Richard Biener <rguenther@suse.de>
33605 PR middle-end/80362
33606 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33607 looking for NEGATE_EXPR in -A / -B to A / B folding.
33609 2017-04-10 Martin Liska <mliska@suse.cz>
33611 PR gcov-profile/80224
33612 * gcov.c (print_usage): Fix usage string.
33613 (get_gcov_intermediate_filename): Remove.
33614 (output_gcov_file): Use both for normal and intermediate format.
33615 (generate_results): Do not initialize special file for
33616 intermediate format.
33618 2017-04-10 Richard Biener <rguenther@suse.de>
33620 PR tree-optimization/80304
33621 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33624 2017-04-10 Nathan Sidwell <nathan@acm.org>
33627 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33628 (rs6000_init_builtins): Use it.
33630 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33632 * config/arm/arm.md (<mrc>): Add mode to SET source.
33633 (<mrrc>): Likewise.
33635 2017-04-10 Richard Biener <rguenther@suse.de>
33637 PR middle-end/80344
33638 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33640 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33643 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33644 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33645 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33646 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33647 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33648 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33649 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33650 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33651 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33652 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33653 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33654 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33655 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33656 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33657 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33658 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33659 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33660 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33661 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33662 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33663 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33664 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33665 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33667 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33669 PR rtl-optimization/70478
33670 * lra-constraints.c: Reverse the last patch.
33672 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33674 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33675 Add comment for WCHAR_T.
33677 2017-04-08 Martin Liska <mliska@suse.cz>
33680 2017-04-07 Martin Liska <mliska@suse.cz>
33683 * ipa-split.c (split_function): Add function part to a same comdat
33686 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33689 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33691 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33693 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33694 (vec_pairing): ...to this.
33695 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33696 (rs6000_sched_init): Adjust for name change.
33697 (struct rs6000_sched_context): Likewise.
33698 (rs6000_init_sched_context): Likewise.
33699 (rs6000_set_sched_context): Likewise.
33701 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33707 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33709 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33710 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33711 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33713 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33715 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33717 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33719 PR rtl-optimization/70703
33720 * ira-color.c (update_conflict_hard_regno_costs): Use
33721 int64_t instead of HOST_WIDE_INT.
33723 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33725 PR rtl-optimization/70478
33726 * lra-constraints.c (process_alt_operands): Disfavor alternative
33727 insn memory operands.
33729 2017-04-07 Jeff Law <law@redhat.com>
33731 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33732 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33734 2017-04-07 Martin Liska <mliska@suse.cz>
33737 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33738 Show error message instead of an ICE.
33740 2017-04-07 Martin Liska <mliska@suse.cz>
33743 * ipa-split.c (split_function): Add function part to a same comdat
33746 2017-04-07 Richard Biener <rguenther@suse.de>
33748 PR middle-end/80341
33749 * tree.c (get_unwidened): Also handle ! for_type case for
33751 * convert.c (do_narrow): Split out from ...
33752 (convert_to_integer_1): ... here. Do not pass final truncation
33753 type to get_unwidened for TRUNC_DIV_EXPR.
33755 2017-04-07 Richard Biener <rguenther@suse.de>
33757 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33759 (aff_combination_const): Adjust.
33760 (aff_combination_scale): Likewise.
33761 (aff_combination_add_elt): Likewise.
33762 (aff_combination_add_cst): Likewise.
33763 (aff_combination_convert): Likewise.
33764 (add_elt_to_tree): Likewise. Remove unused argument.
33765 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33767 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33769 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33771 * config/arm/arm.c (arm_default_short_enums): Use
33772 ARM_DEFAULT_SHORT_ENUMS.
33773 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33775 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33778 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33779 members with redundant out-of-class redeclaration.
33781 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33784 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33785 * config/i386/i386.md (*zero_extendsidi2):
33786 Add (?*x,*x) and (?*v,*v) alternatives.
33788 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33791 * config/i386/i386.c (ix86_expand_builtin)
33792 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33793 mode from insn data. Convert operands to insn operand mode.
33794 Copy operands that don't satisfy insn predicate to a register.
33796 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33798 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33801 2017-04-06 Richard Biener <rguenther@suse.de>
33803 PR tree-optimization/80334
33804 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33805 preserve alignment of accesses.
33807 2017-04-06 Richard Biener <rguenther@suse.de>
33809 PR tree-optimization/80262
33810 * tree-sra.c (build_ref_for_offset): Preserve address-space
33812 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33813 Drop useless address-space information on MEM_REF offsets.
33815 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33817 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33819 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33821 PR rtl-optimization/70703
33822 * ira-color.c (update_conflict_hard_regno_costs): Use
33823 HOST_WIDE_INT instead of long.
33825 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33828 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33829 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33831 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33832 for x86_64 target. Handle -m3dnowa option.
33834 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33836 PR rtl-optimization/70703
33837 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33838 (update_conflict_hard_regno_costs): Use long instead of unsigned
33839 arithmetic for cost calculation.
33841 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33842 Bernd Edlinger <bernd.edlinger@hotmail.de>
33845 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33848 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33851 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33852 ptr_mode with Pmode throughout.
33853 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33854 into probe_stack_range and use DImode.
33856 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33859 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33860 call_eh_return is true.
33862 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33864 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33865 Initialize last_match_fntype_index.
33867 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33870 * tree-nvr.c: Include internal-fn.h.
33871 (pass_return_slot::execute): Ignore internal calls without
33874 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33875 Richard Biener <rguenther@suse.de>
33878 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33879 captures used multiple times, except for the last use.
33880 * generic-match-head.c: Include gimplify.h.
33882 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33884 PR tree-optimization/79390
33885 * target.h (struct noce_if_info): Declare.
33886 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33887 * target.def (noce_conversion_profitable_p): New target hook.
33888 * ifcvt.h (struct noce_if_info): New type, moved from ...
33889 * ifcvt.c (struct noce_if_info): ... here.
33890 (noce_conversion_profitable_p): Renamed to ...
33891 (default_noce_conversion_profitable_p): ... this. No longer
33893 (noce_try_store_flag_constants, noce_try_addcc,
33894 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33895 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33896 instead of noce_conversion_profitable_p.
33897 * config/i386/i386.c: Include ifcvt.h.
33898 (ix86_option_override_internal): Don't override
33899 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33900 (ix86_noce_conversion_profitable_p): New function.
33901 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33902 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33903 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33904 * doc/tm.texi: Regenerated.
33906 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33908 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33911 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33914 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33915 instructions for small multiply cores.
33917 2017-04-04 Jeff Law <law@redhat.com>
33919 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33921 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33922 that are not set by the loop over the elements.
33924 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33927 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33928 int mode, convert_modes it to mode as unsigned, otherwise use
33929 lowpart_subreg to mode rather than SImode.
33930 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33931 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33932 Use DImode instead of SImode for the shift count operand.
33933 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33936 2017-04-04 Richard Biener <rguenther@suse.de>
33938 PR middle-end/80281
33939 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33940 arithmetic done for the negate or the plus. Simplify.
33941 (A - (-B) -> A + B): Likewise.
33942 * fold-const.c (split_tree): Make sure to not negate pointers.
33944 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33946 PR rtl-optimization/60818
33947 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33948 a compare of comparisons with the thing compared if this results
33949 in a different machine mode.
33951 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33953 * alias.c (base_alias_check): Fix typo in comment.
33954 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33955 * cgraphunit.c (symbol_table::compile): Likewise.
33956 * collect2.c (maybe_run_lto_and_relink): Likewise.
33957 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33958 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33959 * config/avr/avr.c (avr_map_op_t): Likewise.
33960 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33961 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33962 * config/epiphany/epiphany.md (movcc): Likewise.
33963 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33964 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33966 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33967 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33968 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33969 * config/sh/sh.c (sh_rtx_costs): Likewise.
33970 * fold-const.c (fold_truth_andor): Likewise.
33971 * genautomata.c (collapse_flag): Likewise.
33972 * gengtype.h (struct type::u::s): Likewise.
33973 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33974 * input.c (FORMAT_AMOUNT): Likewise.
33975 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33976 (known_aggs_to_agg_replacement_list): Likewise.
33977 * ipa-inline-analysis.c: Likewise.
33978 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33979 * ipa-polymorphic-call.c
33980 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33981 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33982 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33984 * modulo-sched.c (apply_reg_moves): Likewise.
33985 * omp-expand.c (build_omp_regions_1): Likewise.
33986 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33987 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33988 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33989 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33990 * value-prof.c: Likewise.
33991 * var-tracking.c (val_reset): Likewise.
33993 2017-04-03 Richard Biener <rguenther@suse.de>
33995 PR tree-optimization/80275
33996 * fold-const.c (split_address_to_core_and_offset): Handle
33999 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34001 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34002 descriptors is at least equal to that of functions.
34004 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34006 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34008 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34011 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34012 (mov<IMOD4:mode>): New expander.
34013 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34015 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34017 PR rtl-optimization/79405
34018 * fwprop.c (propagations_left): New variable.
34019 (forward_propagate_into): Decrement it.
34020 (fwprop_init): Initialize it.
34021 (fw_prop): If the variable has reached zero, stop propagating.
34022 (fwprop_addr): Ditto.
34024 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34027 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34028 a FUNCTION_DECL, pass it as decl instead of origin to
34031 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34033 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34036 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34039 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34040 TARGET_VSX_SMALL_INTEGER.
34042 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34044 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34045 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34047 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34049 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34050 extraction from odd-numbered MSA register.
34052 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34054 PR middle-end/80173
34055 * expmed.c (store_bit_field_1): Don't attempt to create
34056 a word subreg out of hard registers wider than word if they
34057 have HARD_REGNO_NREGS of 1 for their mode.
34059 PR middle-end/80163
34060 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34061 conversions to integer types wider than word and pointer.
34064 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34065 (rtx_equal_for_cselib_p): Pass 0 to it.
34066 * cselib.c (cselib_hasher::equal): Likewise.
34067 (rtx_equal_for_cselib_1): Add depth argument. If depth
34068 is 128, don't look up VALUE locs and punt. Increment
34069 depth in recursive calls when walking VALUE locs.
34071 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34073 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34074 (make_gcov_file_name): Use the canonical path name for generating
34076 (read_line): Fix handling of files with ascii null bytes.
34078 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34080 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34081 to initialise a vector register instead
34082 of using a const_int.
34084 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34086 PR translation/80189
34087 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34088 diagnostic messages.
34090 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34093 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34094 (dfp_diex_<mode>): Update mode of operand 1.
34095 * doc/extend.texi (dxex, dxexq): Document change to return type.
34096 (diex, diexq): Document change to argument type.
34098 2017-03-30 Martin Jambor <mjambor@suse.cz>
34101 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34102 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34103 it reflects the signature changes performed at the callee side.
34104 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34105 to cgraph_build_function_type_skip_args.
34106 (build_function_decl_skip_args): Adjust call to the above function.
34108 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34111 * config/i386/sse.md
34112 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34113 register as dest whenever it is a MEM not rtx_equal_p to the
34114 corresponding dup operand, and when forcing into reg move the
34115 reg into the memory afterwards.
34116 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34117 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34118 for the force_reg mode.
34119 (avx512vl_vextractf128<mode>): Use register as dest either
34120 always when a MEM, or when it is a MEM not rtx_equal_p to the
34121 corresponding dup operand, or even not when it is a CONST_VECTOR
34122 depending on the mode and lo vs. hi.
34123 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34125 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34126 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34127 Likewise. Require that operands[2] is even.
34128 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34129 Remove extraneous parens. Require that operands[2] is a multiple
34131 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34132 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34133 disallow memory then.
34135 2017-03-30 Richard Biener <rguenther@suse.de>
34137 PR tree-optimization/77498
34138 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34139 to non-constants over backedges.
34141 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34143 PR rtl-optimization/80233
34144 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34145 as last_combined_insn. Do not test for BARRIER_P separately.
34147 2017-03-29 Andreas Schwab <schwab@suse.de>
34150 * calls.c (prepare_call_address): Convert funexp to Pmode before
34151 copying to temp reg.
34153 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34155 PR tree-optimization/80158
34156 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34157 Handle possible future case of more than one alternate
34159 (replace_rhs_if_not_dup): Likewise.
34160 (replace_one_candidate): Likewise.
34162 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34164 PR rtl-optimization/80193
34165 * ira.c (ira): Do not check allocation for LRA.
34167 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34169 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34170 (nvptx_output_simt_exit): Declare.
34171 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34172 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34173 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34174 (nvptx_declare_function_name): ...here. Emit declaration of local
34175 memory space buffer for omp_simt_enter insn.
34176 (nvptx_output_unisimt_switch): New.
34177 (nvptx_output_softstack_switch): New.
34178 (nvptx_output_simt_enter): New.
34179 (nvptx_output_simt_exit): New.
34180 * config/nvptx/nvptx.h (struct machine_function): New fields
34181 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34182 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34183 (UNSPECV_SIMT_EXIT): Ditto.
34184 (omp_simt_enter_insn): New insn.
34185 (omp_simt_enter): New expansion.
34186 (omp_simt_exit): New insn.
34187 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34189 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34190 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34191 (expand_GOMP_SIMT_EXIT): New.
34192 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34193 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34194 (GOMP_SIMT_EXIT): Ditto.
34195 * target-insns.def (omp_simt_enter): New insn.
34196 (omp_simt_exit): Ditto.
34197 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34199 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34200 (lower_rec_input_clauses): Likewise.
34201 (lower_lastprivate_clauses): Handle SIMT privatization.
34203 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34204 (ompdevlow_adjust_simt_enter): New.
34205 (find_simtpriv_var_op): New.
34206 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34207 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34209 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34210 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34211 (copy_decl_for_dup_finish): Ditto.
34213 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34215 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34218 * config/i386/i386.c (ix86_option_override_internal): Always
34219 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34221 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34223 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34225 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34227 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34228 mark new edge's irreducible flag accordign to it.
34229 (vect_do_peeling): Check loop preheader edge's irreducible flag
34230 and pass it to function slpeel_add_loop_guard.
34232 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34234 PR tree-optimization/80218
34235 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34236 Update block frequencies and counts.
34238 2017-03-28 Richard Biener <rguenther@suse.de>
34240 PR tree-optimization/78644
34241 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34242 of a simplification result we may not use it at all.
34244 2017-03-28 Richard Biener <rguenther@suse.de>
34247 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34248 without arguments, generate default definition of a SSA name.
34250 2017-03-28 Richard Biener <rguenther@suse.de>
34252 PR middle-end/80222
34253 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34254 TYPE_REF_CAN_ALIAS_ALL references.
34255 * fold-const.c (fold_indirect_ref_1): Likewise.
34257 2017-03-28 Martin Liska <mliska@suse.cz>
34260 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34261 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34263 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34264 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34266 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34267 (EXTRA_SPECS): Define.
34268 (SUBTARGET_EXTRA_SPECS): Likewise.
34269 (SUBTARGET_CPP_SPEC): Likewise.
34270 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34271 SUBTARGET_EXTRA_SPECS.
34272 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34274 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34276 * config/arc/simdext.md (vst64_insn): Update pattern.
34277 (vld32wh_insn): Likewise.
34278 (vld32wl_insn): Likewise.
34279 (vld64_insn): Likewise.
34280 (vld32_insn): Likewise.
34282 2017-03-28 Marek Polacek <polacek@redhat.com>
34285 * fold-const.c (fold_comparison): Use protected_set_expr_location
34286 instead of SET_EXPR_LOCATION.
34288 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34290 * tree.c (add_expr): Avoid name lookup warning.
34292 2017-03-27 Jeff Law <law@redhat.com>
34294 PR tree-optimization/80216
34295 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34296 function name. Limit recursion depth.
34297 (record_temporary_equivalences): Corresponding changes.
34299 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34301 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34304 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34307 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34309 * cfgcleanup.c (reg_note_cfa_p): New array.
34310 (insns_have_identical_cfa_notes): New function.
34311 (old_insns_match_p): Don't cross-jump in between /f
34312 and non-/f instructions. If both i1 and i2 are frame related,
34313 verify all CFA notes, their order and content.
34315 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34318 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34319 HImode and SImode with zero extend to DImode to one insn.
34320 (bswap<mode>2_extenddi): Likewise.
34321 (bswapsi2_extenddi): Likewise.
34322 (bswaphi2_extendsi): Likewise.
34323 (bswaphi2): Combine bswap HImode and SImode into one insn.
34324 Separate memory insns from swapping register.
34325 (bswapsi2): Likewise.
34326 (bswap<mode>2): Likewise.
34327 (bswaphi2_internal): Delete, no longer used.
34328 (bswapsi2_internal): Likewise.
34329 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34330 store, and gpr<-gpr swap insns.
34331 (bswap<mode>2_store): Likewise.
34332 (bswaphi2_reg): Register only splitter, combine with the splitter.
34333 (bswaphi2 splitter): Likewise.
34334 (bswapsi2_reg): Likewise.
34335 (bswapsi2 splitter): Likewise.
34336 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34337 the insns into load, store, and register/register insns.
34338 (bswapdi2_ldbrx): Likewise.
34339 (bswapdi2_load): Likewise.
34340 (bswapdi2_store): Likewise.
34341 (bswapdi2_reg): Likewise.
34343 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34345 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34346 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34348 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34351 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34353 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34354 special handling for target option conflicts between dform
34355 options (-mpower9-dform, -mpower9-dform-vector,
34356 -mpower9-dform-scalar) and -mno-direct-move.
34358 2017-03-27 Richard Biener <rguenther@suse.de>
34360 PR tree-optimization/80181
34361 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34363 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34365 * config/arc/predicates.md (move_double_src_operand): Replace the
34366 call to move_double_src_operand with a call to address_operand.
34368 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34370 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34371 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34372 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34374 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34376 * config/arc/predicates.md (long_immediate_loadstore_operand):
34377 Consider scaled addresses cases.
34379 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34381 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34382 restored when in interrupt.
34383 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34384 doesn't have delay slot.
34386 2017-03-27 Richard Biener <rguenther@suse.de>
34389 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34390 inlined thunk clones.
34392 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34395 * asan.c (instrument_derefs): Copy over last operand from
34396 original COMPONENT_REF to the new COMPONENT_REF with
34397 DECL_BIT_FIELD_REPRESENTATIVE.
34398 * ubsan.c (instrument_object_size): Likewise.
34400 2017-03-27 Richard Biener <rguenther@suse.de>
34402 PR tree-optimization/80170
34403 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34404 sure DR/SCEV didnt fold in constants we do not see when looking
34405 at the reference base alignment.
34407 2017-03-27 Richard Biener <rguenther@suse.de>
34409 PR middle-end/80171
34410 * gimple-fold.c (fold_ctor_reference): Properly guard against
34411 NULL return value from canonicalize_constructor_val.
34413 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34416 * config/i386/i386.c (ix86_expand_builtin)
34417 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34418 flags reg setting and flags reg using instructions.
34419 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34420 clobbering instructions to zero extend op2.
34422 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34424 * doc/install.texi (Configuration) <--with-aix-soname>:
34425 Update link to AIX ld.
34427 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34429 PR rtl-optimization/80160
34430 PR rtl-optimization/80159
34431 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34432 reg_alternate_class into account.
34434 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34437 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34438 to consider in curr_insn_transform.
34440 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34442 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34443 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34444 and emit_mode_inner.
34446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34448 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34449 argument to the overloaded builtin variants. Use the new flag to
34450 deprecate certain builtin variants.
34451 * config/s390/s390-builtin-types.def: Add new builtin types.
34452 * config/s390/s390-builtins.h: Support new flags field for
34453 overloaded builtins.
34454 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34455 (s390_macro_to_expand): Enable vector float data type.
34456 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34457 builtins by incrementing the __VEC__ version number.
34458 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34460 (s390_resolve_overloaded_builtin): Emit error messages depending
34461 on the builtin flags.
34462 * config/s390/s390.c (s390_expand_builtin): Support additional
34463 flags argument. Change error message to match the messages
34464 emitted in s390-c.c.
34465 * config/s390/s390.md: New UNSPEC_* constants.
34466 (op_type): Add new instruction types.
34467 * config/s390/vecintrin.h: Add new builtins and test data class
34469 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34470 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34471 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34472 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34473 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34474 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34476 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34477 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34478 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34479 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34481 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34482 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34483 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34484 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34486 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34487 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34488 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34489 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34490 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34491 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34492 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34494 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34495 ("vec_scatter_element<V_HW_4:mode>_DI")
34496 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34497 ("vec_fpint<mode>", "vflls")
34498 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34499 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34500 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34501 ("*vec_cmphe<mode>_cc"): ... these.
34503 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34504 mode constant instead of magic value.
34506 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34508 * config/s390/s390.c (s390_expand_vec_compare): Support other
34509 vector floating point modes than just V2DF.
34510 (s390_expand_vcond): Likewise.
34511 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34512 (s390_cannot_change_mode_class): Prevent mode changes between TF
34513 and V1TF in vector registers.
34514 * config/s390/s390.md (DF, SF): New mode attributes.
34515 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34516 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34517 SFmode support for VRs.
34518 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34520 (VFT, VF_HW): New mode iterators.
34521 (vw, sdx): New mode attributes.
34522 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34523 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34524 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34525 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34526 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34527 also the new vector floating point modes. Renaming to ...
34529 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34530 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34531 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34532 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34533 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34534 ("vec_unordered<mode>"): ... these.
34536 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34537 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34538 ("*vec_extendv2df"): New insn definitions.
34540 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34542 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34543 ("mulditi3_2", "*muldi3_sign"): New patterns.
34544 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34545 rename the pattern definition.
34547 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34549 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34551 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34553 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34555 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34556 instruction if possible.
34557 * config/s390/vector.md (vec_halfnumelts): New mode
34559 ("*vec_vllezlf<mode>"): New pattern.
34561 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34563 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34564 ("popcountv4si2", "popcountv2di2"): Rename to ...
34565 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34566 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34568 ("popcount<mode>2_vxe"): New pattern.
34570 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34572 * common/config/s390/s390-common.c (processor_flags_table): Add
34574 * config.gcc: Add arch12.
34575 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34576 Default to arch12 for unknown CPU model numbers.
34577 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34578 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34579 PROCESSOR_max sanity check.
34580 * config/s390/s390-opts.h (enum processor_type): Add
34582 * config/s390/s390.c (processor_table): Add arch12.
34583 (s390_expand_builtin): Add check for B_VXE flag.
34584 (s390_issue_rate): Add PROCESSOR_ARCH12.
34585 (s390_get_sched_attrmask): Likewise.
34586 (s390_get_unit_mask): Likewise.
34587 (s390_sched_score): Enable z13 scheduling for arch12.
34588 (s390_sched_reorder): Likewise.
34589 (s390_sched_variable_issue): Likewise.
34590 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34592 (s390_tune_attr): Use z13 scheduling also for arch12.
34593 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34594 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34595 (TARGET_VXE_P): New macros.
34596 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34597 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34598 * config/s390/s390.opt: Add arch12 as processor_type.
34600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34602 * config/s390/s390.md
34603 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34604 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34605 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34607 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34608 Rename expanders to ...
34610 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34611 ("fixuns_truncdddi2_emu"): ... these.
34613 ("fixuns_trunc<mode>si2_emu"): New expander.
34615 ("*fixuns_truncdfdi2_z13"): Rename to ...
34616 ("*fixuns_truncdfdi2_vx"): ... this.
34618 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34620 * config/s390/2964.md: Remove the single element vector compare
34621 instructions which are no longer used.
34622 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34624 (s390_canonicalize_comparison): Remove handling of DFmode
34626 (s390_expand_vec_compare_scalar): Remove function.
34627 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34628 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34630 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34632 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34634 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34636 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34637 will anyway by matched by mov<mode>_64dfp.
34639 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34641 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34642 vlef/vstef. Add missing operand to vleif.
34644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34646 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34647 pair for all vector types with 64 bit elements.
34648 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34649 * config/s390/vector.md (V_HW_64): ... here.
34650 (V_128_NOSINGLE): New mode iterator.
34651 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34652 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34653 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34654 ("*vec_load_pairv2di"): Change to ...
34655 ("*vec_load_pair<mode>"): ... this one.
34657 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34659 * config/s390/constraints.md: Add comments.
34660 (jKK): Reject element sizes > 8 bytes.
34661 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34663 * config/s390/s390.md: Add the s_operand checks formerly in
34664 s390_split_ok_p to various splitters where they are still
34666 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34667 for 128 bit vectors. Plus two splitters.
34669 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34671 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34677 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34678 error if the boundary argument is not constant.
34680 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34682 PR rtl-optimization/80112
34683 * loop-doloop.c (doloop_condition_get): Don't check condition
34684 if cmp isn't SET with IF_THEN_ELSE src.
34686 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34688 PR tree-optimization/80158
34689 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34690 replacing a candidate statement, also replace it for the
34691 candidate's alternate interpretation.
34692 (replace_rhs_if_not_dup): Likewise.
34693 (replace_one_candidate): Likewise.
34695 2017-03-24 Richard Biener <rguenther@suse.de>
34697 PR tree-optimization/80167
34698 * graphite-isl-ast-to-gimple.c
34699 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34701 (translate_isl_ast_to_gimple::get_rename): Likewise.
34703 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34706 handling of certain combinations of target options, including the
34707 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34708 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34710 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34713 * config/arm/arm.md (*load_multiple): Add reload_completed to
34714 matching condition.
34716 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34717 Richard Biener <rguenth@suse.de>
34719 PR tree-optimization/79908
34720 PR tree-optimization/80136
34721 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34722 been cast away, gimplify_and_add suffices.
34724 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34726 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34728 2017-03-23 Richard Biener <rguenther@suse.de>
34730 PR tree-optimization/80032
34731 * gimplify.c (gimple_push_cleanup): Forced unconditional
34732 cleanups still have to go to the conditional_cleanups
34735 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34737 PR tree-optimization/80072
34738 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34740 (next_operand_entry_id): Change type to unsigned int.
34741 (sort_by_operand_rank): Make sure to return the right return value
34742 even if unsigned fields are bigger than INT_MAX.
34743 (struct oecount): Change cnt and id type to unsigned int.
34744 (oecount_hasher::equal): Formatting fix.
34745 (oecount_cmp): Make sure to return the right return value
34746 even if unsigned fields are bigger than INT_MAX.
34747 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34750 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34751 TREE_READONLY on result if writing it more than once.
34754 * doc/invoke.texi (-fsanitize=thread): Document that with
34755 -fnon-call-exceptions atomics are not able to throw
34759 * tsan.c: Include tree-eh.h.
34760 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34761 maybe_clean_or_replace_eh_stmt where needed.
34762 (instrument_memory_accesses): Add cfg_changed argument.
34763 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34764 if it returned true.
34765 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34767 PR rtl-optimization/63191
34768 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34769 wrapper function, moved the whole old content into ...
34770 (ix86_delegitimize_address_1): ... this. New inline function.
34771 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34772 true as last argument instead of ix86_delegitimize_address.
34774 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34776 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34777 cortexa57_branch_cost.
34779 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34781 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34783 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34786 * doc/md.texi (Constraints): Document wA constraint.
34787 * config/rs6000/constraints.md (wA): New.
34788 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34789 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34790 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34791 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34793 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34796 * gimplify.c (is_oacc_declared): New function.
34797 (oacc_default_clause): Use it to set default flags for acc declared
34798 variables inside parallel regions.
34799 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34800 declared variables.
34801 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34802 declare attribute to any decl as necessary.
34804 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34807 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34808 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34809 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34810 (arm_arch_lpae): This.
34811 * config/arm/arm.c (arm_arch7ve): Rename into ...
34812 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34813 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34816 2017-03-22 Martin Liska <mliska@suse.cz>
34819 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34820 error message instead of an ICE.
34822 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34824 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34826 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34828 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34833 2017-03-21 Martin Sebor <msebor@redhat.com>
34835 * doc/extend.texi: Use "cannot" instead of "can't."
34836 * doc/hostconfig.texi: Same.
34837 * doc/install.texi: Same.
34838 * doc/invoke.texi: Same.
34839 * doc/loop.texi: Same.
34840 * doc/md.texi: Same.
34841 * doc/objc.texi: Same.
34842 * doc/rtl.texi: Same.
34843 * doc/tm.texi: Same.
34844 * doc/tm.texi.in: Same.
34845 * doc/trouble.texi: Same.
34847 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34850 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34851 (collect_checksum_attributes): Set it.
34852 (die_checksum_ordered): Use it.
34854 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34856 PR tree-optimization/79908
34857 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34858 change: For a VA_ARG whose LHS has been cast away, use
34859 force_gimple_operand to construct the side effects.
34861 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34863 PR translation/80001
34864 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34865 more amenable to translation.
34866 (oacc_loop_auto_partitions): Likewise.
34868 2017-03-21 Marek Polacek <polacek@redhat.com>
34869 Martin Sebor <msebor@redhat.com>
34871 PR tree-optimization/80109
34872 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34873 on INTEGRAL_TYPE_P.
34875 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34876 Segher Boessenkool <segher@kernel.crashing.org>
34879 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34880 check reg_used_between_p between insn and one of succ or succ2
34881 depending on if succ is artificial insn not inserted into insn
34884 2017-03-21 Martin Liska <mliska@suse.cz>
34886 PR gcov-profile/80081
34887 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34888 * doc/gcc.texi: Include gcov-dump stuff.
34889 * doc/gcov-dump.texi: New file.
34891 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34893 PR rtl-optimization/79150
34894 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34895 conditional jump, if the jump is the last insn of the loop.
34897 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34898 Richard Biener <rguenth@suse.de>
34900 PR tree-optimization/79908
34901 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34902 been cast away, use force_gimple_operand to construct the side
34905 2017-03-21 Martin Liska <mliska@suse.cz>
34907 PR libfortran/79956
34908 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34911 2017-03-21 Brad Spengler <spender@grsecurity.net>
34914 * plugin.c (htab_hash_plugin): New function.
34915 (add_new_plugin): Use it and adjust.
34916 (parse_plugin_arg_opt): Adjust.
34917 (init_one_plugin): Likewise.
34919 2017-03-21 Richard Biener <rguenther@suse.de>
34921 PR tree-optimization/80032
34922 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34923 if set force the cleanup to happen unconditionally.
34924 (gimplify_target_expr): Push inserted clobbers with force_uncond
34925 to avoid them being removed by control-dependent DCE.
34927 2017-03-21 Richard Biener <rguenther@suse.de>
34929 PR tree-optimization/80122
34930 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34931 va_arg_pack_len when the inlined call stmt requires pack
34933 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34935 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34938 * tsan.c (instrument_builtin_call): If the memory model argument
34939 is not a constant, assume it is valid.
34942 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34945 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34947 PR rtl-optimization/79910
34948 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34949 if its dest is used by an insn before I2 (other than the combined
34950 insns themselves, which are properly handled already).
34952 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34955 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34957 * combine.c (record_used_regs): New static function.
34958 (try_combine): Handle situations where there is an additional
34959 instruction between I2 and I3 which needs to have a LOG_LINK
34963 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34965 * combine.c (try_combine): Delete redundant i1 test. Call
34966 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34968 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34971 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34972 alternatives 13/14.
34974 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34976 PR tree-optimization/80054
34977 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34978 the optimization if a PHI or any of its arguments is not dominated
34979 by the candidate's basis. Use gphi* rather than gimple* as
34981 (replace_profitable_candidates): Clean up a gimple* variable that
34982 should be a gphi* variable.
34984 2017-03-20 Martin Sebor <msebor@redhat.com>
34987 * doc/extend.texi (attribute constructor): Document present limitation.
34989 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34992 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34993 __POWER9_VECTOR__ #ifdef control, change template definition to
34994 use Power9-specific built-in function.
34995 (vec_any_eq): Likewise.
34996 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34997 to control outcomes from this test.
34998 (vector_ae_<mode>p): For VEC_F modes, likewise.
35000 2017-03-20 Ian Lance Taylor <iant@google.com>
35002 * config/i386/i386.c (ix86_function_regparm): Save an extra
35003 register for -fsplit-stack with DECL_STATIC_CHAIN.
35005 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35008 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35009 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35011 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35013 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35015 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35018 2017-03-20 Marek Polacek <polacek@redhat.com>
35021 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35023 2017-03-20 Richard Biener <rguenther@suse.de>
35025 PR tree-optimization/80113
35026 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35027 allocate extra SSA name for PHI def.
35028 (add_close_phis_to_outer_loops): Likewise.
35029 (add_close_phis_to_merge_points): Likewise.
35030 (copy_loop_close_phi_args): Likewise.
35031 (copy_cond_phi_nodes): Likewise.
35033 2017-03-20 Martin Liska <mliska@suse.cz>
35035 PR middle-end/79753
35036 * tree-chkp.c (chkp_build_returned_bound): Do not build
35037 returned bounds for a LHS that's not a BOUNDED_P type.
35039 2017-03-20 Martin Liska <mliska@suse.cz>
35043 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35044 COMPLEX_CST and VECTOR_CST.
35046 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35049 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35050 target operand. A new splitter adds the clobber statement in case
35051 the target operand is dead anyway.
35053 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35055 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35056 to age-old versions of binutils and glibc.
35058 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35060 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35062 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35064 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35066 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35068 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35069 requirement for binutils 2.13.
35071 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35073 * combine.c (try_combine): Delete redundant i1 test. Call
35074 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35076 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35078 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35079 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35081 <riscv64-*-elf>: Re-arrange section
35082 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35083 <riscv32-*-linux>: Likewise.
35084 <riscv64-*-elf>: Likewise
35085 <riscv64-*-linux>: Likewise.
35087 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35090 * aarch64.opt(verbose-cost-dump): Fix typo.
35092 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35095 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35096 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35098 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35100 * reload.c (find_reloads): When reloading a nonoffsettable address,
35101 use RELOAD_OTHER for it and its address reloads.
35103 PR rtl-optimization/79910
35104 * combine.c (record_used_regs): New static function.
35105 (try_combine): Handle situations where there is an additional
35106 instruction between I2 and I3 which needs to have a LOG_LINK
35109 2017-03-17 Jeff Law <law@redhat.com>
35111 PR tree-optimization/71437
35112 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35113 conditional in the hash table first.
35114 (vrp_dom_walker::before_dom_children): Extract condition from
35115 ASSERT_EXPR. Record condition, its inverion and any implied
35116 conditions as well.
35118 2017-03-17 Marek Polacek <polacek@redhat.com>
35119 Markus Trippelsdorf <markus@trippelsdorf.de>
35121 PR tree-optimization/80079
35122 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35125 2017-03-17 Richard Biener <rguenther@suse.de>
35127 PR middle-end/80075
35128 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35129 Properly verify the LHS before the RHS possibly claims to be
35131 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35134 2017-03-17 Martin Jambor <mjambor@suse.cz>
35136 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35137 (List of -O2 options): Likewise.
35138 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35141 2017-03-17 Tom de Vries <tom@codesourcery.com>
35143 * gcov-dump.c (print_usage): Print bug_report_url.
35145 2017-03-17 Richard Biener <rguenther@suse.de>
35147 PR middle-end/80050
35148 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35149 (parser::peek): Likewise.
35151 2017-03-17 Richard Biener <rguenther@suse.de>
35153 PR tree-optimization/80048
35154 * sese.c (free_sese_info): Properly release rename_map and
35155 copied_bb_map elements.
35157 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35159 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35160 Add linked-list forward and backlinks. Insert on
35161 construction, remove on destruction.
35162 (class pass_store_merging): Add m_stores_head field.
35163 (pass_store_merging::terminate_and_process_all_chains):
35164 Iterate over m_stores_head list.
35165 (pass_store_merging::terminate_all_aliasing_chains):
35167 (pass_store_merging::execute): Check for debug stmts first.
35168 Push new chains onto the m_stores_head stack.
35170 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35173 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35174 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35175 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35177 2017-03-16 Jeff Law <law@redhat.com>
35179 PR tree-optimization/71437
35180 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35181 member function. Implementation moved into after_dom_children
35182 member function and into the threader's thread_outgoing_edges
35184 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35185 some code into new thread_outgoing_edges.
35186 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35187 definition. Simplify marker handling (do it here). Assume we always
35188 have the available expression and the const/copies tables.
35189 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35191 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35192 * tree-vrp.c (equiv_stack): No longer file scoped.
35193 (vrp_dom_walker): New class.
35194 (vrp_dom_walker::before_dom_children): New member function.
35195 (vrp_dom_walker::after_dom_children): Likewise.
35196 (identify_jump_threads): Setup domwalker. Use it rather than
35197 walking edges in a random order by hand. Simplify setup/finalization.
35198 (finalize_jump_threads): Remove.
35199 (vrp_finalize): Do not call identify_jump_threads here.
35200 (execute_vrp): Do it here instead and call thread_through_all_blocks
35203 PR tree-optimization/71437
35204 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35206 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35208 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35209 (dom_opt_dom_walker::thread_across_edge): Remove
35210 handle_dominating_asserts argument. All callers changed.
35211 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35212 changes. Remove calls to lhs_of_dominating_assert. Other
35213 uses of handle_dominating_asserts turn into unconditional code
35214 (simplify_control_stmt_condition_1): Likewise.
35215 (simplify_control_stmt_condition): Likewise.
35216 (thread_through_normal_block, thread_across_edge): Likewise.
35217 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35218 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35219 object if it is not an SSA_NAME.
35220 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35221 before calling into the VRP specific simplifiers.
35222 (identify_jump_threads): Remove handle_dominating_asserts
35225 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35228 * tree-diagnostic.c (default_tree_printer): No longer static.
35229 * tree-diagnostic.h (default_tree_printer): New prototype.
35231 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35233 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35234 Change ins into fmov.
35236 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35238 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35239 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35240 Use h_con constraint for operand 1.
35241 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35242 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35244 2017-03-15 Jeff Law <law@redhat.com>
35246 PR tree-optimization/71437
35247 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35248 (record_temporary_equivalences): Use it.
35250 PR tree-optimization/71437
35251 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35252 tree-ssa-scopedtables.
35253 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35254 (record_conditions, record_cond, vuse_eq): Likewise.
35255 (record_edge_info): Adjust to API tweak of record_conditions.
35256 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35257 (record_temporary_equivalences, optimize_stmt): Likewise.
35258 (eliminate_redundant_computations): Likewise.
35259 (record_equivalences_from_stmt): Likewise.
35260 * tree-ssa-scopedtables.c: Include options.h and params.h.
35261 (vuse_eq): New function, moved from tree-ssa-dom.c
35262 (build_and_record_new_cond): Likewise.
35263 (record_conditions): Likewise. Accept vector of conditions rather
35264 than edge_equivalence structure for first argument.
35265 for the first argument.
35266 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35267 from tree-ssa-dom.c.
35268 (avail_exprs_stack::record_cond): Likewise.
35269 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35270 from tree-ssa-dom.c.
35271 (avail_exprs_stack): Add new member functions lookup_avail_expr
35273 (record_conditions): Declare.
35275 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35278 * lra-constraints.c (process_alt_operands): Increase reject for
35279 reloading an input/output operand.
35281 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35284 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35285 insns to convert from signed/unsigned char/short to IEEE 128-bit
35287 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35289 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35292 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35293 subreg of inner mode for values already in registers.
35295 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35297 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35298 iteration reg is used after the loop.
35300 2017-03-14 Martin Sebor <msebor@redhat.com>
35302 PR tree-optimization/79800
35303 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35304 precision in negative-positive range.
35305 (format_floating): Call non-const overload with adjusted precision.
35307 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35310 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35313 2017-03-14 Martin Sebor <msebor@redhat.com>
35315 PR middle-end/80020
35316 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35317 * builtins.def (aligned_alloc): Use it.
35320 * Makefile.in (GTFILES): Add calls.c.
35321 * calls.c: Include "gt-calls.h".
35323 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35325 PR rtl-optimization/79728
35326 * regs.h (struct target_regs): New field
35327 x_contains_allocatable_regs_of_mode.
35328 (contains_allocatable_regs_of_mode): New macro.
35329 * reginfo.c (init_reg_sets_1): Initialize it, and change
35330 contains_reg_of_mode so it includes global regs as well.
35331 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35332 rather than contains_regs_of_mode.
35334 2017-03-14 Martin Liska <mliska@suse.cz>
35336 * doc/invoke.texi: Document options that can't be combined with
35337 -fcheck-pointer-bounds.
35339 2017-03-14 Martin Liska <mliska@suse.cz>
35341 PR middle-end/79831
35342 * doc/invoke.texi (-Wchkp): Document the option.
35344 2017-03-14 Martin Liska <mliska@suse.cz>
35346 * Makefile.in: Install gcov-dump.
35348 2017-03-14 Martin Liska <mliska@suse.cz>
35350 * multiple_target.c (expand_target_clones): Bail out for
35351 an invalid attribute.
35353 2017-03-14 Richard Biener <rguenther@suse.de>
35355 * alias.c (struct alias_set_entry): Pack properly.
35356 * cfgloop.h (struct loop): Likewise.
35357 * cse.c (struct set): Likewise.
35358 * ipa-utils.c (struct searchc_env): Likewise.
35359 * loop-invariant.c (struct invariant): Likewise.
35360 * lra-remat.c (struct cand): Likewise.
35361 * recog.c (struct change_t): Likewise.
35362 * rtl.h (struct address_info): Likewise.
35363 * symbol-summary.h (function_summary): Likewise.
35364 * tree-loop-distribution.c (struct partition): Likewise.
35365 * tree-object-size.c (struct object_size_info): Likewise.
35366 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35367 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35368 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35369 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35370 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35371 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35372 (struct _stmt_vec_info): Likewise.
35374 2017-03-14 Martin Liska <mliska@suse.cz>
35377 * multiple_target.c (create_dispatcher_calls): Check that
35378 a target can create a function dispatcher.
35380 2017-03-14 Martin Liska <mliska@suse.cz>
35383 * multiple_target.c (expand_target_clones): Drop local.local
35384 flag for default implementation.
35386 2017-03-14 Richard Biener <rguenther@suse.de>
35388 PR tree-optimization/80030
35389 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35391 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35393 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35394 gcc_fallthrough() instead of __attribute__((fallthrough));
35396 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35398 * doc/gcc.texi: Remove "up" link to (DIR).
35399 * doc/gccint.texi: Ditto.
35401 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35403 * doc/install.texi (Specific) <avr>: Remove reference to
35406 2017-03-13 Jeff Law <law@redhat.com>
35408 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35409 attribute rather than comments.
35411 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35412 match_scratch operand is highest.
35414 2017-03-13 Martin Liska <mliska@suse.cz>
35416 PR middle-end/78339
35417 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35418 is a CHKP clone, use original declaration.
35420 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35422 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35423 (arc_conditional_register_usage): Use a different allocation order
35424 when optimizing for size.
35425 * common/config/arc/arc-common.c (arc_option_optimization_table):
35426 Section anchors default on when optimizing for size.
35428 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35430 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35432 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35434 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35435 * config/arc/arc.md (cpu_facility): Add cd variant.
35436 (*movqi_insn): Add code density variant.
35437 (*movhi_insn): Likewise.
35438 (*movqi_insn): Likewise.
35439 (*addsi3_mixed): Likewise.
35440 (subsi3_insn): Likewise.
35442 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35444 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35446 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35448 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35449 expressions with MINUS and UNARY ops.
35451 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35454 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35456 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35457 between vec_select and vector argument.
35458 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35459 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35460 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35461 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35462 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35463 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35465 2017-03-13 Richard Biener <rguenther@suse.de>
35468 * params.def (vect-max-peeling-for-alignment): Fix typo.
35470 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35472 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35473 issue that only occurred with binutils below 2.18.
35475 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35477 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35478 refer to binutils 2.11/2.12 minimum.
35480 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35482 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35483 ftp.kernel.org and simplify binutils requirement.
35485 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35487 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35489 (Optimize Options): Ditto. Also remove redundancy.
35491 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35493 PR translation/79848
35494 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35496 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35497 to G_ to avoid double translation.
35499 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35501 PR translation/79923
35502 * auto-profile.c (get_combined_location): Convert leading
35503 character of diagnostics to lower case and remove trailing period.
35504 (read_profile): Likewise for various diagnostics.
35505 * config/arm/arm.c (arm_option_override): Remove trailing period
35506 from various diagnostics.
35507 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35508 (msp430_expand_delay_cycles): Likewise.
35510 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35513 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35514 full command-line argument, rather than just "str".
35515 (aarch64_validate_march): Likewise.
35516 (aarch64_validate_mtune): Likewise.
35518 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35520 PR rtl-optimization/78911
35521 * lra-assigns.c (must_not_spill_p): New function.
35522 (spill_for): Use it.
35524 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35526 PR tree-optimization/79981
35527 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35528 ATOMIC_COMPARE_EXCHANGE ifn result.
35529 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35530 IFN_ATOMIC_COMPARE_EXCHANGE.
35532 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35535 * opts.c (parse_sanitizer_options): Add missing question mark to
35536 "did you mean" message.
35538 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35540 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35542 (VMULEUH_UNS): Likewise.
35543 (VMULOUB_UNS): Likewise.
35544 (VMULOUH_UNS): Likewise.
35545 * config/rs6000/rs6000.c (builtin_function_type): Remove
35546 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35548 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35551 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35552 x with result of extra_parsing_for_operand_code_0.
35553 (function_reader::extra_parsing_for_operand_code_0): Convert
35554 return type from void to rtx, returning x. When reading
35555 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35556 larger size containing struct block_symbol.
35558 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35560 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35561 -mfloat128-hardware without -m64.
35563 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35566 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35567 entries to the case statement that marks unsigned arguments to
35568 overloaded functions.
35570 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35572 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35573 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35575 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35578 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35579 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35581 2017-03-10 Martin Liska <mliska@suse.cz>
35585 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35586 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35589 2017-03-10 Olivier Hainque <hainque@adacore.com>
35591 * tree-switch-conversion (array_value_type): Start by resetting
35592 candidate type to it's main variant.
35594 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35596 PR rtl-optimization/79909
35597 * combine.c (try_combine): Use simplify_replace_rtx on individual
35598 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35599 of the whole CALL_INSN_FUNCTION_USAGE.
35601 PR tree-optimization/79972
35602 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35603 get_range_info on SSA_NAMEs. Formatting fixes.
35605 2017-03-10 Richard Biener <rguenther@suse.de>
35606 Jakub Jelinek <jakub@redhat.com>
35608 PR tree-optimization/77975
35609 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35610 edge to be constant.
35611 (get_val_for): For constant x return it. Formatting fix.
35612 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35613 would use the same bases as the current one.
35615 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35617 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35618 instead of vec_select for V1TImode.
35619 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35621 (VSX_LE_128): Add V1TI to this mode iterator.
35622 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35623 (*vsx_le_perm_store_<mode>): Likewise.
35624 (pre-reload splitter for VSX stores): Likewise.
35625 (post-reload splitter for VSX stores): Likewise.
35626 (*vsx_xxpermdi2_le_<mode>): Likewise.
35627 (*vsx_lxvd2x2_le_<mode>): Likewise.
35628 (*vsx_stxvd2x2_le_<mode>): Likewise.
35630 2017-03-09 Michael Eager <eager@eagercon.com>
35632 Correct failures with --enable-checking=yes,rtl.
35634 * config/microblaze/microblaze.c (microblaze_expand_shift):
35635 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35636 test for const0_rtx.
35637 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35638 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35640 2017-03-09 Richard Biener <rguenther@suse.de>
35642 PR tree-optimization/79977
35643 * graphite-scop-detection.c (scop_detection::merge_sese):
35644 Handle the case of extra exits to blocks dominating the entry.
35646 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35648 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35651 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35653 PR rtl-optimization/79949
35654 * lra-constraints.c (process_alt_operands): Check memory when
35655 trying to predict a cycle. Print about the overall increase.
35657 2017-03-09 Richard Biener <rguenther@suse.de>
35659 PR middle-end/79971
35660 * gimple-expr.c (useless_type_conversion_p): Preserve
35661 TYPE_SATURATING for fixed-point types.
35663 2017-03-09 Richard Biener <rguenther@suse.de>
35666 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35667 alignment of BLKmode params.
35669 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35672 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35673 (VALL_NO_V2Q): Likewise.
35675 * config/aarch64/aarch64-simd.md
35676 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35678 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35679 VALL_NO_V2Q mode iterator.
35680 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35682 2017-03-09 Martin Liska <mliska@suse.cz>
35684 PR tree-optimization/79631
35685 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35686 tree_int_cst_sign_bit just for INTEGER constants.
35688 2017-03-09 Martin Liska <mliska@suse.cz>
35692 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35695 2017-03-09 Marek Polacek <polacek@redhat.com>
35698 * tree.c (inchash::add_expr): Handle TREE_VEC.
35700 2017-03-09 Martin Liska <mliska@suse.cz>
35703 (chkp_narrow_size_and_offset): New function.
35704 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35705 (void chkp_parse_bit_field_ref): New function.
35706 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35707 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35709 2017-03-09 Martin Liska <mliska@suse.cz>
35712 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35713 (chkp_find_bounds_1): Remove gcc_unreachable.
35715 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35718 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35719 BUILT_IN_SYNC*, determine the access type from the size suffix and
35720 always build a MEM_REF with that type. Handle forgotten
35721 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35724 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35725 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35726 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35727 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35728 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35729 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35730 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35731 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35732 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35733 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35734 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35735 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35736 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35737 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35738 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35739 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35740 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35741 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35742 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35743 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35744 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35745 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35746 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35747 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35748 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35749 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35750 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35751 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35752 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35753 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35754 definitions outside of __OPTIMIZE__ guarded section.
35757 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35758 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35759 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35760 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35763 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35765 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35766 ("vfenez<mode>"): Add missing constraints.
35768 2017-03-08 Martin Sebor <msebor@redhat.com>
35771 * config/nds32/nds32.c (nds32_option_override):
35772 Fix misspelled diagnostic.
35774 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35777 * gimplify.c (gimplify_omp_for): Replace index var in outer
35778 taskloop statement with an artificial variable and add
35779 OMP_CLAUSE_PRIVATE clause for it.
35781 2017-03-08 Richard Biener <rguenther@suse.de>
35783 PR tree-optimization/79955
35784 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35785 for accesses that are completely outside of the variable.
35787 2017-03-08 Andrew Haley <aph@redhat.com>
35789 PR tree-optimization/79943
35790 * tree-ssa-loop-split.c (compute_new_first_bound): When
35791 calculating the new upper bound, (END-BEG) should be added, not
35794 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35796 * config/avr/avr.md (setmemhi): Make sure match_dup
35797 operand number comes before match_scratch.
35799 2017-03-08 Richard Biener <rguenther@suse.de>
35801 PR tree-optimization/79920
35802 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35803 with ncopies == 1 to ...
35804 (vect_transform_slp_perm_load): ... here. Properly compute
35805 all element loads by iterating VF times over the group. Do
35806 not handle ncopies (computed in a broken way) in
35807 vect_create_mask_and_perm.
35809 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35812 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35813 is a uniform vector, use uniform_vector_p return value instead of
35814 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35816 2017-03-07 Marek Polacek <polacek@redhat.com>
35818 PR middle-end/79809
35819 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35820 (alloca_call_type): Likewise.
35822 2017-03-07 Martin Liska <mliska@suse.cz>
35824 * gcov.c (process_args): Put comment to correct location.
35826 2017-03-07 Martin Liska <mliska@suse.cz>
35828 PR middle-end/68270
35829 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35830 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35831 (chkp_narrow_bounds_for_field): Likewise.
35832 (chkp_parse_array_and_component_ref): Pass one more argument to
35835 2017-03-07 Richard Biener <rguenther@suse.de>
35837 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35840 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35842 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35843 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35845 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35848 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35849 to end of description.
35850 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35852 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35854 PR rtl-optimization/79901
35855 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35857 (*avx512f_<code><mode>3<mask_name>): ... this.
35858 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35859 iterator instead of VI8_AVX2_AVX512BW.
35861 PR rtl-optimization/79901
35862 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35863 min/max expander, expand it using expand_vec_cond_expr.
35866 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35869 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35872 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35873 to void * for PCH reasons.
35874 * dwarf2out.c (output_loc_operands, output_die): Cast
35875 v.val_vec.array to unsigned char *.
35877 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35880 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35883 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35885 PR rtl-optimization/79571
35886 * lra-constraints.c (process_alt_operands): Calculate static
35887 reject and subtract it from overall when only addresses will be
35890 2017-03-06 Julia Koval <julia.koval@intel.com>
35893 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35894 incoming stack boundary to 128 for 64-bit targets.
35896 2017-03-06 Richard Biener <rguenther@suse.de>
35898 PR tree-optimization/79894
35899 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35900 to NULL after folding it.
35902 2017-03-06 Richard Biener <rguenther@suse.de>
35904 PR tree-optimization/79824
35905 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35906 check disabling peeling for gaps.
35908 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35910 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35911 attributes): Document gettimeofday.
35913 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35915 * config/s390/s390.c (s390_option_override_internal): Set
35916 PARAM_MIN_VECT_LOOP_BOUND
35918 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35920 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35921 * config/s390/s390.md: Likewise.
35923 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35926 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35927 (<avx2_avx512>_perm<mode>): Rename to ...
35928 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35930 (<avx512>_perm<mode>_mask): Rename to ...
35931 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35933 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35934 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35935 instead of VI8F_256_512.
35936 (avx512f_perm<mode>): New define_expand.
35937 (avx512f_perm<mode>_mask): Likewise.
35938 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35939 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35941 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35943 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35944 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35946 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35948 2017-03-06 Martin Liska <mliska@suse.cz>
35951 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35952 when having a SSA NAME w/o VAR_DECL assigned to it.
35954 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35956 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35957 msa_dpsub_<su>_d): Fix MODE for vec_select.
35959 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35961 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35963 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35965 2017-03-06 Richard Biener <rguenther@suse.de>
35967 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35968 * plugin.c (register_plugin_info): Likewise.
35969 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35971 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35973 * config/i386/sse.md (sse_storehps, sse_storelps,
35974 avx_<castmode><avxsizesuffix>_<castmode>,
35975 avx512f_<castmode><avxsizesuffix>_<castmode>,
35976 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35977 in condition that at least one operand is not a MEM.
35979 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35981 PR middle-end/79805
35982 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35983 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35985 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35986 gimple_call_nothrow_p flag based on whether original builtin can throw.
35987 If it can, emit following stmts on the fallthrough edge.
35988 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35989 don't create new bb if inserting just debug stmts on the edge, try to
35990 insert them on the fallthru bb or just reset debug stmts.
35992 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35995 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35996 restore recog_data (including the operand rtxes inside it) around
35997 the call to get_insn_template.
35999 2017-03-03 Martin Sebor <msebor@redhat.com>
36001 PR tree-optimization/79699
36002 * context.c (context::~context): Free MPFR caches to avoid
36003 a memory leak on program exit.
36005 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36007 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36008 Use wide_int::ulow () instead of .elt (0).
36010 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36012 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36013 (*pushxf): Limit oF constraint to 32bit targets and add oC
36014 constraint for 64bit targets.
36015 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36016 (*pushdf): Change rmF constraint to rmC.
36018 2017-03-03 Martin Liska <mliska@suse.cz>
36020 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36021 Remove unused variable.
36023 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36026 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36027 is a memory operand, increase num_memory.
36028 (ix86_expand_args_builtin): Likewise.
36030 2017-03-03 Jan Hubicka <jh@suse.cz>
36033 * ipa-devirt.c (maybe_record_node): Properly handle
36034 __cxa_pure_virtual visibility.
36036 2017-03-03 Martin Liska <mliska@suse.cz>
36038 PR tree-optimization/79803
36039 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36041 (pass_loop_prefetch::execute): Disabled optimization if an
36042 assumption about L1 cache size is not met.
36044 2017-03-03 Martin Liska <mliska@suse.cz>
36046 PR rtl-optimization/79574
36047 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36048 (hash_scan_set): Likewise.
36049 (dump_hash_table): Likewise.
36050 (hoist_code): Likewise.
36052 2017-03-03 Richard Biener <rguenther@suse.de>
36054 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36055 in place of uhigh ().
36056 (fixed_convert_from_real): Likewise.
36058 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36061 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36063 2017-03-03 Richard Biener <rguenther@suse.de>
36065 PR middle-end/79818
36066 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36067 TYPE_OVERFLOW_UNDEFINED check.
36069 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36071 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36073 (vector_ae_<mode>_p): Likewise.
36074 (vector_nez_<mode>_p): Likewise.
36075 (vector_ne_v2di_p): Likewise.
36076 (vector_ae_v2di_p): Likewise.
36077 (vector_ne_<mode>_p): Likewise.
36078 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36080 (vsx_tsqrt<mode>2_fe): Likewise.
36082 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36085 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36087 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36089 PR rtl-optimization/79780
36090 * cprop.c (one_cprop_pass): When second and further conditional trap
36091 in a single basic block is turned into an unconditional trap, turn it
36092 into a deleted note to avoid RTL verification failures.
36094 2017-03-02 Richard Biener <rguenther@suse.de>
36096 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36098 2017-03-02 Richard Biener <rguenther@suse.de>
36100 PR tree-optimization/79345
36102 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36103 param and abort the walk, returning -1 if it is hit.
36104 (walk_aliased_vdefs): Take a limit param and pass it on.
36105 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36106 defaulting to 0 and return a signed int.
36107 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36108 (check_defs): New helper.
36109 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36110 about uninitialized memory.
36111 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36112 bogus uninitialized warning.
36113 (fixed_convert_from_real): Likewise.
36115 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36117 PR tree-optimization/66768
36118 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36119 iv_use if base object can't be determined.
36121 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36123 PR tree-optimization/79345
36124 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36125 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36126 (get_pattern_stats): Initialize it.
36127 * genemit.c (gen_expand): Verify match_scratch numbers come after
36128 match_operand/match_dup numbers.
36129 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36130 match_scratch numbers.
36131 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36133 * config/s390/s390.md (trunctdsd2): Likewise.
36135 2017-03-02 Richard Biener <rguenther@suse.de>
36137 * wide-int.h (wide_int_storage::operator=): Implement in terms
36140 2017-03-02 Richard Biener <rguenther@suse.de>
36142 PR tree-optimization/79777
36143 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36144 the to insert expression to sth existing.
36146 2017-03-01 Martin Sebor <msebor@redhat.com>
36148 PR middle-end/79692
36149 * gimple-ssa-sprintf.c
36150 (directive::known_width_and_precision): New function.
36151 (format_integer): Use it.
36152 (get_mpfr_format_length): Consider the full range of precision
36153 when computing %g output with the # flag. Set the likely byte
36154 count to 3 rather than 1 when precision is indeterminate.
36155 (format_floating): Correct the lower bound of precision.
36157 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36159 * doc/invoke.texi: Document default code model for 64-bit Linux.
36161 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36164 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36165 udiv rather than div since input pattern is unsigned.
36167 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36169 * config/i386/i386.c (print_reg): Warn for values of
36170 unsupported size in integer register.
36172 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36175 * config/rs6000/predicates.md (current_file_function_operand): Do
36176 not allow self calls to be local if the function is replaceable.
36178 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36181 * config/rs6000/altivec.h (vec_ctz and others): Change the
36182 preprocessor macro that controls conditional compilation from
36183 _ARCH_PWR9 to __POWER9_VECTOR__.
36184 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36185 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36186 control (instead of _ARCH_PWR9 control) so that template
36187 definition uses power9-specific function.
36188 (vec_any_eq): Likewise.
36189 (vec_all_ne): Change macro definition to use a power9-specific
36190 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36191 _ARCH_PWR9 control).
36192 (vec_any_eq) Likewise.
36193 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36194 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36195 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36196 support for xvcmpnedp instruction.
36197 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36198 macro expansion so that Power9 implementation of vec_all_ne does
36199 not use the AltiVec predicate framework.
36200 (VCMPNEH_P): Likewise.
36201 (VCMPNEW_P): Likewise.
36202 (VCMPNED_P): Likewise.
36203 (VCMPNEFP_P): Likewise.
36204 (VCMPNEDP_P): Likewise.
36205 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36206 implementation of vec_any_eq to not use AltiVec predicate
36208 (VCMPAEH_P): Likewise.
36209 (VCMPAEW_P): Likewise.
36210 (VCMPAED_P): Likewise.
36211 (VCMPAEFP_P): Likewise.
36212 (VCMPAEDP_P): Likewise.
36213 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36214 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36215 not use the AltiVec predicate framework.
36216 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36217 of vec_any_eq to not use AltiVec predicate framework.
36218 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36219 support for predefined __POWER9_VECTOR__ macro to indicate that
36220 Power9 instruction selection is enabled.
36221 (altivec_overloaded_builtins): Remove extraneous
36222 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36223 function argument types RS6000_BTI_bool_V16QI and
36224 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36225 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36226 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36227 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36228 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36229 Power9 for implementations of vec_cmpne. Change the signature for
36230 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36231 (representing vec_all_ne) to remove the previously described first
36232 argument of type RS6000_BTI_INTSI, as this was an artifact of
36233 reliance on the AltiVec predicate framework, which is no longer
36234 used in the implementation of these functions. Add
36235 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36236 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36237 since, unlike the AltiVec predicate framework implementation, we
36238 do not share function descriptors between vec_alle and vec_anyeq.
36239 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36240 set of modes that receive special treatment even when
36241 TARGET_P9_VECTOR is true. The special treatment emits code that
36242 does not depend on Power9 instructions.
36243 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36244 define_expand to not rely on AltiVec predicate framework.
36245 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36247 (vector_ne_v2di_p): Change this define_expand to not rely on
36248 AltiVec predicate framework.
36249 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36251 (vector_ne_<mode>_p): Change this define_expand to not rely on
36252 AltiVec predicate framework.
36253 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36255 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36256 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36257 define_insn pattern.
36258 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36259 define_insn pattern because the xvcmpne<VSs>. instruction is not
36261 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36262 instruction is not supported.
36264 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36266 * config/nvptx/nvptx.c: Include intl.h.
36268 2017-03-01 Martin Jambor <mjambor@suse.cz>
36271 * ipa-prop.h (ipa_bits): Removed field known.
36272 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36273 to pointers. Adjusted their comments to warn about their sharing.
36274 (ipcp_transformation_summary): Change bits to a vector of pointers.
36275 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36276 (ipa_get_ipa_bits_for_value): Declare.
36277 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36278 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36279 (ipa_bits_hash_table): Likewise.
36280 (ipa_vr_ggc_hash_traits): Likewise.
36281 (ipa_vr_hash_table): Likewise.
36282 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36283 being pointers and vr_known being removed.
36284 (ipa_set_jf_unknown): Likewise.
36285 (ipa_get_ipa_bits_for_value): New function.
36286 (ipa_set_jfunc_bits): Likewise.
36287 (ipa_get_value_range): New overloaded functions.
36288 (ipa_set_jfunc_vr): Likewise.
36289 (ipa_compute_jump_functions_for_edge): Use the above functions to
36290 construct bits and vr parts of jump functions.
36291 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36292 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36294 (ipcp_grow_transformations_if_necessary): Also allocate
36295 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36297 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36298 them. Fix too long lines.
36299 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36300 vr_known being removed.
36301 (ipa_read_jump_function): Use new setter functions to construct bits
36302 and vr parts of jump functions or set them to NULL.
36303 (write_ipcp_transformation_info): Adjust for bits being pointers.
36304 (read_ipcp_transformation_info): Likewise.
36305 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36307 Include gt-ipa-prop.h.
36308 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36310 (ipcp_store_bits_results): Likewise.
36311 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36312 Do not write to existing jump functions but use a temporary instead.
36314 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36317 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36318 attempt to use its first operand as BIT_FIELD_REF base.
36320 2017-03-01 Richard Biener <rguenther@suse.de>
36322 PR middle-end/79721
36323 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36324 interpolating formula in wrapping arithmetic.
36325 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36327 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36329 PR tree-optimization/79734
36330 * tree-vect-generic.c (expand_vector_condition): Optimize
36331 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36332 Handle VEC_COND_EXPR where comparison has different inner width from
36333 type's inner width.
36335 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36337 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36338 markup, and similar issues. Remove @opindex entries for things
36339 that aren't options. Add missing -mmpy-option entries.
36341 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36343 PR tree-optimization/79737
36344 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36345 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36346 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36347 instead of byte_size. Formatting fix.
36348 (shift_bytes_in_array_right): Formatting fix.
36350 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36353 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36354 condition on optimize for the leaf function test.
36356 2017-02-28 Martin Liska <mliska@suse.cz>
36359 * read-rtl-function.c (function_reader::handle_unknown_directive):
36360 Bail out when one uses -flto.
36362 2017-02-28 Martin Liska <mliska@suse.cz>
36364 * common.opt: Replace space with tabular for options of <number>
36366 * config/i386/i386.opt: Show <number> value for
36367 -mlarge-data-threshold.
36368 * opts.c (print_filtered_help): Do not display number in hexadecimal
36371 2017-02-28 Martin Liska <mliska@suse.cz>
36373 * common.opt: Fix --help=option -Q for options which are of
36376 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36378 * config/i386/i386.c (print_reg): Error out for values
36379 of 8-bit size in invalid integer register.
36381 2017-02-28 Martin Sebor <msebor@redhat.com>
36383 PR tree-optimization/79691
36384 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36386 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36389 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36390 gcc_unreachable with output_operand_lossage.
36392 2017-02-28 Richard Biener <rguenther@suse.de>
36394 PR tree-optimization/79740
36395 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36397 (visit_nary_op): Insert the nary into the hashtable if we
36398 pattern-matched sth.
36399 * tree-ssa-pre.c (eliminate_insert): Robustify.
36401 2017-02-28 Richard Biener <rguenther@suse.de>
36403 PR middle-end/79731
36404 * fold-const.c (decode_field_reference): Reject out-of-bound
36407 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36409 * config/i386/i386.c: Include intl.h.
36410 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36411 instead of just cond ? "..." : "...".
36412 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36413 * coverage.c (read_counts_file): Likewise.
36414 * omp-offload.c: Include intl.h.
36415 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36416 of just cond ? "..." : "...".
36417 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36418 of just cond ? "..." : "...".
36420 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36423 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36425 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36426 'tune for' CPU name.
36427 * config/arm/arm-cpu-data.h: Regenerated.
36429 2017-02-28 Richard Biener <rguenther@suse.de>
36431 PR tree-optimization/79732
36432 * tree-inline.c (expand_call_inline): Do not shadow var.
36434 2017-02-28 Richard Biener <rguenther@suse.de>
36436 PR tree-optimization/79723
36437 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36438 address-space properly.
36440 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36442 * doc/optinfo.texi (Optimization groups): Fix option used for
36444 * doc/invoke.texi (-fopt-info): Document "omp".
36445 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36446 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36447 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36448 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36449 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36451 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36453 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36454 OPTGROUP_OMP with "omp".
36456 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36459 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36460 for arithmetic shift of unsigned V2DI.
36462 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36464 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36465 arc/linux.h headers.
36466 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36467 (LINK_SPEC): Likewise.
36468 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36469 (EXTRA_SPECS): Likewise.
36470 (STARTFILE_SPEC): Likewise.
36471 (ENDFILE_SPEC): Likewise.
36472 (LIB_SPEC): Likewise.
36473 (TARGET_SDATA_DEFAULT): Likewise.
36474 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36475 (MULTILIB_DEFAULTS): Likewise.
36476 (DWARF2_UNWIND_INFO): Likewise.
36477 * config/arc/big.h: New file.
36478 * config/arc/elf.h: Likewise.
36479 * config/arc/linux.h: Likewise.
36480 * config/arc/t-uClibc: Remove.
36482 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36484 PR tree-optimization/77536
36485 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36486 (tree_transform_and_unroll_loop): Use above function to compute the
36487 estimated niter of unrolled loop and use it when scaling profile.
36488 Also use count info rather than frequency if it's non-zero.
36489 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36490 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36491 (vect_transform_loop): Call above function.
36493 2017-02-27 Richard Biener <rguenther@suse.de>
36495 PR tree-optimization/45397
36496 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36497 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36498 (visit_nary_op): Add pattern matching for CSEing sign-changed
36499 or truncated operations with wider ones.
36501 2017-02-27 Richard Biener <rguenther@suse.de>
36503 PR tree-optimization/79690
36504 * tree-vect-stmts.c (vectorizable_store): Use vector type
36505 built from the DR with address-space.
36507 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36509 * doc/invoke.texi (Optimize Options): Refine the description
36510 of asan-use-after-return.
36512 2017-02-25 Alan Modra <amodra@gmail.com>
36514 PR rtl-optimization/79584
36515 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36516 base, not ad->base_term, the reg within base. Remove assertion
36517 that ad->base == ad->base_term. Replace gen_int_mode using
36518 bogus mode with const0_rtx.
36520 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36522 PR middle-end/79396
36523 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36524 FMA_EXPR like tcc_binary or tcc_unary.
36526 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36529 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36531 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36532 (output_loc_operands): Handle DW_OP_call_ref and
36533 DW_OP_GNU_variable_value.
36534 (struct variable_value_struct): New type.
36535 (struct variable_value_hasher): Likewise.
36536 (variable_value_hash): New variable.
36537 (string_types): Remove.
36538 (copy_loc_descr): New function.
36539 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36540 (prepend_loc_descr_to_each): New function.
36541 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36542 instead of add_loc_descr_to_each if the first argument is single
36543 location list and the second has multiple.
36544 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36545 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36546 when looking for variable value which doesn't have other location info.
36547 (loc_list_from_tree): Formatting fix.
36548 (gen_array_type_die): Simplify DW_AT_string_length handling.
36549 (adjust_string_types): Remove.
36550 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36551 string_types. Call resolve_variable_values.
36552 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36553 (resolve_addr_in_expr): Likewise. Add A argument.
36554 (copy_deref_exprloc): Remove deref argument. Adjust for the
36555 original expression being DW_OP_GNU_variable_value with optionally
36556 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36557 optionally after it.
36558 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36559 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36560 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36561 (variable_value_hasher::hash, variable_value_hasher::equal): New
36563 (resolve_variable_value_in_expr, resolve_variable_value,
36564 resolve_variable_values, note_variable_value_in_expr,
36565 note_variable_value): New functions.
36566 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36569 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36572 * opts.h (handle_generated_option): Add GENERATED_P argument.
36573 * opts-common.c (handle_option): Adjust function comment.
36574 (handle_generated_option): Add GENERATED_P argument, pass it to
36576 (control_warning_option): Pass false to handle_generated_option
36578 * opts.c (maybe_default_option): Pass true to handle_generated_option
36580 * optc-gen.awk: Likewise.
36582 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36584 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36585 a REG, look at the REG it is a SUBREG of.
36586 (splitter for cmpeqsi_t): Ditto.
36588 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36590 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36591 the special USEs with the pattern of the insn, not the insn itself.
36593 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36596 * doc/invoke.texi: Document -mload-store-pairs.
36598 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36599 Sandra Loosemore <sandra@codesourcery.com>
36601 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36602 argument isn't a CONST_INT.
36603 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36604 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36605 (nios2_validate_compare): Bypass alternate compare logic if *op2
36606 is not a CONST_INT.
36607 (ldstwm_operation_p): Return false if first_base is not a REG or
36608 if first_offset is not a CONST_INT.
36610 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36612 * config/cris/cris.md: Use correct operand in a define_peephole2.
36614 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36616 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36618 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36620 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36621 this_insn if it is an INSN or JUMP_INSN.
36622 (force_offsettable): Look at base, not at addr.
36623 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36624 on things that aren't necessarily CONST_INTs.
36626 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36628 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36629 -mfpmath=sse is the default also for x86-32 targets with SSE2
36630 instruction set when @option{-ffast-math} is enabled
36632 2017-02-24 Jeff Law <law@redhat.com>
36634 PR rtl-optimizatoin/79286
36635 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36638 2017-02-24 Richard Biener <rguenther@suse.de>
36640 PR tree-optimization/79389
36641 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36644 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36646 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36647 function comment to reflect reality.
36648 (loop_exits_before_overflow): Fix typo in function description.
36650 2017-02-24 Richard Biener <rguenther@suse.de>
36652 PR tree-optimization/79389
36653 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36654 properly that a threading opportunity exists. Detect conditional
36655 copy/constant propagation opportunities.
36657 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36659 * config/visium/visium.md (type): Add trap.
36660 (b): New mode attribute.
36661 (*btst): Rename into...
36662 (*btst<mode>): ...this and adjust.
36663 (*cbranchsi4_btst_insn): Rename into...
36664 (*cbranch<mode>4_btst_insn): ...this and adjust.
36665 (trap): New define_insn.
36667 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36669 PR tree-optimization/79389
36670 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36671 (noce_reversed_cond_code): New function.
36672 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36673 reversed_comparison_code. Formatting fix.
36674 (noce_try_store_flag): Test rev_cond != NULL in addition to
36675 reversed_comparison_code.
36676 (noce_try_store_flag_constants): Likewise.
36677 (noce_try_store_flag_mask): Likewise.
36678 (noce_try_addcc): Use rev_cond if non-NULL instead of
36679 reversed_comparison_code.
36680 (noce_try_cmove_arith): Likewise. Formatting fixes.
36681 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36682 (noce_find_if_block): Initialize rev_cond.
36683 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36684 instead of false as last argument never attempt to reverse it
36687 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36689 PR tree-optimization/79663
36690 * tree-predcom.c (combine_chains): Process refs in reverse order
36691 only for ZERO length chains, and add explaining comment.
36693 2017-02-23 Jeff Law <law@redhat.com>
36695 PR tree-optimization/79578
36696 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36697 in call to operand_equal_p.
36699 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36702 * config/i386/cpuid.h: Fix another undefined behavior.
36704 2017-02-23 Richard Biener <rguenther@suse.de>
36706 PR tree-optimization/79683
36707 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36708 vector types for data-refs.
36710 2017-02-23 Martin Liska <mliska@suse.cz>
36712 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36714 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36716 PR middle-end/79665
36717 * internal-fn.c (get_range_pos_neg): Moved to ...
36718 * tree.c (get_range_pos_neg): ... here. No longer static.
36719 * tree.h (get_range_pos_neg): New prototype.
36720 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36721 are known to be in between 0 and signed maximum inclusive, try to
36722 expand both unsigned and signed divmod and use the cheaper one from
36725 2017-02-22 Jeff Law <law@redhat.com>
36727 PR tree-optimization/79578
36728 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36729 to compare base operands.
36731 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36734 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36735 gpc_reg_operand instead of fpr_reg_operand.
36737 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36739 * config/mips/mips.c (mips_return_in_memory): Force FP
36740 vector types to be returned in memory for o32 ABI.
36742 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36744 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36745 instead of DW_TAG_member for static data member declarations and don't
36746 set no_linkage_name for static inline data members.
36747 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36750 2017-02-22 Martin Liska <mliska@suse.cz>
36752 * doc/invoke.texi: Replace inequality signs with square brackets
36755 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36758 * lra-constraints.c (simplify_operand_subreg): Handle
36759 WORD_REGISTER_OPERATIONS targets.
36761 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36764 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36765 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36766 elimination by swapping fld*.
36768 2017-02-22 Richard Biener <rguenther@suse.de>
36770 PR tree-optimization/79673
36771 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36772 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36773 irrelevant address-space qualifiers and avoiding a
36774 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36776 2017-02-22 Richard Biener <rguenther@suse.de>
36778 PR tree-optimization/79666
36779 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36780 to not symbolically negate if that may introduce undefined
36783 2017-02-22 Martin Liska <mliska@suse.cz>
36786 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36787 * data-streamer-out.c (streamer_write_gcov_count_stream):
36789 * value-prof.c (stream_out_histogram_value): Make assert more
36790 precise based on type of counter.
36792 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36795 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36796 Use nonimmediate_operand instead of memory_operand for operand 1.
36797 (float-extend standard_x87sse_constant_load splitter): Ditto.
36799 2017-02-21 Jeff Law <law@redhat.com>
36801 PR tree-optimization/79621
36802 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36803 blocks with edges to themselves.
36805 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36808 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36809 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36810 Use gimple_call_builtin_p.
36813 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36814 on temporarily removed DEBUG_INSNs.
36816 PR tree-optimization/79649
36817 * tree-loop-distribution.c (classify_partition): Give up on
36818 non-generic address space loads/stores.
36820 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36822 * doc/loop.texi (Loop manipulation): Remove nonexistent
36823 tree_ssa_loop_version from the documentation.
36824 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36826 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36829 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36830 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36831 * config/rs6000/rs6000.c: Include except.h.
36832 (rs6000_expand_split_stack_prologue): Call
36833 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36835 2017-02-21 Martin Jambor <mjambor@suse.cz>
36838 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36839 have been analyzed.
36841 2017-02-21 Martin Jambor <mjambor@suse.cz>
36843 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36844 for backward compatibility only.
36845 * doc/invoke.texi (Option Summary): Remove all references to
36846 -fipa-cp-alignment.
36848 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36852 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36854 * lra-constraints.c (curr_insn_transform): Handle
36855 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36857 2017-02-21 Martin Liska <mliska@suse.cz>
36859 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36861 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36863 PR translation/79638
36864 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36866 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36869 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36870 (arm_function_ok_for_sibcall): Return false for an indirect call by
36871 descriptor if all the argument registers are used.
36872 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36873 alignment of the function.
36875 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36877 PR tree-optimization/61441
36878 * simplify-rtx.c (simplify_const_unary_operation): For
36879 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36880 the sNaN unmodified.
36882 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36884 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36885 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36886 instead of SYSTEM_HEADER_DIR.
36888 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36889 Martin Liška <mliska@suse.cz>
36891 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36892 Fix typos and grammar, use active voice, and clarify.
36894 2017-02-20 Marek Polacek <polacek@redhat.com>
36896 PR middle-end/79537
36897 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36900 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36902 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36905 * config/i386/i386.c (ix86_expand_builtin): Handle
36906 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36907 ix86_builtins_isa[fcode].isa as a requirement of those
36908 flags and any other flag in the bitmask.
36909 (ix86_init_mmx_sse_builtins): Use 0 instead of
36910 ~OPTION_MASK_ISA_64BIT as mask.
36911 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36912 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36913 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36914 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36916 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36919 * lra-constraints.c (split_reg): Check requested split mode
36920 is supported by the register.
36922 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36924 * lra-constraints.c (simplify_operand_subreg): Remove early
36927 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36930 * lra-constraints.c (curr_insn_transform): Tighten condition
36931 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36933 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36936 * lra-constraints.c (curr_insn_transform): Handle
36937 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36939 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36942 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36944 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36946 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36949 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36952 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36954 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36957 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36959 * final.c (last_columnnum, override_columnnum): New variables.
36960 (final_start_function): Set last_columnnum, pass it to begin_prologue
36961 hook and pass 0 to dwarf2out_begin_prologue.
36962 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36963 to source_line debug hook.
36964 (notice_source_line): Compute last_columnnum and for debug_column_info
36965 return true on column changes.
36966 * debug.h (struct gcc_debug_hooks): Add column argument to
36967 source_line and begin_prologue hooks.
36968 (debug_nothing_int_charstar_int_bool): Remove prototype.
36969 (debug_nothing_int_int_charstar,
36970 debug_nothing_int_int_charstar_int_bool): New prototypes.
36971 (dwarf2out_begin_prologue): Add column argument.
36972 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36973 begin_prologue hooks.
36974 (debug_nothing_int_charstar_int_bool): Remove.
36975 (debug_nothing_int_int_charstar,
36976 debug_nothing_int_int_charstar_int_bool): New functions.
36977 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36978 through to dwarf2out_source_line.
36979 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36980 (dwarf2out_source_line): Add column argument, emit it if requested.
36981 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36983 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36984 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36985 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36986 through to dwarf2out_begin_prologue.
36987 (vmsdbgout_source_line): Add column argument, pass it through to
36988 dwarf2out_source_line.
36989 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36990 dbxout_source_line caller.
36991 (dbxout_source_line): Add column argument.
36993 * common.opt (gno-column-info, gcolumn-info): New options.
36994 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36995 (check_die): Also test for multiple DW_AT_decl_column attributes.
36996 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36997 DW_AT_decl_column if requested.
36998 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37000 (gen_variable_die): Likewise.
37001 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37002 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37005 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37006 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37007 (ix86_handle_option): Handle OPT_m3dnowa.
37008 * doc/invoke.texi (-m3dnowa): Document.
37009 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37010 -m3dnowa instead of -m3dnow -march=athlon.
37013 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37014 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37016 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37019 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37020 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37021 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37022 generator for vsx_xxpermdi_<mode>_be.
37023 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37024 force big-endian semantics.
37025 (vsx_xxpermdi_<mode>_be): New define_expand with same
37026 implementation as previous version of vsx_xxpermdi_<mode>.
37028 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37030 PR tree-optimization/79327
37031 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37032 variable, its initialization and use.
37034 2017-02-17 Julia Koval <julia.koval@intel.com>
37036 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37037 (OPTION_MASK_ISA_PKU_UNSET): New.
37038 (ix86_handle_option): Handle -mrdpid.
37039 * config/i386/cpuid.h (bit_RDPID): New.
37040 * config/i386/driver-i386.c (host_detect_local_cpu):
37041 Detect RDPID feature.
37042 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37043 * config/i386/i386-c.c (ix86_target_macros_internal):
37045 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37046 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37047 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37048 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37049 * config/i386/i386.md (define_insn "rdpid"): New.
37050 * config/i386/i386.opt Add -mrdpid.
37051 * config/i386/immintrin.h (_rdpid_u32): New.
37053 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37055 PR rtl-optimization/79541
37056 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37057 instead of transforming it into USE.
37059 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37061 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37062 If HONOR_SNANS (SFmode) force the input to a register.
37063 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37064 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37065 an frsp or similar insn.
37067 2017-02-17 Martin Liska <mliska@suse.cz>
37069 PR rtl-optimization/79577
37070 * params.def (selsched-max-sched-times): Increase minimum to 1.
37072 2017-02-17 Martin Liska <mliska@suse.cz>
37074 PR rtl-optimization/79574
37075 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37077 2017-02-17 Martin Liska <mliska@suse.cz>
37079 PR tree-optimization/79529
37080 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37081 ssa_defined_default_def_p to handle cases which are implicitly
37083 * tree-ssa.c (ssa_defined_default_def_p): New function.
37084 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37085 which are implicitly defined.
37086 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37088 2017-02-17 Richard Biener <rguenther@suse.de>
37090 PR middle-end/79576
37091 * params.def (max-ssa-name-query-depth): Limit to 10.
37093 2017-02-17 Richard Biener <rguenther@suse.de>
37095 PR tree-optimization/79552
37096 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37099 2017-02-17 Richard Biener <rguenther@suse.de>
37102 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37104 2017-02-17 Marek Polacek <polacek@redhat.com>
37106 PR middle-end/79536
37107 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37108 (fold_negate_expr): New wrapper.
37110 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37112 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37113 Correct terminology and de-emphasize pre-standard behavior.
37115 2017-02-16 Alan Modra <amodra@gmail.com>
37117 PR rtl-optimization/79286
37118 * ira.c (def_dominates_uses): New function.
37119 (update_equiv_regs): Don't create an equivalence for insns that
37120 may trap where the register def does not dominate the use.
37122 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37124 PR rtl-optimization/78127
37125 * lra.c (lra): Call lra_eliminate before finish the loop after
37128 2017-02-16 Richard Biener <rguenther@suse.de>
37130 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37132 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37133 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37134 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37135 (isl_val_int_from_wi): New function.
37136 (extract_affine_gmp): Rename to ...
37137 (extract_affine_wi): ... this, take a widest_int.
37138 (extract_affine_int): Just wrap extract_affine_wi.
37139 (add_param_constraints): Use isl_val_int_from_wi.
37140 (add_loop_constraints): Likewise, and extract_affine_wi.
37142 2017-02-15 Jeff Law <law@redhat.com>
37144 PR middle-end/79521
37145 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37146 ira_init_register_move_cost_if_necessary.
37148 2017-02-15 Martin Sebor <msebor@redhat.com>
37150 PR middle-end/32003
37151 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37152 removed in a prior commit.
37154 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37156 PR tree-optimization/79347
37157 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37158 counters during peeling.
37160 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37162 * Makefile.in (site.exp): Remove "set ISLVER".
37164 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37167 * real.c (real_from_integer): Call real_convert even for decimal.
37169 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37172 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37174 2017-02-14 Andrew Pinski <apinski@cavium.com>
37176 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37177 cores and change the partno/implementer to be correct.
37178 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37179 the 'B" as the implementer.
37180 * config/aarch64/aarch64-tune.md: Regenerate.
37182 2017-02-14 Carl Love <cel@us.ibm.com>
37184 * config/rs6000/rs6000.c: Add case statement entry to make the
37185 xvcvuxdsp built-in argument unsigned.
37186 * config/rs6000/vsx.md: Fix the source and return operand types so they
37187 match the instruction definitions from the ISA document. Fix typo
37188 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37191 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37194 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37195 member early_clobber_alts.
37196 * lra-lives.c (reg_early_clobber_p): New.
37197 (process_bb_lives): Use it.
37198 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37199 (debug_operand_data): Initialize early_clobber_alts.
37200 (setup_operand_alternative): Set up early_clobber_alts.
37201 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37202 alternatives to new_insn_reg.
37203 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37205 (lra_update_insn_regno_info): Pass the new arg.
37207 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37209 PR middle-end/79505
37210 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37211 (new_oacc_loop_raw): Don't clear already cleared fields.
37214 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37215 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37216 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37218 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37221 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37223 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37226 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37227 the extra instruction to the right place to store 128-bit constant
37230 2017-02-14 Martin Sebor <msebor@redhat.com>
37232 PR middle-end/79448
37233 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37234 warning for strings of unknown length.
37236 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37238 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37240 2017-02-14 Jeff Law <law@redhat.com>
37243 * ira-costs.c (scan_one_insn): Initialize register move costs
37244 for pseudos seen in USE/CLOBBER insns.
37246 PR tree-optimization/79095
37247 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37248 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37249 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37250 if the operands are known to be not equal, then the resulting range
37252 (intersect_ranges): If the new range is ~[0,0] and the old range is
37253 wide, then prefer ~[0,0].
37254 * tree-vrp.c (overflow_comparison_p_1): New function.
37255 (overflow_comparison_p): New function.
37256 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37257 if NAME is used in an overflow test.
37258 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37259 overflow check that can be expressed as an equality test, then adjust
37260 ops to be that equality test.
37262 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37264 * config/s390/s390-builtin-types.def: Remove flags argument.
37265 * config/s390/s390.c (s390_init_builtins): Likewise.
37267 2017-02-14 Martin Liska <mliska@suse.cz>
37269 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37270 vector. Fix trailing white spaces.
37272 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37274 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37277 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37279 PR rtl-optimization/68664
37280 * config/arm/arm.c (arm_sched_can_speculate_insn):
37281 New function. Declare prototype.
37282 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37284 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37286 PR rtl-optimization/68664
37287 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37289 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37291 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37293 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37294 max skip bytes for function, loop and jump.
37296 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37298 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37299 ABS_EXPR for gimple dump.
37301 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37304 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37306 PR tree-optimization/79408
37307 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37308 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37309 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37310 also if rhs1 is INTEGER_CST.
37312 2017-02-14 Richard Biener <rguenther@suse.de>
37314 PR middle-end/79432
37315 * tree-into-ssa.c (insert_phi_nodes): When the function can
37316 have abnormal edges rewrite SSA names with broken use-def
37317 dominance out of SSA and register them for PHI insertion.
37319 2017-02-13 Martin Sebor <msebor@redhat.com>
37321 PR middle-end/79496
37322 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37323 clearing info.nowrite flag when snprintf size argument is a range.
37325 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37327 * cprop.c (cprop_jump): Add missing space in string literal.
37328 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37329 (get_constraint_for_component_ref): Likewise.
37330 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37331 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37332 * lra-constraints.c (process_alt_operands): Likewise.
37333 * ipa-inline.c (inline_small_functions): Likewise.
37334 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37335 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37336 * trans-mem.c (diagnose_tm_1_op): Likewise.
37337 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37338 (grid_parallel_clauses_gridifiable): Likewise.
37340 * config/nvptx/mkoffload.c (process): Add space in between
37343 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37344 "MOD4_SSE_REGS" and "ALL_REGS".
37346 * spellcheck.c (test_data): Add , in between "foo" and "food".
37348 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37351 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37352 boundary crossing check and subsequent code generation agree.
37354 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37356 * config/aarch64/aarch64.c (has_memory_op): Delete.
37357 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37360 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37362 PR rtl-optimization/79388
37363 PR rtl-optimization/79450
37364 * combine.c (distribute_notes): When removing TEM_INSN for which
37365 corresponding dest has last value recorded, invalidate that last
37368 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37370 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37371 of explicit '@'. Add missing assembly comment marker on branch costs
37374 2017-02-13 Nathan Sidwell <nathan@acm.org>
37376 * gengtype-lex.l (<in_struct>): Add '/'.
37378 2017-02-13 Martin Liska <mliska@suse.cz>
37381 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37383 2017-02-13 Richard Biener <rguenther@suse.de>
37385 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37387 * configure: Re-generate.
37388 * config.in: Likewise.
37389 * graphite-dependences.c: Simplify as if
37390 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37391 * graphite-isl-ast-to-gimple.c: Likewise.
37392 * graphite-optimize-isl.c: Likewise.
37393 * graphite-poly.c: Likewise.
37394 * graphite-sese-to-poly.c: Likewise.
37395 * graphite.h: Likewise.
37396 * toplev.c: Include isl/version.h and use isl_version () for
37397 printing the ISL version.
37398 * doc/install.texi: Update ISL requirement.
37400 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37402 * doc/standards.texi (Standards): Update reference to
37405 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37407 * doc/extend.texi (Named Address Spaces): sourceware.org now
37409 * doc/install.texi (Binaries): Ditto.
37412 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37414 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37415 standard terminology "stringize"/"stringizing" throughout.
37416 * doc/cppinternals.texi: Likewise.
37418 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37420 * doc/extend.texi: Fix some spelling mistakes and typos.
37421 * doc/invoke.texi: Likewise.
37423 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37426 * params.def (inline-min-speedup) Change from 10 to 8.
37428 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37430 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37433 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37436 * ipa-inline-analysis.c (get_minimal_bb): New function.
37437 (record_modified): Use it.
37438 (remap_edge_change_prob): Handle also ancestor functions.
37440 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37442 * doc/contrib.texi (Contributors): Remove broken link into
37443 the Mauve CVS repository.
37445 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37447 PR middle-end/79454
37448 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37449 result computation whenever lhs doesn't have vector mode, not
37450 just when it has BLKmode.
37452 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37454 * doc/makefile.texi (profiledbootstrap): Refer to the
37455 installation instructions only in textual form.
37457 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37460 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37462 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37464 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37465 (Specific): Update mingw-w64 reference.
37467 (Specific): Remove broken link to Renesas RX processor.
37469 2017-02-10 Richard Biener <rguenther@suse.de>
37471 * toplev.c (process_options): Do not mention obsolete graphite
37472 options when printing sorry message about missing graphite support.
37473 Mention -floop-nest-optimize.
37475 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37477 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37478 (vtst_p16): Likewise.
37479 (vtstq_p8): Likewise.
37480 (vtstq_p16): Likewise.
37482 (vtstq_p64): Likewise.
37483 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37484 (vset_lane_p64): New.
37485 (vsetq_lane_p64): New.
37487 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37489 PR tree-optimization/79411
37490 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37491 stmt operands are SSA_NAMEs used in abnormal phis.
37492 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37495 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37498 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37501 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37503 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37505 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37507 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37511 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37512 not arbitrary TREE_CONSTANT.
37515 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37516 "omp declare target link" attribute unless is_global_var.
37517 * omp-offload.c (find_link_var_op): Likewise.
37519 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37520 Chung-Lin Tang <cltang@codesourcery.com>
37522 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37524 (gimplify_adjust_omp_clauses): Don't delete TILE.
37525 (gimplify_omp_for): Deal with TILE.
37526 * internal-fn.c (expand_GOACC_TILE): New function.
37527 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37529 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37530 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37532 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37533 avoid DIV for outermost collapse var.
37534 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37535 Remove out of date comments, fix whitespace.
37536 * omp-general.c (omp_extract_for_data): Deal with tiling.
37537 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37538 adjust OLF_DIM_BASE value.
37539 (struct omp_for_data): Add tiling field.
37540 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37541 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37542 for auto loops. Remove default auto determining, moved to
37543 oacc_loop_fixed_partitions.
37544 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37545 stmts, add e_mask field.
37546 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37547 (oacc_thread_numbers): Use oacc_dim_call.
37548 (oacc_xform_tile): New.
37549 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37550 (finish_oacc_loop): Adjust for ifns vector.
37551 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37552 add case for GOACC_TILE fns.
37553 (oacc_loop_xform_loop): Delete.
37554 (oacc_loop_process): Iterate over call list directly, and add
37555 handling for GOACC_TILE fns.
37556 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37558 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37559 vector partitioning to outer loops. Assign 2 partitions to loops
37560 when available. Add TILE handling.
37561 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37562 (execite_oacc_device_lower): Process GOACC_TILE fns,
37563 ignore unknown specs.
37564 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37565 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37566 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37568 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37570 * configure.ac (ACX_BUGURL): Update.
37571 * configure: Regenerate.
37573 2017-02-09 Richard Biener <rguenther@suse.de>
37575 PR tree-optimization/69823
37576 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37577 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37579 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37581 * config/arc/arc-c.def: Add __NPS400__ definition.
37582 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37583 (TARGET_NPS400): Define.
37585 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37587 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37589 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37590 pointer, arch_info.
37591 (arc_cpu_types): Fill the arch_info field with a pointer into the
37592 arc_arch_types table.
37593 (arc_selected_cpu): Declare.
37594 * config/arc/arc.c (arc_selected_cpu): Make global.
37595 (arc_selected_arch): Delete.
37596 (arc_base_cpu): Delete.
37597 (arc_override_options): Remove references to deleted variables,
37598 update access to arch information.
37599 (ARC_OPT): Update access to arch information.
37600 (ARC_OPTX): Likewise.
37601 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37602 (TARGET_ARC600): Update access to arch information.
37603 (TARGET_ARC601): Likewise.
37604 (TARGET_ARC700): Likewise.
37605 (TARGET_EM): Likewise.
37606 (TARGET_HS): Likewise.
37607 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37610 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37613 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37614 condition/operands for integer GE/LE/GEU/LEU operations.
37616 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37618 PR translation/79397
37619 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37622 2017-02-08 Martin Jambor <mjambor@suse.cz>
37625 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37626 whether allocation happened.
37627 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37628 nothing was allocated.
37630 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37632 PR tree-optimization/79408
37633 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37634 constant, but SSA_NAME with a known integer range, use the minimum
37635 of that range instead of op1 to determine if modulo can be replaced
37636 with its first operand.
37638 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37640 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37642 2017-02-08 Richard Biener <rguenther@suse.de>
37644 PR tree-optimization/71824
37645 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37646 Check all loops contained in the merged region.
37648 2017-02-07 Andrew Pinski <apinski@cavium.com>
37650 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37652 2017-02-07 Andrew Pinski <apinski@cavium.com>
37654 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37655 (thunderxt88): Likewise.
37656 (thunderxt81): Disable LSE and change v8.1 to v8.
37657 (thunderxt83): Likewise.
37659 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37660 Richard Biener <rguenther@suse.de>
37662 PR middle-end/79399
37663 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37664 type from int to size_t.
37665 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37667 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37669 PR rtl-optimization/79386
37670 * cprop.c (bypass_conditional_jumps): Initialize
37671 bypass_last_basic_block already before splitting bbs after
37672 unconditional traps...
37673 (bypass_conditional_jumps): ... rather than here.
37676 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37677 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37678 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37679 fix -masm=intel patterns.
37681 2017-02-07 Richard Biener <rguenther@suse.de>
37683 PR tree-optimization/79256
37684 PR middle-end/79278
37685 * builtins.c (get_object_alignment_2): Use min_align_of_type
37686 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37687 and ADJUST_FIELD_ALIGN.
37689 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37691 * doc/tm.texi: Regenerate.
37692 * stor-layout.c (layout_decl): Adjust.
37693 (update_alignment_for_field): Likewise.
37694 (place_field): Likewise.
37695 (min_align_of_type): Likewise.
37696 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37697 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37698 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37699 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37700 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37701 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37702 * config/i386/i386.c (x86_field_alignment): Likewise.
37703 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37704 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37705 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37706 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37707 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37708 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37712 2017-01-30 Richard Biener <rguenther@suse.de>
37714 PR tree-optimization/79256
37715 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37716 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37719 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37721 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37722 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37723 builtins to SImode and emit a zero-extend, if necessary.
37725 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37727 * docs/invoke.texi (RISC-V Options): Alphabetize.
37729 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37731 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37734 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37736 * config/riscv/riscv.c: New file.
37737 * common/config/riscv/riscv-common.c: Likewise.
37738 * config.gcc: Likewise.
37739 * config/riscv/constraints.md: Likewise.
37740 * config/riscv/elf.h: Likewise.
37741 * config/riscv/generic.md: Likewise.
37742 * config/riscv/linux.h: Likewise.
37743 * config/riscv/multilib-generator: Likewise.
37744 * config/riscv/peephole.md: Likewise.
37745 * config/riscv/pic.md: Likewise.
37746 * config/riscv/predicates.md: Likewise.
37747 * config/riscv/riscv-builtins.c: Likewise.
37748 * config/riscv/riscv-c.c: Likewise.
37749 * config/riscv/riscv-ftypes.def: Likewise.
37750 * config/riscv/riscv-modes.def: Likewise.
37751 * config/riscv/riscv-opts.h: Likewise.
37752 * config/riscv/riscv-protos.h: Likewise.
37753 * config/riscv/riscv.h: Likewise.
37754 * config/riscv/riscv.md: Likewise.
37755 * config/riscv/riscv.opt: Likewise.
37756 * config/riscv/sync.md: Likewise.
37757 * config/riscv/t-elf-multilib: Likewise.
37758 * config/riscv/t-linux: Likewise.
37759 * config/riscv/t-linux-multilib: Likewise.
37760 * config/riscv/t-riscv: Likewise.
37761 * configure.ac: Likewise.
37762 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37763 Waterman as RISC-V maintainers.
37764 * doc/install.texi: Add RISC-V entries.
37765 * doc/invoke.texi: Add RISC-V options section.
37766 * doc/md.texi: Add RISC-V constraints section.
37767 * configure: Regenerated.
37769 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37772 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37773 false values to be constant vectors with all 0 or all 1 bits set.
37774 (vcondu<mode><mode>): Likewise.
37775 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37777 (fpmask_comparison_operator): Update comment.
37778 (vecint_comparison_operator): New predicate.
37779 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37780 vector conditionals when the true and false values are constant
37781 vectors with all 0 bits or all 1 bits set.
37783 2017-02-06 Martin Sebor <msebor@redhat.com>
37785 PR tree-optimization/79376
37786 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37788 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37790 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37791 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37792 to simplify split condition.
37794 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37796 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37797 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37800 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37802 PR rtl-optimization/68664
37803 * target.def (can_speculate_insn): New hook.
37804 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37805 * doc/tm.texi: Regenerate.
37806 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37807 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37808 (rs6000_sched_can_speculate_insn): New function.
37810 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37812 PR tree-optimization/79284
37813 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37814 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37815 vectorizable_mask_load_store, vectorizable_operation,
37816 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37817 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37818 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37819 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37820 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37821 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37822 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37823 is_gimple_assign (stmt). Replace another such test with
37824 is_gimple_assign (stmt).
37826 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37829 * config/avr/avr.c (rtl-iter.h): Include it.
37830 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37831 (avr_legitimate_combined_insn): ...and implementation.
37833 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37835 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37836 * config/s390/s390.c (s390_const_operand_ok)
37837 (s390_canonicalize_comparison, s390_extract_part)
37838 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37839 (s390_contiguous_bitmask_p, s390_rtx_costs)
37840 (legitimize_pic_address): Likewise.
37841 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37842 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37843 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37844 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37845 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37847 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37849 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37850 REGNO($0) == REGNO($1).
37852 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37854 * config/s390/linux.h(SIZE_TYPE): Add comment.
37856 2017-02-06 Julian Brown <julian@codesourcery.com>
37857 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37858 Virendra Pathak <virendra.pathak@broadcom.com>
37860 * config/aarch64/aarch64-cores.def: Change the scheduler
37862 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37863 * config/aarch64/thunderx2t99.md: New file.
37865 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37867 * doc/standards.texi (Go Language): Update link to language
37870 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37872 * tree-eh.c (lower_resx): Sanitize profile.
37873 (cleanup_empty_eh_move_lp): Likewise.
37875 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37878 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37880 * cfgloopmanip.h (loop_version): Update prototype.
37881 * modulo-sched.c (sms_schedule): Update call of loop_version.
37882 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37883 * tree-parloops.c (gen_parallel_loop): Likewise.
37884 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37885 * tree-ssa-loop-split.c (split_loop): Likewise.
37886 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37887 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37889 2017-02-05 Martin Liska <mliska@suse.cz>
37892 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37894 (print_operand_address): Initialize a struct to zero.
37896 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37898 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37899 garbage collector only in textual form.
37901 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37903 * doc/extend.texi (x86 specific memory model extensions for
37904 transactional memory): Simplify a phrase.
37906 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37909 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37910 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37911 (atomic_storedi_1): Likewise.
37913 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37915 PR tree-optimization/79338
37916 * tree-parloops.c (gather_scalar_reductions): Don't call
37917 vect_analyze_loop_form for loop->inner before destroying loop's
37920 2017-02-03 Martin Sebor <msebor@redhat.com>
37922 PR tree-optimization/79327
37923 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37924 when precision has resulted in leading zeros.
37925 (format_integer): Adjust the likely counter to assume an unknown
37926 argument that may be zero is non-zero.
37928 2017-02-03 Jason Merrill <jason@redhat.com>
37931 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37932 avoid copying non-taken branch.
37934 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37936 PR tree-optimization/79340
37937 * tree-vect-loop.c (vectorizable_reduction): Release
37938 vec_defs elements after safe_splicing them into other vectors.
37941 PR tree-optimization/79327
37942 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37943 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37945 (format_integer): Use wide_int_to_tree instead of build_int_cst
37946 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37947 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37948 of shortest and longest sequence.
37950 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37952 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37953 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37955 2017-02-03 Walter Lee <walt@tilera.com>
37958 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37959 after initial stackframe link reg save.
37960 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37962 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37965 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37966 wu for stxssp alternative.
37968 2017-02-03 Martin Sebor <msebor@redhat.com>
37970 PR tree-optimization/79352
37971 * gimple-fold.c (get_range_strlen): Add argument.
37972 (get_range_strlen): Change return type to bool.
37973 (get_maxval_strlen): Pass in a dummy argument.
37974 * gimple-fold.h (get_range_strlen): Change return type to bool.
37975 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37976 * tree.h (array_at_struct_end_p): Add argument.
37977 * tree.c (array_at_struct_end_p): Handle it.
37979 2017-02-03 Martin Liska <mliska@suse.cz>
37982 * multiple_target.c (create_dispatcher_calls): Redirect edge
37983 from a caller of a dispatcher.
37984 (expand_target_clones): Make the clones local.
37985 (ipa_target_clone): Do both target clones and resolvers.
37986 (ipa_dispatcher_calls): Remove the pass.
37987 (pass_dispatcher_calls::gate): Likewise.
37988 (make_pass_dispatcher_calls): Likewise.
37989 * passes.def (pass_target_clone): Put as very first IPA early
37992 2017-02-03 Martin Liska <mliska@suse.cz>
37994 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37995 in case of a function with ifunc attribute.
37997 2017-02-03 Martin Liska <mliska@suse.cz>
37999 * cgraph.c (cgraph_node::dump): Dump function version info.
38000 * symtab.c (symtab_node::dump_base): Add missing new line.
38002 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38004 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38005 (ifcombine_ifandif): Use it.
38007 2017-02-03 Martin Liska <mliska@suse.cz>
38009 * doc/invoke.texi: Document default value for
38010 use-after-scope-direct-emission-threshold.
38012 2017-02-03 Martin Liska <mliska@suse.cz>
38014 PR tree-optimization/79339
38015 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38016 (format_floating): Likewise.
38018 2017-02-03 Martin Liska <mliska@suse.cz>
38021 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38023 (ipa_node_params_t::remove): Likewise.
38024 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38025 initialization from removed ipa_node_params_t::insert.
38026 (ipa_node_params::~ipa_node_params): Move from removed
38027 ipa_node_params_t::release.
38028 * symbol-summary.h (symbol_summary::m_released): New member.
38029 Do not release a summary twice. Do not allow to call finalizer
38030 for types of a summary that live in GGC memory.
38032 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38034 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38037 2017-02-02 Martin Sebor <msebor@redhat.com>
38039 PR middle-end/79275
38040 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38041 (format_string): Tighten up the range of output for non-constant
38042 strings and correct the expected range for wide non-constant strings.
38044 2017-02-02 Martin Sebor <msebor@redhat.com>
38046 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38048 PR middle-end/32003
38049 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38051 (-fdump-tree-@var): Add to index and document how to come up
38052 with pass-specific option and dump file names.
38053 (-fdump-passes): Clarify where to look for output.
38055 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38057 PR middle-end/77445
38058 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38059 statistics of the analyzed path; allow threading for speed when
38060 any of BBs along the path are optimized for speed.
38062 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38064 PR middle-end/78468
38065 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38066 settings of the virtual registers.
38069 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38071 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38072 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38075 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38077 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38078 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38080 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38082 * config/s390/s390.md: Add missing comments with the expanded
38084 * config/s390/vector.md: Likewise.
38085 * config/s390/vx-builtins.md: Likewise.
38087 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38090 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38091 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38092 conditions on a single line.
38094 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38096 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38097 __S390_VX__ to __VX__.
38099 2017-02-01 Andrew Pinski <apinski@cavium.com>
38101 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38102 stmt_info to record_stmt_cost.
38103 (vect_get_known_peeling_cost): Pass stmt_info if known to
38105 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38106 cpu_vector_cost field into
38107 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38108 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38109 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38110 splitting of scalar_stmt_cost and vec_stmt_cost.
38111 (thunderx_vector_cost): Likewise.
38112 (cortexa57_vector_cost): LIkewise.
38113 (exynosm1_vector_cost): Likewise.
38114 (xgene1_vector_cost): Likewise.
38115 (thunderx2t99_vector_cost): Improve after the splitting of the two
38117 (aarch64_builtin_vectorization_cost): Update for the splitting of
38118 scalar_stmt_cost and vec_stmt_cost.
38120 2017-02-01 Torvald Riegel <triegel@redhat.com>
38121 Richard Henderson <rth@redhat.com>
38123 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38124 conditional on existance of a fast atomic load.
38125 * optabs-query.c (can_atomic_load_p): New function.
38126 * optabs-query.h (can_atomic_load_p): Declare it.
38127 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38128 no fast atomic load is available for the particular size of access.
38129 (expand_atomic_compare_and_swap): Likewise.
38130 (expand_atomic_load): Likewise.
38131 (expand_atomic_store): Likewise.
38132 (expand_atomic_fetch_op): Likewise.
38133 * testsuite/lib/target-supports.exp
38134 (check_effective_target_sync_int_128): Remove x86 because it provides
38135 no fast atomic load.
38136 (check_effective_target_sync_int_128_runtime): Likewise.
38138 2017-02-01 Richard Biener <rguenther@suse.de>
38140 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38141 (graphite_transform_loops): Provide opt-info for optimized nests.
38142 * tree-parloop.c (parallelize_loops): Provide opt-info for
38143 parallelized loops.
38145 2017-02-01 Richard Biener <rguenther@suse.de>
38147 PR middle-end/79315
38148 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38149 was not set before.
38151 2017-02-01 Richard Biener <rguenther@suse.de>
38153 PR tree-optimization/71824
38154 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38155 Verify the loops are valid in the merged SESE region.
38156 (scop_detection::can_represent_loop_1): Check analyzing the
38157 evolution of the number of iterations in the region succeeds.
38159 2017-01-31 Ian Lance Taylor <iant@golang.org>
38161 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38162 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38164 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38166 PR preprocessor/79210
38167 * input.c (get_substring_ranges_for_loc): Replace line_width
38168 assertion with error-handling.
38170 2017-01-31 Richard Biener <rguenther@suse.de>
38172 PR tree-optimization/77318
38173 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38174 (create_pw_aff_from_tree): Take loop parameter.
38175 (add_condition_to_pbb): Pass loop of the condition to
38176 create_pw_aff_from_tree.
38178 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38180 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38181 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38183 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38187 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38189 (convert_int_to_float128): Likewise.
38190 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38191 (convert_int_to_float128): Likewise.
38192 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38193 (UNSPEC_IEEE128_CONVERT): Likewise.
38194 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38195 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38196 Use local variables for IBM extended format.
38197 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38198 (fix_trunc<mode>si2_fprs): Likewise.
38199 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38200 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38201 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38202 to know that we can now have integers of all sizes in vector
38204 (fix<uns>_<mode>di2_hw): Likewise.
38205 (float<uns>_<mode>si2_hw): Likewise.
38206 (fix_<mode>si2_hw): Likewise.
38207 (fixuns_<mode>si2_hw): Likewise.
38208 (float<uns>_<mode>di2_hw): Likewise.
38209 (float_<mode>di2_hw): Likewise.
38210 (float_<mode>si2_hw): Likewise.
38211 (floatuns_<mode>di2_hw): Likewise.
38212 (floatuns_<mode>si2_hw): Likewise.
38213 (xscvqp<su>wz_<mode>): Delete, no longer used.
38214 (xscvqp<su>dz_<mode>): Likewise.
38215 (xscv<su>dqp_<mode>): Likewise.
38216 (ieee128_mfvsrd_64bit): Likewise.
38217 (ieee128_mfvsrd_32bit): Likewise.
38218 (ieee128_mfvsrwz): Likewise.
38219 (ieee128_mtvsrw): Likewise.
38220 (ieee128_mtvsrd_64bit): Likewise.
38221 (ieee128_mtvsrd_32bit): Likewise.
38223 2017-01-31 Martin Liska <mliska@suse.cz>
38226 * ipa-prop.c (ipa_free_all_node_params): Call release method
38227 instead of ~sumbol_summary to not to trigger double times
38230 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38232 PR tree-optimization/71691
38233 * bitmap.h (class auto_bitmap): New.
38234 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38235 is_maybe_undefined instead of ssa_undefined_value_p.
38237 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38239 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38240 __S390_ARCH_LEVEL__ to __ARCH__.
38242 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38244 PR tree-optimization/79267
38245 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38246 if should_remove_lhs_p is true.
38248 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38251 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38252 (add_alignment_attribute): New.
38253 (base_type_die): Add alignment attribute.
38254 (subrange_type_die): Likewise.
38255 (modified_type_die): Likewise.
38256 (gen_array_type_die): Likewise.
38257 (gen_descr_array_type_die: Likewise.
38258 (gen_enumeration_type_die): Likewise.
38259 (gen_subprogram_die): Likewise.
38260 (gen_variable_die): Likewise.
38261 (gen_field_die): Likewise.
38262 (gen_ptr_to_mbr_type_die): Likewise.
38263 (gen_struct_or_union_type_die): Likewise.
38264 (gen_subroutine_type_die): Likewise.
38265 (gen_typedef_die): Likewise.
38266 (base_type_cmp): Compare alignment attribute.
38268 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38271 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38272 (setb_unsigned) New pattern for setb with CCUNS.
38273 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38274 subfc./subfe sequence to avoid overflow problems. Generate a
38275 shorter sequence with cmpld/setb for power9.
38276 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38277 for generating subfc. instruction.
38278 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38279 now uses this instruction.
38281 2017-01-30 Ian Lance Taylor <iant@google.com>
38284 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38285 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38287 2017-01-30 Martin Sebor <msebor@redhat.com>
38289 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38290 Move constant to the right of a relational operator.
38291 (get_mpfr_format_length, format_character, format_string): Ditto.
38292 (should_warn_p, maybe_warn): Same.
38294 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38296 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38299 * asan.c (get_translation_unit_decl): Remove function.
38300 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38302 2017-01-30 Martin Liska <mliska@suse.cz>
38304 PR gcov-profile/79259
38305 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38306 -fprofile-generate.
38308 2017-01-30 Martin Liska <mliska@suse.cz>
38311 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38312 Initialize variables with NULL value.
38314 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38317 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38319 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38321 2017-01-30 Richard Biener <rguenther@suse.de>
38323 PR tree-optimization/79276
38324 * tree-vrp.c (process_assert_insertions): Properly adjust common
38325 when removing a duplicate.
38327 2017-01-30 Richard Biener <rguenther@suse.de>
38329 PR tree-optimization/79256
38330 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38331 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38333 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38335 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38338 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38339 ("*r<noxa>sbg_<mode>_sll_bitmask")
38340 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38341 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38342 Use contiguous_bitmask_nowrap_operand.
38344 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38347 * config/rs6000/altivec.h (vec_xl): Revise #define.
38348 (vec_xst): Likewise.
38350 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38352 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38354 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38356 PR rtl-optimization/79194
38357 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38358 traps before call to bypass_conditional_jumps.
38360 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38362 PR tree-optimization/71374
38363 * lra-constraints.c (check_conflict_input_operands): New.
38364 (match_reload): Use it.
38366 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38369 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38370 account to calculate conflict_set.
38372 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38374 PR rtl-optimization/78559
38375 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38376 other_insn in combine.
38378 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38380 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38381 uint16_type_node for BT_UINT16.
38383 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38385 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38386 "RTL Tests" to menu.
38387 (GIMPLE Tests): New node.
38388 (RTL Tests): New node.
38390 2017-01-27 Richard Biener <rguenther@suse.de>
38392 PR tree-optimization/79245
38393 * tree-loop-distribution.c (distribute_loop): Apply cost
38394 modeling also to detected patterns.
38396 2017-01-27 Richard Biener <rguenther@suse.de>
38398 PR tree-optimization/71433
38399 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38400 (compare_assert_loc): New function.
38401 (process_assert_insertions): Sort and optimize assert locations
38402 to remove duplicates and push down identical assertions on
38403 edges to their destination block.
38405 2017-01-27 Richard Biener <rguenther@suse.de>
38407 PR tree-optimization/79244
38408 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38409 out SSA names even if abnormal.
38411 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38413 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38414 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38415 instead of MPFR_RNDN.
38417 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38420 * arm.c (arm_option_override): Don't call build_target_option_node
38421 until after doing all option overrides.
38422 (arm_valid_target_attribute_tree): Likewise.
38424 2017-01-27 Martin Liska <mliska@suse.cz>
38426 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38427 for {cd}tors and C++ {cd}tors.
38429 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38431 * config/s390/s390.md ("*setmem_long_and")
38432 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38434 2017-01-26 Martin Sebor <msebor@redhat.com>
38436 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38439 2017-01-26 Martin Sebor <msebor@redhat.com>
38441 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38442 HAVE_DFmode before using XFmode or DFmode.
38443 (parse_directive): Avoid using the z length modifier to avoid
38444 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38446 PR middle-end/78703
38447 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38448 to accept adjustment as an array.
38449 (get_int_range): New function.
38450 (struct directive): Make width and prec arrays.
38451 (directive::set_width, directive::set_precision): Call get_int_range.
38452 (format_integer, format_floating): Handle width and precision ranges.
38453 (format_string, parse_directive): Same.
38455 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38458 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38459 comdat_type_p set, just clone them, but keep the children in the
38463 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38464 which have direct callers with -fvar-tracking-assignments enabled
38466 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38467 inside of type units.
38469 2017-01-26 Martin Sebor <msebor@redhat.com>
38471 PR middle-end/78703
38472 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38474 (struct format_result): Replace number_chars, number_chars_min,
38475 and number_chars_max with a single member of struct result_range.
38477 (format_result::operator+=): Adjust.
38478 (struct fmtresult): Remove bounded. Handle likely and unlikely
38480 (fmtresult::adjust_for_width_or_precision): New function.
38481 (fmtresult:type_max_digits): New function.
38482 (bytes_remaining): Handle likely and unlikely counters.
38483 (min_bytes_remaining): Remove.
38484 (format_percent): Simplify.
38485 (format_integer, format_floating): Set likely and unlikely counters.
38486 (get_string_length, format_character, format_string): Same.
38487 (format_plain, should_warn_p): New function.
38488 (maybe_warn): Call should_warn_p. Update diagnostic messages
38489 and handle those for all directives, including plain strings.
38490 (format_directive): Handle likely and unlikely counters.
38491 Remove unnecessary quoting from diagnostics. Add an informational
38493 (add_bytes): Remove.
38494 (pass_sprintf_length::compute_format_length): Simplify.
38495 (try_substitute_return_value): Handle likely and unlikely counters.
38497 2017-01-26 Carl Love <cel@us.ibm.com>
38499 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38500 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38502 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38505 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38506 endianess for subregs into account.
38507 * lra-constraints.c (lra_constraints): Do risky transformations
38508 always on the first iteration.
38509 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38510 last_call_used_reg_set.
38511 (process_bb_lives): Define and use last_call_used_reg_set.
38512 * lra.c (lra): Always continue after lra_constraints on the first
38515 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38517 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38519 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38521 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38523 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38524 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38525 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38526 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38527 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38528 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38529 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38530 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38531 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38533 2017-01-26 Marek Polacek <polacek@redhat.com>
38536 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38537 for the third operand.
38539 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38541 PR middle-end/79236
38542 * omp-low.c (struct omp_context): Add simt_stmt field.
38543 (scan_omp_for): Return omp_context *.
38544 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38545 context to the _simt_ SIMD stmt.
38546 (lower_omp_for): For combined SIMD with sibling _simt_
38547 SIMD, make sure to use the same decls in _looptemp_
38548 clauses as in the sibling.
38550 2017-01-26 David Sherwood <david.sherwood@arm.com>
38552 PR middle-end/79212
38553 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38556 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38559 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38560 emit fld b; fld a; if possible.
38562 * brig-builtins.def: Update copyright years.
38563 * config/arm/arm_acle_builtins.def: Update copyright years.
38565 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38568 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38569 constraint instead of o for the stxsd instruction.
38571 2017-01-25 Carl Love <cel@us.ibm.com>
38573 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38574 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38576 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38578 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38580 2017-01-25 Richard Biener <rguenther@suse.de>
38582 PR tree-optimization/69264
38583 * target.def (vector_alignment_reachable): Improve documentation.
38584 * doc/tm.texi: Regenerate.
38585 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38587 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38588 earlier changes with respect to TYPE_USER_ALIGN.
38589 (vector_alignment_reachable_p): Likewise. Improve dumping.
38591 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38594 * config/arm/arm.md (xordi3): Force constant operand into a register
38597 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38599 * doc/invoke.texi (-fstore-merging): Correct default optimization
38600 levels at which it is enabled.
38601 (-O): Move -fstore-merging from list to...
38604 2017-01-25 Richard Biener <rguenther@suse.de>
38607 * omp-expand.c: Include debug.h.
38608 (expand_omp_taskreg): Make sure to generate early debug before
38609 outlining anything from a function.
38610 (expand_omp_target): Likewise.
38611 (grid_expand_target_grid_body): Likewise.
38613 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38616 * asan.c (get_translation_unit_decl): New function.
38617 (asan_add_global): Extract modules file name from globals
38618 TRANSLATION_UNIT_DECL name.
38620 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38623 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38624 for long calls with APCS frame and VFP.
38626 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38628 * cfg.c (original_copy_tables_initialized_p): New function.
38629 * cfg.h (original_copy_tables_initialized_p): New decl.
38630 * cfgrtl.c (relink_block_chain): Guard the call to
38631 free_original_copy_tables with a call to
38632 original_copy_tables_initialized_p.
38633 * cgraph.h (symtab_node::native_rtl_p): New decl.
38634 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38635 (symtab_node::needed_p): Don't assert for early assembly output
38636 for __RTL functions.
38637 (cgraph_node::finalize_function): Set "force_output" for __RTL
38639 (cgraph_node::analyze): Bail out early for __RTL functions.
38640 (analyze_functions): Update assertion to support __RTL functions.
38641 (cgraph_node::expand): Bail out early for __RTL functions.
38642 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38644 * function.h (struct function): Update comment for field
38646 * gimple-expr.c: Include "tree-pass.h".
38647 (gimple_has_body_p): Return false for __RTL functions.
38648 * Makefile.in (OBJS): Add run-rtl-passes.o.
38649 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38651 (gcc::pass_manager::get_clean_slate): New accessor.
38652 * passes.c: Include "insn-addr.h".
38653 (should_skip_pass_p): Add logging. Update logic for running
38654 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38655 property-provider override so it is only done for gimple passes.
38657 (skip_pass): New function.
38658 (execute_one_pass): Call skip_pass when skipping passes.
38659 * read-md.c (md_reader::read_char): Support filtering
38660 the input to a subset of line numbers.
38661 (md_reader::md_reader): Initialize fields
38662 m_first_line and m_last_line.
38663 (md_reader::read_file_fragment): New function.
38664 * read-md.h (md_reader::read_file_fragment): New decl.
38665 (md_reader::m_first_line): New field.
38666 (md_reader::m_last_line): New field.
38667 * read-rtl-function.c (function_reader::create_function): Only
38668 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38669 curr_properties. Set DECL_INITIAL to a dummy block.
38670 (read_rtl_function_body_from_file_range): New function.
38671 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38673 * run-rtl-passes.c: New file.
38674 * run-rtl-passes.h: New file.
38676 2017-01-24 Jeff Law <law@redhat.com>
38678 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38681 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38683 PR tree-optimization/79159
38684 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38685 (record_nonwrapping_iv): Improve boundary using above function if no
38686 value range information.
38688 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38689 Martin Jambor <mjambor@suse.cz>
38691 * brig-builtins.def: New file.
38692 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38693 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38694 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38695 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38696 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38697 * builtin-types.def (BT_INT8): New.
38698 (BT_INT16): Likewise.
38699 (BT_UINT8): Likewise.
38700 (BT_UINT16): Likewise.
38701 (BT_FN_ULONG): Likewise.
38702 (BT_FN_UINT_INT): Likewise.
38703 (BT_FN_UINT_ULONG): Likewise.
38704 (BT_FN_UINT_LONG): Likewise.
38705 (BT_FN_UINT_PTR): Likewise.
38706 (BT_FN_ULONG_PTR): Likewise.
38707 (BT_FN_INT8_FLOAT): Likewise.
38708 (BT_FN_INT16_FLOAT): Likewise.
38709 (BT_FN_UINT32_FLOAT): Likewise.
38710 (BT_FN_UINT16_FLOAT): Likewise.
38711 (BT_FN_UINT8_FLOAT): Likewise.
38712 (BT_FN_UINT64_FLOAT): Likewise.
38713 (BT_FN_UINT16_UINT32): Likewise.
38714 (BT_FN_UINT32_UINT16): Likewise.
38715 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38716 (BT_FN_INT_PTR_INT): Likewise.
38717 (BT_FN_UINT_PTR_UINT): Likewise.
38718 (BT_FN_LONG_PTR_LONG): Likewise.
38719 (BT_FN_ULONG_PTR_ULONG): Likewise.
38720 (BT_FN_VOID_UINT64_UINT64): Likewise.
38721 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38722 (BT_FN_INT8_INT8_INT8): Likewise.
38723 (BT_FN_INT16_INT16_INT16): Likewise.
38724 (BT_FN_INT_INT_INT): Likewise.
38725 (BT_FN_UINT_FLOAT_UINT): Likewise.
38726 (BT_FN_FLOAT_UINT_UINT): Likewise.
38727 (BT_FN_ULONG_UINT_UINT): Likewise.
38728 (BT_FN_ULONG_UINT_PTR): Likewise.
38729 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38730 (BT_FN_UINT_UINT_UINT): Likewise.
38731 (BT_FN_VOID_UINT_PTR): Likewise.
38732 (BT_FN_UINT_UINT_PTR: Likewise.
38733 (BT_FN_UINT32_UINT64_PTR): Likewise.
38734 (BT_FN_INT_INT_UINT_UINT): Likewise.
38735 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38736 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38737 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38738 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38739 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38740 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38741 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38742 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38743 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38744 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38745 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38746 * doc/frontends.texi: List BRIG FE.
38747 * doc/install.texi (Testing): Add BRIG tesring requirements.
38748 * doc/invoke.texi (Overall Options): Mention BRIG.
38749 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38751 2017-01-24 Richard Biener <rguenther@suse.de>
38753 PR translation/79208
38754 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38756 2017-01-24 Martin Jambor <mjambor@suse.cz>
38759 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38760 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38761 and known_contexts.
38763 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38765 PR middle-end/79123
38766 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38767 casts from signed to unsigned really don't have a range.
38769 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38771 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38772 GMP_RNDx for compatiblity.
38774 2017-01-24 Martin Liska <mliska@suse.cz>
38777 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38778 that would prevent us to call alloca with -1 as argument.
38780 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38782 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38783 Avoid -Wformat-security warning.
38785 2017-01-23 Andrew Pinski <apinski@cavium.com>
38787 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38790 2017-01-23 Martin Sebor <msebor@redhat.com>
38792 PR middle-end/78703
38793 * gimple-ssa-sprintf.c (warn_level): New global.
38794 (format_integer): Use it here and throughout the rest of the file.
38795 Use the same switch to compute sign as base.
38796 (maybe_warn): New function.
38797 (format_directive): Factor out warnings into maybe_warn.
38798 Add debugging output. Use warn_level.
38799 (add_bytes): Use warn_level.
38800 (pass_sprintf_length::compute_format_length): Add debugging output.
38801 (try_substitute_return_value): Same.
38802 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38804 PR middle-end/78703
38805 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38806 (struct fmtresult, format_integer, format_floating): Adjust.
38807 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38808 (get_string_length, format_string,format_directive): Same.
38809 (pass_sprintf_length::compute_format_length): Same.
38810 (try_substitute_return_value): Simplify slightly.
38812 PR middle-end/78703
38813 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38814 (fmtresult::operator+=): Outlined.
38815 (struct fmtresult): Add ctors.
38816 (struct conversion_spec): Rename...
38817 (struct directive): ...to this. Add and remove data members.
38818 (directive::set_width, directive::set_precision): New functions.
38819 (format_percent): Use fmtresult ctor.
38820 (get_width_and_precision): Remove.
38821 (format_integer): Make naming changes. Avoid computing width and
38823 (format_floating): Same. Adjust indentation.
38824 (format_character, format_none): New functions.
38825 (format_string): Moved character handling to format_character.
38826 (format_directive): Remove arguments, change return type.
38827 (parse_directive): New function.
38828 (pass_sprintf_length::compute_format_length): Move directive
38829 parsing to parse_directive.
38831 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38833 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38834 (assign_assembler_name_if_needed): ... this.
38835 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38836 (assign_assembler_name_if_needed): ... this.
38837 (free_lang_data_in_cgraph): Adjust callers.
38838 * cgraphunit.c (cgraph_node::analyze): Likewise.
38839 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38842 2017-01-23 Richard Biener <rguenther@suse.de>
38844 PR tree-optimization/79088
38845 PR tree-optimization/79188
38846 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38847 resetting loop bounds after last path deletion. Reset loop
38848 bounds of the target loop, make code match the comments.
38849 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38850 Make sure loops need no fixups.
38852 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38854 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38855 exponent support with double type for first argument.
38856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38857 type returned by __builtin_vec_extract_sig,
38858 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38859 functions from "vector int" to "vector unsigned int" or from
38860 "vector long long int" to "vector unsigned long long int".
38861 Changed type returned by __builtin_vec_extract_exp,
38862 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38863 functions from "vector int" to "vector unsigned int" or from
38864 "vector long long int" to "vector unsigned long long int".
38865 Changed return type of __builtin_vec_test_data_class,
38866 __builtin_vec_test_data_class_sp, and
38867 __builtin_vec_test_data_class_dp from "vector int" to
38868 "vector bool int" or from "vector long long int" to "vector bool
38869 long long int" and changed second argument type from "unsigned
38870 int" to "int". Added new overloaded function forms "vector float
38871 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38872 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38873 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38874 double, vector unsigned long long int)" and "vector double
38875 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38876 long int)". Changed return type of
38877 __builtin_scalar_test_data_class and
38878 __builtin_scalar_test_data_class_sp and
38879 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38880 int" and changed second argument from "unsigned int" to "int".
38881 Changed type returned by __builtin_scalar_test_neg,
38882 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38883 from "int" to "bool int". Added new overloaded function form
38884 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38885 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38886 exponent double-precision with floating point first argument.
38887 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38888 documentation of scalar_test_data_class, scalar_test_neg,
38889 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38890 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38891 vec_test_data_class built-in functions to reflect refinements in
38892 their type signatures.
38894 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38896 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38898 (aarch64_elf_asm_destructor): Likewise.
38900 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38902 PR rtl-optimization/78634
38903 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38904 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38905 * ifcvt.c (noce_try_cmove): Add missing cost check.
38907 PR rtl-optimization/71724
38908 * combine.c (if_then_else_cond): Look for situations where it is
38909 beneficial to undo the work of one of the recursive calls.
38911 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38913 PR tree-optimization/70754
38914 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38915 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38916 combined stmt before it if not NULL.
38917 (combine_chains): Process refs reversely and compute dominance point
38920 2017-01-23 Martin Liska <mliska@suse.cz>
38922 PR tree-optimization/79196
38923 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38924 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38926 (strlen_optimize_stmt): Call the renamed function.
38928 2017-01-23 Michael Matz <matz@suse.de>
38930 PR tree-optimization/78384
38931 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38933 2017-01-23 Richard Biener <rguenther@suse.de>
38935 PR tree-optimization/79186
38936 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38937 both incoming edges before moving an assert.
38939 2017-01-23 Martin Jambor <mjambor@suse.cz>
38941 * ipa-prop.c (load_from_param_1): Removed.
38942 (load_from_unmodified_param): Bits from load_from_param_1 put back
38944 (load_from_param): Removed.
38945 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38946 with stmt. Reverted back to use of load_from_unmodified_param.
38948 2017-01-23 Martin Jambor <mjambor@suse.cz>
38951 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38952 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38953 field a pointer to garbage collected vector, mark lattices and
38954 ipcp_orig_node with GTY((skip)).
38955 (ipa_get_param_count): Adjust to descriptors being a pointer.
38956 (ipa_get_param): Likewise.
38957 (ipa_get_type): Likewise.
38958 (ipa_get_param_move_cost): Likewise.
38959 (ipa_set_param_used): Likewise.
38960 (ipa_get_controlled_uses): Likewise.
38961 (ipa_set_controlled_uses): Likewise.
38962 (ipa_is_param_used): Likewise.
38963 (ipa_node_params_t): Move into garbage collector. New methods insert
38965 (ipa_node_params_sum): Annotate wth GTY(()).
38966 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38968 (ipa_load_from_parm_agg): Adjust declaration.
38969 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38970 * ipa-profile.c (ipa_profile): Likewise.
38971 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38972 (ipa_populate_param_decls): Make descriptors parameter garbage
38974 (ipa_dump_param): Adjust to descriptors being a pointer.
38975 (ipa_alloc_node_params): Likewise.
38976 (ipa_initialize_node_params): Likewise.
38977 (load_from_param_1): Make descriptors parameter garbage collected.
38978 (load_from_unmodified_param): Likewise.
38979 (load_from_param): Likewise.
38980 (ipa_load_from_parm_agg): Likewise.
38981 (ipa_node_params::~ipa_node_params): Removed.
38982 (ipa_free_all_node_params): Remove call to delete operator.
38983 (ipa_node_params_t::insert): New.
38984 (ipa_node_params_t::remove): Likewise.
38985 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38986 copy known_csts and known_contexts vectors.
38987 (ipa_read_node_info): Adjust to descriptors being a pointer.
38988 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38990 (ipcp_transform_function): Make descriptors variable garbage
38993 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38995 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38996 * config/i386/avx512dqintrin.h: Ditto.
38997 * config/i386/avx512fintrin.h: Ditto.
38998 * config/i386/i386.c: Handle new builtins.
38999 * config/i386/i386-builtin.def: Add new builtins.
39000 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39001 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39003 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39004 Martin Liska <mliska@suse.cz>
39006 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39007 * asan.c (asan_expand_poison_ifn): Support stores and use
39008 appropriate ASAN report function.
39009 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39010 * internal-fn.def (ASAN_POISON_USE): Declare.
39011 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39012 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39013 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39014 ASAN_POISON calls w/o LHS.
39015 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39016 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39017 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39018 * gimplify.c (asan_poison_variables): Add attribute
39019 use_after_scope_memory to variables that really needs to live
39021 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39022 having the attribute.
39024 2017-01-23 Martin Liska <mliska@suse.cz>
39026 * asan.c (create_asan_shadow_var): New function.
39027 (asan_expand_poison_ifn): Likewise.
39028 * asan.h (asan_expand_poison_ifn): New declaration.
39029 * internal-fn.c (expand_ASAN_POISON): Likewise.
39030 * internal-fn.def (ASAN_POISON): New builtin.
39031 * sanopt.c (pass_sanopt::execute): Expand
39032 asan_expand_poison_ifn.
39033 * tree-inline.c (copy_decl_for_dup_finish): Make function
39035 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39036 * tree-ssa.c (is_asan_mark_p): New function.
39037 (execute_update_addresses_taken): Rewrite local variables
39038 (identified just by use-after-scope as addressable) into SSA.
39040 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39042 * doc/install.texi (Specific): opensource.apple.com uses https
39043 now. Remove trailing slash.
39045 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39047 * README.Portability: Remove note on an Irix compatibility issue.
39049 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39051 * gcov.c (INCLUDE_ALGORITHM): Define.
39052 (INCLUDE_VECTOR): Define.
39053 No longer include <vector> and <algorithm> directly.
39055 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39057 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39059 * doc/invoke.texi (Code Gen Options): Ditto.
39061 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39064 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39066 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39068 rtl-optimization/79125
39069 * cprop.c (local_cprop_pass): Handle cases where we make an
39070 unconditional trap.
39072 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39076 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39077 read from, for big endian.
39079 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39081 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39082 register pauth builtins for LP64 only.
39084 2017-01-20 Marek Polacek <polacek@redhat.com>
39087 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39090 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39092 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39094 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39095 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39096 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39098 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39101 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39102 in big-endian mode when they are not a single duplicated value.
39104 2017-01-20 Richard Biener <rguenther@suse.de>
39106 * BASE-VER: Bump to 7.0.1.
39108 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39110 * omp-low.c (omplow_simd_context): New struct. Use it...
39111 (lower_rec_simd_input_clauses): ...here and...
39112 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39113 references to idx, lane, max_vf, is_simt.
39115 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39117 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39120 2017-01-20 Martin Jambor <mjambor@suse.cz>
39122 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39123 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39125 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39126 (GTFILES): Rename hsa.c to hsa-common.c.
39127 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39128 * hsa-dump.c: Likewise.
39129 * hsa-gen.c: Likewise.
39130 * hsa-regalloc.c: Likewise.
39131 * ipa-hsa.c: Likewise.
39132 * omp-expand.c: Likewise.
39133 * omp-low.c: Likewise.
39134 * toplev.c: Likewise.
39136 2017-01-20 Marek Polacek <polacek@redhat.com>
39139 * doc/invoke.texi: Document -Wduplicated-branches.
39140 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39141 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39142 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39143 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39144 return 0 only when not OEP_LEXICOGRAPHIC.
39145 (fold_build_cleanup_point_expr): Use the expression
39146 location when building CLEANUP_POINT_EXPR.
39147 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39148 * tree.c (add_expr): Handle error_mark_node.
39150 2017-01-20 Martin Liska <mliska@suse.cz>
39153 * tree-profile.c (init_ic_make_global_vars): Do not call
39155 (gimple_init_gcov_profiler): Likewise.
39157 2017-01-20 Martin Liska <mliska@suse.cz>
39160 * cgraph.h (maybe_create_reference): Remove argument and
39162 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39164 * ipa-cp.c (create_specialized_node): Likewise.
39165 * symtab.c (symtab_node::maybe_create_reference): Handle
39166 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39168 2017-01-20 Martin Liska <mliska@suse.cz>
39170 * read-rtl-function.c (function_reader::create_function): Use
39171 build_decl instread of build_decl_stat.
39173 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39175 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39176 * config/i386/avx512dqintrin.h: Ditto.
39177 * config/i386/avx512fintrin.h: Ditto.
39178 * config/i386/i386-builtin-types.def: Add new types.
39179 * config/i386/i386.c: Handle new types.
39180 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39181 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39182 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39183 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39184 (__builtin_ia32_kshiftridi): New.
39185 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39187 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39191 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39192 define to rs6000_init_stack_protect_guard.
39193 (rs6000_init_stack_protect_guard): New function.
39195 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39196 Yunqiang Su <yunqiang.su@imgtec.com>
39198 * config.gcc (supported_defaults): Add madd4.
39199 (with_madd4): Add validation.
39200 (all_defaults): Add madd4.
39201 * config/mips/mips.opt (mmadd4): New option.
39202 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39204 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39206 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39207 (ISA_HAS_FUSED_MADD4): Likewise.
39208 * doc/invoke.texi (-mmadd4): Document the new option.
39209 * doc/install.texi (--with-madd4): Document the new option.
39211 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39213 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39214 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39215 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39216 (aarch64_init_pauth_hint_builtins): New.
39217 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39218 (aarch64_expand_builtin): Expand new builtins.
39220 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39222 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39223 * combine-stack-adj.c (no_unhandled_cfa): Handle
39224 REG_CFA_TOGGLE_RA_MANGLE.
39225 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39226 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39227 info for return address signing.
39228 (aarch64_expand_epilogue): Likewise.
39230 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39232 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39233 * config/aarch64/aarch64-protos.h
39234 (aarch64_return_address_signing_enabled): New declaration.
39235 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39237 (aarch64_expand_prologue): Sign return address before it's pushed onto
39239 (aarch64_expand_epilogue): Authenticate return address fetched from
39241 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39242 (aarch64_attributes): New function attributes for "sign-return-address".
39243 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39244 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39245 ("*do_return"): Generate combined instructions according to key index.
39246 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39247 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39249 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39250 * config/aarch64/aarch64.opt (msign-return-address=): New.
39251 * doc/extend.texi (AArch64 Function Attributes): Documents
39252 "sign-return-address=".
39253 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39255 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39257 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39258 overall option summary.
39260 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39262 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39263 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39264 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39265 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39267 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39269 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39270 -mpower9-minmax by default for -mcpu=power9.
39271 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39272 128-bit floating point.
39274 2017-01-20 Alan Modra <amodra@gmail.com>
39276 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39277 optimizing for size.
39279 2017-01-20 Alan Modra <amodra@gmail.com>
39282 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39283 for strcmp and strncmp from corresponding builtin decl.
39285 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39287 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39288 instead of i386/rtems-64.h.
39289 * config/i386/rtems-64.h: Remove.
39291 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39295 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39297 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39299 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39301 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39302 Change int to HOST_WIDE_INT.
39303 * config/aarch64/aarch64-protos.h
39304 (aarch64_simd_gen_const_vector_dup): Likewise.
39305 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39307 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39309 * langhooks-def.h (lhd_type_for_size): New decl.
39310 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39311 * langhooks.c (lhd_type_for_size): New function, taken from
39314 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39316 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39317 define_bypass for CR latency.
39318 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39319 (power9-alu2): Add define_bypass for CR latency.
39321 (power9-mul): Update insn latency.
39322 (power9-mul-compare): Update insn latency, bypass latency and remove
39325 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39327 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39329 * config/aarch64/aarch64.md
39330 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39331 aarch64_nopcrelative_literal_loads.
39332 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39334 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39336 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39337 TARGET_LOONGSON_3A.
39338 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39340 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39343 * config.gcc (supported_defaults): Add lxc1-sxc1.
39344 (with_lxc1_sxc1): Add validation.
39345 (all_defaults): Add lxc1-sxc1.
39346 * config/mips/mips.opt (mlxc1-sxc1): New option.
39347 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39349 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39350 __mips_no_lxc1_sxc1.
39351 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39352 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39353 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39355 2017-01-19 Richard Biener <rguenther@suse.de>
39357 PR tree-optimization/72488
39358 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39359 sure to restore SSA info.
39360 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39362 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39364 PR rtl-optimization/79121
39365 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39366 of the inner type when shifting an extended value.
39368 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39371 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39372 interposable aliases.
39374 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39377 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39378 Use the evmergelohi instruction.
39379 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39380 (mov_sitf_e500_subreg8_2_be): Likewise.
39381 (mov_sitf_e500_subreg12_2_le): Likewise.
39382 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39383 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39384 (mov_sitf_e500_subreg8_2_le): Likewise.
39385 (mov_sitf_e500_subreg12_2_be): Likewise.
39387 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39389 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39390 attribute from vecsimple to vecperm.
39391 (altivec_vbpermq2): Likewise.
39393 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39396 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39398 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39399 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39400 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39401 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39402 case where N arg is SIZE_MAX.
39403 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39404 (cmpstrsi): Add pattern.
39406 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39408 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39409 __builtin_vec_revb builtins.
39410 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39411 built-in functions to support generation of the ISA 3.0 XXBR<x>
39412 vector byte reverse instructions.
39413 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39414 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39415 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39416 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39417 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39418 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39419 (P9V_BUILTIN_VEC_REVB): Likewise.
39420 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39421 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39422 (p9_xxbrq_v16qi): Likewise.
39423 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39424 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39425 (p9_xxbrh_v8hi): Likewise.
39426 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39427 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39428 vec_revb built-in functions.
39430 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39432 PR rtl-optimization/78952
39433 * config/i386/i386.md (any_extract): New code iterator.
39434 (*insvqi_2): Use any_extract for source operand.
39435 (*insvqi_3): Use any_shiftrt for source operand.
39437 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39439 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39441 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39443 2017-01-18 Matthias Klose <doko@ubuntu.com>
39445 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39447 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39449 * config/rs6000/altivec.h (vec_bperm): Change #define.
39450 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39451 (altivec_vbpermq2): New define_insn.
39452 (altivec_vbpermd): Likewise.
39453 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39454 function interface.
39455 (VBPERMD): Likewise.
39456 (VBPERM): New polymorphic function interface.
39457 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39458 Add entries for P9V_BUILTIN_VEC_VBPERM.
39459 * doc/extend.texi: Add interfaces for vec_bperm.
39461 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39463 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39464 first letter of error messages.
39465 (s390_resolve_overloaded_builtin): Likewise.
39466 * config/s390/s390.c (s390_expand_builtin): Likewise.
39467 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39468 (s390_valid_target_attribute_inner_p): Likewise.
39469 * config/s390/s390.md ("tabort"): Likewise.
39471 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39473 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39474 (ISA_AVOID_DIV_HILO): New macro.
39475 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39476 (ISA_HAS_DDIV): Likewise.
39478 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39480 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39482 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39484 * doc/invoke.texi (fabi-version): Spelling fix.
39486 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39489 * doc/invoke.texi (fabi-version): Mention mangling fix for
39492 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39495 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39497 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39500 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39503 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39505 (stack_protect_test): Ditto.
39506 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39507 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39509 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39510 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39511 -mstack-protector-guard-offset=.
39512 (RS/6000 and PowerPC Options): Ditto.
39514 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39516 * config/i386/i386.h (MASK_CLASS_P): New define.
39517 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39518 there are no registers from different register sets also when
39519 mask registers are used. Update function comment.
39520 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39521 to (*k/*r) and (*k/*km) alternatives.
39523 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39525 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39526 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39527 (EH_RETURN_HANDLER_RTX): New define.
39528 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39529 Force frame pointer in EH return functions.
39530 (aarch64_expand_epilogue): Add barrier for eh_return.
39531 (aarch64_final_eh_return_addr): Remove.
39532 (aarch64_eh_return_handler_rtx): New function.
39533 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39535 (aarch64_eh_return_handler_rtx): New prototype.
39537 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39539 * config/rs6000/altivec.h (vec_rlmi): New #define.
39540 (vec_vrlnm): Likewise.
39541 (vec_rlnm): Likewise.
39542 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39543 (UNSPEC_VRLNM): Likewise.
39544 (VIlong): New mode iterator.
39545 (altivec_vrl<VI_char>mi): New define_insn.
39546 (altivec_vrl<VI_char>nm): Likewise.
39547 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39549 (VRLDNM): Likewise.
39550 (RLNM): New polymorphic function entry.
39551 (VRLWMI): New monomorphic function entry.
39552 (VRLDMI): Likewise.
39553 (RLMI): New polymorphic function entry.
39554 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39555 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39556 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39559 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39562 * dwarf2out.c (field_byte_offset): Restore the
39563 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39564 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39565 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39568 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39571 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39573 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39576 * dwarf2out.c (add_data_member_location_attribute): For constant
39577 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39578 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39579 DW_AT_byte_size attributes.
39581 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39583 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39584 after forcing to constant memory when the code model is medium.
39586 2017-01-17 Julia Koval <julia.koval@intel.com>
39589 * config/i386/avx512fintrin.h
39590 (_mm512_i32gather_ps): Change __addr type to void const*.
39591 (_mm512_mask_i32gather_ps): Ditto.
39592 (_mm512_i32gather_pd): Ditto.
39593 (_mm512_mask_i32gather_pd): Ditto.
39594 (_mm512_i64gather_ps): Ditto.
39595 (_mm512_mask_i64gather_ps): Ditto.
39596 (_mm512_i64gather_pd): Ditto.
39597 (_mm512_mask_i64gather_pd): Ditto.
39598 (_mm512_i32gather_epi32): Ditto.
39599 (_mm512_mask_i32gather_epi32): Ditto.
39600 (_mm512_i32gather_epi64): Ditto.
39601 (_mm512_mask_i32gather_epi64): Ditto.
39602 (_mm512_i64gather_epi32): Ditto.
39603 (_mm512_mask_i64gather_epi32): Ditto.
39604 (_mm512_i64gather_epi64): Ditto.
39605 (_mm512_mask_i64gather_epi64): Ditto.
39606 (_mm512_i32scatter_ps): Change __addr type to void*.
39607 (_mm512_mask_i32scatter_ps): Ditto.
39608 (_mm512_i32scatter_pd): Ditto.
39609 (_mm512_mask_i32scatter_pd): Ditto.
39610 (_mm512_i64scatter_ps): Ditto.
39611 (_mm512_mask_i64scatter_ps): Ditto.
39612 (_mm512_i64scatter_pd): Ditto.
39613 (_mm512_mask_i64scatter_pd): Ditto.
39614 (_mm512_i32scatter_epi32): Ditto.
39615 (_mm512_mask_i32scatter_epi32): Ditto.
39616 (_mm512_i32scatter_epi64): Ditto.
39617 (_mm512_mask_i32scatter_epi64): Ditto.
39618 (_mm512_i64scatter_epi32): Ditto.
39619 (_mm512_mask_i64scatter_epi32): Ditto.
39620 (_mm512_i64scatter_epi64): Ditto.
39621 (_mm512_mask_i64scatter_epi64): Ditto.
39622 * config/i386/avx512pfintrin.h
39623 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39624 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39625 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39626 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39627 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39628 (_mm512_prefetch_i32scatter_ps): Ditto.
39629 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39630 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39631 (_mm512_prefetch_i64scatter_pd): Ditto.
39632 (_mm512_prefetch_i64scatter_ps): Ditto.
39633 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39634 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39635 * config/i386/avx512vlintrin.h
39636 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39637 (_mm_mmask_i32gather_ps): Ditto.
39638 (_mm256_mmask_i32gather_pd): Ditto.
39639 (_mm_mmask_i32gather_pd): Ditto.
39640 (_mm256_mmask_i64gather_ps): Ditto.
39641 (_mm_mmask_i64gather_ps): Ditto.
39642 (_mm256_mmask_i64gather_pd): Ditto.
39643 (_mm_mmask_i64gather_pd): Ditto.
39644 (_mm256_mmask_i32gather_epi32): Ditto.
39645 (_mm_mmask_i32gather_epi32): Ditto.
39646 (_mm256_mmask_i32gather_epi64): Ditto.
39647 (_mm_mmask_i32gather_epi64): Ditto.
39648 (_mm256_mmask_i64gather_epi32): Ditto.
39649 (_mm_mmask_i64gather_epi32): Ditto.
39650 (_mm256_mmask_i64gather_epi64): Ditto.
39651 (_mm_mmask_i64gather_epi64): Ditto.
39652 (_mm256_i32scatter_ps): Change __addr type to void*.
39653 (_mm256_mask_i32scatter_ps): Ditto.
39654 (_mm_i32scatter_ps): Ditto.
39655 (_mm_mask_i32scatter_ps): Ditto.
39656 (_mm256_i32scatter_pd): Ditto.
39657 (_mm256_mask_i32scatter_pd): Ditto.
39658 (_mm_i32scatter_pd): Ditto.
39659 (_mm_mask_i32scatter_pd): Ditto.
39660 (_mm256_i64scatter_ps): Ditto.
39661 (_mm256_mask_i64scatter_ps): Ditto.
39662 (_mm_i64scatter_ps): Ditto.
39663 (_mm_mask_i64scatter_ps): Ditto.
39664 (_mm256_i64scatter_pd): Ditto.
39665 (_mm256_mask_i64scatter_pd): Ditto.
39666 (_mm_i64scatter_pd): Ditto.
39667 (_mm_mask_i64scatter_pd): Ditto.
39668 (_mm256_i32scatter_epi32): Ditto.
39669 (_mm256_mask_i32scatter_epi32): Ditto.
39670 (_mm_i32scatter_epi32): Ditto.
39671 (_mm_mask_i32scatter_epi32): Ditto.
39672 (_mm256_i32scatter_epi64): Ditto.
39673 (_mm256_mask_i32scatter_epi64): Ditto.
39674 (_mm_i32scatter_epi64): Ditto.
39675 (_mm_mask_i32scatter_epi64): Ditto.
39676 (_mm256_i64scatter_epi32): Ditto.
39677 (_mm256_mask_i64scatter_epi32): Ditto.
39678 (_mm_i64scatter_epi32): Ditto.
39679 (_mm_mask_i64scatter_epi32): Ditto.
39680 (_mm256_i64scatter_epi64): Ditto.
39681 (_mm256_mask_i64scatter_epi64): Ditto.
39682 (_mm_i64scatter_epi64): Ditto.
39683 (_mm_mask_i64scatter_epi64): Ditto.
39684 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39685 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39686 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39687 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39688 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39689 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39690 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39691 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39692 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39693 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39694 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39695 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39696 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39697 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39698 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39699 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39700 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39701 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39702 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39703 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39704 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39705 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39706 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39707 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39708 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39709 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39710 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39711 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39712 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39713 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39714 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39715 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39716 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39717 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39718 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39719 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39720 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39721 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39722 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39723 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39724 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39725 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39726 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39727 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39728 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39729 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39730 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39731 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39732 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39733 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39734 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39735 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39736 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39737 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39738 definitions accordingly.
39740 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39741 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39744 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39747 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39750 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39751 hard regs for inner regno.
39753 2017-01-17 Martin Liska <mliska@suse.cz>
39756 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39757 assumption and add comment.
39759 2017-01-17 Nathan Sidwell <nathan@acm.org>
39761 * ipa-visibility.c (localize_node): New function, broken out of ...
39762 (function_and_variable_visibility): ... here. Call it.
39764 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39766 PR middle-end/77445
39767 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39768 correctly set frequency of oudgoing edge.
39769 (duplicate_thread_path): Fix profile updating.
39771 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39774 * configure.ac: Add GCC_BASE_VER.
39775 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39776 version from BASE-VER file.
39777 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39778 (gcc.o): Depend on $(BASEVER).
39779 * common.opt (dumpfullversion): New option.
39780 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39781 * doc/invoke.texi: Document -dumpfullversion.
39782 * doc/install.texi: Document --with-gcc-major-version-only.
39783 * configure: Regenerated.
39785 2017-01-17 Richard Biener <rguenther@suse.de>
39787 PR tree-optimization/71433
39788 * tree-vrp.c (register_new_assert_for): Merge same asserts
39789 on all incoming edges.
39790 (process_assert_insertions_for): Handle insertions at the
39793 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39795 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39796 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39798 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39801 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39804 2017-01-17 Alan Modra <amodra@gmail.com>
39807 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39808 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39809 symbolic stack limit when pic.
39811 2017-01-16 Martin Sebor <msebor@redhat.com>
39813 PR tree-optimization/78608
39814 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39816 2017-01-16 Jeff Law <law@redhat.com>
39819 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39820 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39821 for several include directories that may be relative to sysroot.
39822 * config/i386/x-mingw32 (gplus_includedir): Define.
39823 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39824 (native_system_includedir): Likewise.
39825 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39826 override if TARGET_SYSTEM_ROOT is defined.
39827 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39829 PR tree-optimization/79090
39830 PR tree-optimization/33562
39831 PR tree-optimization/61912
39832 PR tree-optimization/77485
39833 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39834 and computed trims into the dump file.
39836 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39838 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39840 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39843 * gimplify.c (gimplify_init_constructor): If want_value and
39844 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39848 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39849 sequence. Formatting fixes.
39850 (doloop_optimize): Formatting fixes.
39853 * gcc.c (debug_level_greater_than_spec_func): New function.
39854 (static_spec_functions): Add debug-level-gt spec function.
39855 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39857 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39858 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39859 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39860 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39861 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39862 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39864 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39866 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39867 QImode fixups to general and mask registers only.
39869 2017-01-16 Carl Love <cel@us.ibm.com>
39871 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39872 for built-in functions
39873 vector signed char vec_nabs (vector signed char)
39874 vector signed short vec_nabs (vector signed short)
39875 vector signed int vec_nabs (vector signed int)
39876 vector signed long long vec_nabs (vector signed long long)
39877 vector float vec_nabs (vector float)
39878 vector double vec_nabs (vector double)
39879 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39881 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39882 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39883 * doc/extend.texi: Update the documentation file for the new built-in
39886 2017-01-16 Martin Sebor <msebor@redhat.com>
39888 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39891 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39893 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39894 UNSPEC_VSX__XXSPLTD to require special splat handling.
39896 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39899 * system.h: Poison strndup.
39901 2017-01-16 Alan Modra <amodra@gmail.com>
39904 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39907 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39909 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39911 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39913 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39914 call recog here. Assert that INSN_CODE (insn) is non-negative.
39916 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39919 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39921 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39922 in the currently scheduled RTL fragment.
39924 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39926 PR rtl-optimization/78751
39927 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39930 2017-01-14 Jeff Law <law@redhat.com>
39932 PR tree-optimization/79090
39933 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39934 variable length stores.
39935 (compute_trims): Delete dead assignment to *trim_tail.
39936 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39939 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39941 PR rtl-optimization/78626
39942 PR rtl-optimization/78727
39943 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39944 of a block, and split such blocks after everything else is finished.
39946 2017-01-14 Alan Modra <amodra@gmail.com>
39949 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39950 target legitimate_combined_insn.
39951 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39952 (rs6000_legitimate_combined_insn): New function.
39953 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39955 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39956 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39957 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39959 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39961 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39963 2017-01-13 Jeff Law <law@redhat.com>
39965 PR tree-optimization/33562
39966 PR tree-optimization/61912
39967 PR tree-optimization/77485
39968 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39970 (delete_dead_assignment): Likewise.
39971 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39972 statement to delete_dead_call and delete_dead_assignment.
39974 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39977 * substring-locations.c (format_warning_va): Strengthen case 1 so
39978 that both endpoints of the substring must be within the format
39979 range for just the substring to be printed.
39981 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39983 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39984 * config/i386/i386.c (ix86_target_string): Add missing options
39985 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39986 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39987 flags_other and ix86_target_other to flags2_other. Display unknown
39989 (ix86_valid_target_attribute_inner_p): Add missing options and
39990 reorder options by implied ISAs, as in ix86_target_string.
39992 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39994 * hash-table.h (hash_table::too_empty_p): New function.
39995 (hash_table::expand): Use it.
39996 (hash_table::traverse): Likewise.
39997 (hash_table::empty_slot): Use sizeof (value_type) instead of
39998 sizeof (PTR) to convert bytes to elements. Shrink the table
39999 if the current size is excessive for the current number of
40002 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40004 * ira-costs.c (record_reg_classes): Break from the inner loop
40005 early once alt_fail is known to be true. Update outer loop
40006 handling accordingly.
40008 2017-01-13 Jeff Law <law@redhat.com>
40010 * tree-ssa-dse.c (decrement_count): New function.
40011 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40012 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40013 when we know the partially dead statement is a mem* function.
40015 PR tree-optimization/61912
40016 PR tree-optimization/77485
40017 * tree-ssa-dse.c: Include expr.h.
40018 (maybe_trim_constructor_store): New function.
40019 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40021 PR tree-optimization/33562
40022 PR tree-optimization/61912
40023 PR tree-optimization/77485
40024 * doc/invoke.texi: Document new dse-max-object-size param.
40025 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40026 * tree-ssa-dse.c: Include params.h.
40027 (dse_store_status): New enum.
40028 (initialize_ao_ref_for_dse): New, partially extracted from
40030 (valid_ao_ref_for_dse, normalize_ref): New.
40031 (setup_live_bytes_from_ref, compute_trims): Likewise.
40032 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40033 (maybe_trim_partially_dead_store): Likewise.
40034 (maybe_trim_complex_store): Likewise.
40035 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40036 Track what bytes live from the original store. Return tri-state
40037 for dead, partially dead or live.
40038 (dse_dom_walker): Add constructor, destructor and new private members.
40039 (delete_dead_call, delete_dead_assignment): New extracted from
40041 (dse_optimize_stmt): Make a member of dse_dom_walker.
40042 Use initialize_ao_ref_for_dse.
40044 PR tree-optimization/33562
40045 PR tree-optimization/61912
40046 PR tree-optimization/77485
40047 * sbitmap.h (bitmap_count_bits): Prototype.
40048 (bitmap_clear_range, bitmap_set_range): Likewise.
40049 * sbitmap.c (bitmap_clear_range): New function.
40050 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40052 2017-01-13 Martin Liska <mliska@suse.cz>
40055 * function.c (set_cfun): Add new argument force.
40056 * function.h (set_cfun): Likewise.
40057 * ipa-inline-transform.c (inline_call): Use the function when
40058 strict alising from is dropped for function we inline to.
40060 2017-01-13 Richard Biener <rguenther@suse.de>
40062 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40063 for dumping GIMPLE INTEGER_CSTs.
40065 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40067 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40068 to 201112L since C++17.
40070 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40073 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40074 if -fsanitize=kernel-address is present.
40076 2017-01-13 Richard Biener <rguenther@suse.de>
40078 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40079 as _Literal ( type ) number in case usual suffixes do not
40080 preserve all information.
40082 2017-01-13 Richard Biener <rguenther@suse.de>
40084 PR tree-optimization/77283
40085 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40086 and ssa-iterators.h.
40087 (is_feasible_trace): Implement a cost model based on joiner
40090 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40093 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40094 char or short to __float128/_Float128 directly.
40096 2017-01-12 Martin Sebor <msebor@redhat.com>
40098 to -Wformat-overflow.
40099 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40100 (min_bytes_remaining): Same.
40101 (get_string_length): Same.
40102 (format_string): Same.
40103 (format_directive): Same.
40105 (pass_sprintf_length::handle_gimple_call): Same.
40107 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40109 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40110 info.nowrite calls with no lhs that can't throw. Return bool
40111 whether gsi_remove has been called or not.
40112 (pass_sprintf_length::handle_gimple_call): Return bool whether
40113 try_substitute_return_value called gsi_remove. Formatting fix.
40114 (pass_sprintf_length::execute): Don't use gsi_remove if
40115 handle_gimple_call returned true.
40118 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40119 be removed due to side-effects, don't remove following barrier nor
40120 turn the successor edge into fallthru edge.
40122 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40125 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40126 element-reversing loads and stores as not swappable.
40128 2017-01-12 Nathan Sidwell <nathan@acm.org>
40129 Nicolai Stange <nicstange@gmail.com>
40131 * combine.c (try_combine): Don't ignore result of overlap checking
40132 loop. Combine overlap & asm check into single loop.
40134 2017-01-12 Richard Biener <rguenther@suse.de>
40136 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40137 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40139 2017-01-12 Richard Biener <rguenther@suse.de>
40141 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40142 and TS_TARGET_OPTION directly derive from TS_BASE.
40143 * tree-core.h (tree_optimization_option): Derive from tree_base.
40144 (tree_target_option): Likewise.
40146 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40148 * config/i386/i386.c (memory_address_length): Increase len
40149 only when rip_relative_addr_p returns false.
40151 2017-01-11 Julia Koval <julia.koval@intel.com>
40153 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40154 (OPTION_MASK_ISA_SGX_SET): New.
40155 (ix86_handle_option): Handle OPT_msgx.
40156 * config.gcc: Added sgxintrin.h.
40157 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40158 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40159 * config/i386/i386.c (ix86_target_string): Add -msgx.
40161 (ix86_option_override_internal): Handle new options.
40162 (ix86_valid_target_attribute_inner_p): Add sgx.
40163 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40164 * config/i386/i386.opt: Add msgx.
40165 * config/i386/sgxintrin.h: New file.
40166 * config/i386/x86intrin.h: Add sgxintrin.h.
40168 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40171 * fold-const.c (maybe_nonzero_address): Return 1 for function
40173 (tree_single_nonzero_warnv_p): Don't handle function local objects
40177 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40180 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40183 * opts.c: Include "spellcheck.h"
40184 (struct string_fragment): New struct.
40185 (struct edit_distance_traits<const string_fragment &>): New
40187 (get_closest_sanitizer_option): New function.
40188 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40190 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40192 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40194 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40195 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40196 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40197 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40198 for initial die_offset if dwarf_split_debug_info.
40199 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40200 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40202 (output_skeleton_debug_sections): Formatting fix. Use
40203 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40204 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40206 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40208 * config/arm/cortex-a53.md: Add bypasses for
40209 cortex_a53_r2f_cvt.
40210 (cortex_a53_r2f): Only use for transfers.
40211 (cortex_a53_f2r): Likewise.
40212 (cortex_a53_r2f_cvt): Add reservation for conversions.
40213 (cortex_a53_f2r_cvt): Likewise.
40215 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40217 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40218 to all inlined functions, change static to extern.
40220 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40223 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40225 (arm_assemble_integer): Likewise.
40227 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40229 * config.gcc: Use new awk script to check CPU, FPU and architecture
40230 parameters for --with-... options.
40231 * config/arm/parsecpu.awk: New file
40232 * config/arm/arm-cpus.in: New file.
40233 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40235 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40237 * config/arm/t-arm: Update dependency rules.
40238 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40239 of processing .def files.
40240 * config/arm/genopt.sh: Deleted.
40241 * config/arm/gentune.sh: Deleted.
40242 * config/arm/arm-cores.def: Deleted.
40243 * config/arm/arm-arches.def: Deleted.
40244 * config/arm/arm-fpus.def: Deleted.
40245 * config/arm/arm-tune.md: Regenerated.
40246 * config/arm/arm-tables.opt: Regenerated.
40247 * config/arm/arm-cpu.h: New generated file.
40248 * config/arm/arm-cpu-data.h: New generated file.
40249 * config/arm/arm-cpu-cdata.h: New generated file.
40251 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40254 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40256 (input_varpool_node): Unpack dynamically_initialized bit.
40258 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40260 PR rtl-optimization/79032
40261 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40262 the alignment of the adjusted memory reference against that of MODE,
40263 instead of the alignment of the original memory reference.
40265 2017-01-11 Martin Jambor <mjambor@suse.cz>
40267 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40269 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40270 decorated functions.
40272 2017-01-11 Richard Biener <rguenther@suse.de>
40274 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40275 set range/nonnull info for PHI results. Do not set it on
40276 stmts marked for removal.
40278 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40280 * expr.c (store_field): In the bitfield case, fetch the return value
40281 from the registers before applying a single big-endian adjustment.
40282 Always do a final load for a BLKmode value not larger than a word.
40284 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40287 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40288 that we correctly handle column numbers greater than
40289 LINE_MAP_MAX_COLUMN_NUMBER.
40291 2017-01-10 Martin Sebor <msebor@redhat.com>
40293 PR middle-end/78245
40294 * gimple-ssa-sprintf.c (get_destination_size): Call
40295 {init,fini}object_sizes.
40296 * tree-object-size.c (addr_object_size): Adjust.
40297 (pass_through_call): Adjust.
40298 (pass_object_sizes::execute): Adjust.
40299 * tree-object-size.h (fini_object_sizes): Declare.
40301 2017-01-10 Martin Sebor <msebor@redhat.com>
40303 PR tree-optimization/78775
40304 * builtins.c (get_size_range): Move...
40305 * calls.c: ...to here.
40306 (alloc_max_size): Accept zero argument.
40307 (operand_signed_p): Remove.
40308 (maybe_warn_alloc_args_overflow): Call get_size_range.
40309 * calls.h (get_size_range): Declare.
40311 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40313 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40314 from TI's devices.csv file as of September 2016.
40315 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40317 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40319 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40320 * doc/invoke.texi: Likewise.
40321 * doc/md.texi: Likewise.
40322 * doc/objc.texi: Likewise.
40324 2017-01-10 Joshua Conner <joshconner@google.com>
40326 * config/arm/fuchsia-elf.h: New file.
40327 * config/fuchsia.h: New file.
40328 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40329 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40331 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40333 2016-01-10 Richard Biener <rguenther@suse.de>
40335 PR tree-optimization/79034
40336 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40337 Propagate out degenerate PHIs in the joiner.
40339 2017-01-10 Martin Liska <mliska@suse.cz>
40341 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40342 (sort_congruence_classes_by_decl_uid): Likewise.
40343 (sort_congruence_class_groups_by_decl_uid): Likewise.
40344 (sem_item_optimizer::merge_classes): Sort class, groups in these
40345 classes and members in the groups by DECL_UID of declarations.
40346 This would make merge operations stable.
40348 2017-01-10 Martin Liska <mliska@suse.cz>
40350 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40351 usage of m_classes_vec.
40352 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40353 (sem_item_optimizer::get_group_by_hash): Likewise.
40354 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40355 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40356 (sem_item_optimizer::verify_classes): Likewise.
40357 (sem_item_optimizer::process_cong_reduction): Likewise.
40358 (sem_item_optimizer::dump_cong_classes): Likewise.
40359 (sem_item_optimizer::merge_classes): Likewise.
40360 * ipa-icf.h (congruence_class_hash): Rename from
40361 congruence_class_group_hash. Remove declaration of m_classes_vec.
40363 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40365 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40366 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40367 * config.gcc: Add avx512vpopcntdqintrin.h.
40368 * config/i386/avx512vpopcntdqintrin.h: New.
40369 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40370 * config/i386/i386-builtin-types.def: Add new types.
40371 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40372 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40373 __builtin_ia32_vpopcountq_v8di_mask): New.
40374 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40375 __AVX512VPOPCNTDQ__.
40376 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40377 (PTA_AVX512VPOPCNTDQ): Define.
40378 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40379 TARGET_AVX512VPOPCNTDQ_P): Define.
40380 * config/i386/i386.opt: Add mavx512vpopcntdq.
40381 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40382 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40384 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40386 PR middle-end/77484
40387 * predict.def (PRED_CALL): Set to 67.
40389 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40391 * expr.c (store_field): In the bitfield case, if the value comes from
40392 a function call and is of an aggregate type returned in registers, do
40393 not modify the field mode; extract the value in all cases if the mode
40394 is BLKmode and the size is not larger than a word.
40396 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40399 * config/i386/cpuid.h: Fix undefined behavior.
40401 2017-01-04 Jeff Law <law@redhat.com>
40403 PR tree-optimization/79007
40404 PR tree-optimization/67955
40405 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40406 conservative for pt.null when flag_non_call_exceptions is on.
40408 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40410 PR translation/79019
40411 PR translation/79020
40412 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40413 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40414 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40416 * config/avr/avr.opt (maccumulate-args): Likewise.
40417 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40418 * common.opt (freport-bug): Likewise.
40419 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40420 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40421 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40422 translatable string.
40423 * config/i386/i386.c (function_value_32): Likewise.
40424 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40425 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40427 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40428 * common/config/msp430/msp430-common.c (msp430_handle_option):
40430 * symtab.c (symtab_node::verify_base): Likewise.
40431 * opts.c (set_debug_level): Likewise.
40432 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40433 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40434 missing whitespace to translatable strings.
40435 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40436 * config/sh/superh.h: Likewise.
40437 * config/i386/xopintrin.h: Likewise.
40438 * config/i386/znver1.md: Likewise.
40439 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40440 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40441 * double-int.h (struct double_int): Likewise.
40442 * double-int.c (div_and_round_double): Likewise.
40443 * wide-int.cc: Likewise.
40444 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40445 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40446 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40447 (crossjumps_occurred): ... this.
40448 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40451 PR tree-optimization/78899
40452 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40453 returning bool return struct loop *, NULL for failure and the new
40455 (versionable_outer_loop_p): Don't version outer loop if it has
40456 dont_vectorized bit set.
40457 (tree_if_conversion): When versioning outer loop, ensure
40458 tree_if_conversion is performed also on the inner loop of the
40459 non-vectorizable outer loop copy.
40460 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40461 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40462 prevent vectorization of it.
40463 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40464 the outer loop vectorization of the non-scalar version is attempted
40465 before vectorization of the inner loop in scalar version. If
40466 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40467 vectorization of its inner loop.
40468 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40469 has 2 inner loops, rename also on edges from bb whose single pred
40470 is outer_loop->header. Fix typo in function comment.
40472 2017-01-09 Martin Sebor <msebor@redhat.com>
40475 * asan.c (asan_emit_stack_protection): Increase local buffer size
40476 to avoid snprintf truncation warning.
40478 2017-01-09 Andrew Pinski <apinski@cavium.com>
40480 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40481 to reference thunderx2t99 for the tuning structure
40482 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40484 (thunderx2t99_extra_costs): This.
40485 * config/aarch64/aarch64-tune.md: Regenerate.
40486 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40487 (vulcan_addrcost_table): This.
40488 (vulcan_regmove_cost): Rename to ...
40489 (thunderx2t99_regmove_cost): This.
40490 (vulcan_vector_cost): Rename to ...
40491 (thunderx2t99_vector_cost): this.
40492 (vulcan_branch_cost): Rename to ...
40493 (thunderx2t99_branch_cost): This.
40494 (vulcan_tunings): Rename to ...
40495 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40496 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40498 2017-01-09 Martin Jambor <mjambor@suse.cz>
40502 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40503 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40504 (propagate_vr_accross_jump_function): Use the above function for all
40505 value range computations for pass-through jump functions and type
40506 converasion from explicit value range values.
40507 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40508 parameters from TYPE_ARG_TYPES.
40509 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40510 (ipa_write_node_info): Stream type of the actual argument.
40511 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40513 2017-01-09 Martin Liska <mliska@suse.cz>
40516 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40517 (lookup_compiler): Do not show error message with have_E.
40519 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40521 PR tree-optimization/78938
40522 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40523 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40524 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40525 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40528 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40530 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40533 2017-01-09 Richard Biener <rguenther@suse.de>
40535 PR tree-optimization/78997
40536 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40537 name condition properly.
40539 2017-01-09 Richard Biener <rguenther@suse.de>
40542 * dwarf2out.c (is_cxx): New overload with context.
40543 (is_naming_typedef_decl): Use it.
40545 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40547 * invoke.texi (Option Summary): Correct spacing in option lists
40548 and add line breaks to fix over-long lines.
40550 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40552 PR middle-end/17660
40554 * extend.texi (Common Variable Attributes): Add xref to GCC
40555 Internals manual to explain mode attribute keywords.
40557 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40560 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40561 and Preprocessor Options.
40562 (Options for Linking): Document -pthread here....
40563 (RS/6000 and PowerPC Options): ...not here.
40564 (Solaris 2 Options): ...or here.
40565 * doc/cppopts.texi: Document -pthread.
40567 2017-01-08 Martin Sebor <msebor@redhat.com>
40569 PR middle-end/77708
40570 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40571 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40572 New member functions.
40573 (format_directive): Used them.
40575 (pass_sprintf_length::handle_gimple_call): Same.
40576 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40577 to avoid truncation for any argument.
40578 (extract_affine_mul): Same.
40579 * tree.c (get_file_function_name): Same.
40581 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40583 PR middle-end/77484
40584 * predict.def (PRED_INDIR_CALL): Set to 86.
40586 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40588 PR preprocessor/54124
40589 * doc/cppopts.texi: Reformat -d subtable to list the full name
40590 of the options. Add cross-reference to the docs for the general
40591 compiler -d options.
40592 * doc/invoke.texi (Developer Options): Add cross-reference to the
40593 preprocessor-specific -d option documentation.
40595 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40597 PR preprocessor/13498
40598 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40599 redudant material, and reflect new command-line options.
40600 (System Headers): Likewise.
40602 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40604 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40605 -isystem, and -idirafter. Copy-edit.
40606 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40607 default for -ftrack-macro-expansion. Delete obsolete and
40608 badly-formatted implementation details about -fdebug-cpp output.
40609 * doc/cppwarnopts.texi: Copy-edit.
40611 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40614 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40615 that the transition from a max line width >= 1<<10 to narrower
40616 lines works correctly.
40618 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40620 * doc/options.texi (PerFunction): New.
40621 * opt-functions.awk (switch_flags): Map both Optimization and
40622 PerFunction to CL_OPTIMIZATION.
40623 * opth-gen.awk: Test for PerFunction flag along with
40625 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40626 it only when the latter is present. Skip those that don't in
40627 the hash function generator.
40628 * common.opt (fvar-tracking): Mark as PerFunction instead of
40630 (fvar-tracking-assignments): Likewise.
40631 (fvar-tracking-assignments-toggle): Likewise.
40632 (fvar-tracking-uninit): Likewise.
40634 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40636 PR translation/79018
40637 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40640 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40643 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40644 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40645 TARGET_LONG_JUMP_TABLE_OFFSETS.
40646 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40647 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40648 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40649 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40650 * config/m68k/m68k.md (tablejump expander): Likewise.
40651 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40652 TARGET_LONG_JUMP_TABLE_OFFSETS.
40653 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40654 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40656 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40657 David Holsgrove <david.holsgrove@xilinx.com>
40659 * common/config/microblaze/microblaze-common.c
40660 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40661 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40663 * config/microblaze/microblaze.c (microblaze_must_save_register)
40664 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40666 (microblaze_eh_return): New function.
40667 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40668 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40669 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40670 * config/microblaze/microblaze.md (eh_return): New pattern.
40672 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40674 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40675 GCC_DIAGNOSTIC_STRINGIFY): Define.
40677 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40679 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40681 * config/arm/arm.md (<mcrr>): New.
40683 * config/arm/arm.c (arm_arch5te): New.
40684 (arm_option_override): Set arm_arch5te.
40685 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40687 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40688 (arm_mcrr_qualifiers): ... this. New.
40689 (MRRC_QUALIFIERS): Define to...
40690 (arm_mrrc_qualifiers): ... this. New.
40691 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40693 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40694 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40695 (MRRCI, mrrc, MRRC): New.
40696 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40697 VUNSPEC_MRRC2): New.
40699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40701 * config/arm/arm.md (<mcr>): New.
40703 * config/arm/arm.c (arm_coproc_builtin_available): Add
40704 support for mcr, mrc, mcr2 and mrc2.
40705 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40706 (arm_mcr_qualifiers): ... this. New.
40707 (MRC_QUALIFIERS): Define to ...
40708 (arm_mrc_qualifiers): ... this. New.
40709 (MCR_QUALIFIERS): Define to ...
40710 (arm_mcr_qualifiers): ... this. New.
40711 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40713 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40714 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40715 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40716 VUNSPEC_MRC2): New.
40718 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40720 * config/arm/arm.md (*ldc): New.
40724 * config/arm/arm.c (arm_coproc_builtin_available): Add
40725 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40726 (arm_coproc_ldc_stc_legitimate_address): New.
40727 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40728 'qualifier_const_pointer'.
40729 (LDC_QUALIFIERS): Define to...
40730 (arm_ldc_qualifiers): ... this. New.
40731 (STC_QUALIFIERS): Define to...
40732 (arm_stc_qualifiers): ... this. New.
40733 * config/arm/arm-protos.h
40734 (arm_coproc_ldc_stc_legitimate_address): New.
40735 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40736 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40737 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40738 stc2, stcl, stc2l): New.
40739 * config/arm/constraints.md (Uz): New.
40740 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40741 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40742 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40743 VUNSPEC_STC2L): New.
40745 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40747 * config/arm/arm.md (<cdp>): New.
40748 * config/arm/arm.c (neon_const_bounds): Rename this ...
40749 (arm_const_bounds): ... this.
40750 (arm_coproc_builtin_available): New.
40751 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40752 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40753 (CDP_QUALIFIERS): Define to...
40754 (arm_cdp_qualifiers): ... this. New.
40756 (arm_expand_builtin_args): Add case for 6 arguments.
40757 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40758 (arm_const_bounds): ... this.
40759 (arm_coproc_builtin_available): New.
40760 * config/arm/arm_acle.h (__arm_cdp): New.
40762 * config/arm/arm_acle_builtins.def (cdp): New.
40764 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40765 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40766 'arm_const_bounds'.
40767 * config/arm/types.md (coproc): New.
40768 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40769 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40770 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40771 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40773 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40775 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40776 (UBINOP_QUALIFIERS): New.
40778 (acle_builtin_data): New. Change comment.
40779 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40780 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40781 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40782 arm_acle_builtins.def.
40783 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40784 (arm_init_acle_builtins): New.
40785 (CRC32_BUILTIN): Remove.
40786 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40787 crc32cb, crc32ch and crc32cw.
40788 (arm_init_crc32_builtins): Remove.
40789 (arm_init_builtins): Use arm_init_acle_builtins rather
40790 than arm_init_crc32_builtins.
40791 (arm_expand_acle_builtin): New.
40792 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40793 * config/arm/arm_acle_builtins.def: New.
40795 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40797 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40798 (arm_builtin_datum): ... this.
40799 (arm_init_neon_builtin): Rename to ...
40800 (arm_init_builtin): ... this. Add a new parameters PREFIX
40801 and USE_SIG_IN_NAME.
40802 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40803 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40804 'arm_builtin_datum'.
40805 (arm_init_vfp_builtins): Likewise.
40806 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40807 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40808 (arm_expand_neon_args): Rename to ...
40809 (arm_expand_builtin_args): ... this. Rename builtin_arg
40810 enum values and differentiate between ARG_BUILTIN_MEMORY
40811 and ARG_BUILTIN_NEON_MEMORY.
40812 (arm_expand_neon_builtin_1): Rename to ...
40813 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40814 values, arm_expand_builtin_args and add bool parameter NEON.
40815 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40816 (arm_expand_vfp_builtin): Likewise.
40817 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40819 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40821 PR middle-end/77484
40822 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40823 * predict.c (tree_estimate_probability_bb): Reverse direction of
40824 polymorphic call predictor.
40826 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40828 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40829 (determine_pass_name_match): ...this new function and...
40830 (should_skip_pass_p): ...this new function.
40832 2017-01-06 Nathan Sidwell <nathan@acm.org>
40834 * ipa-visibility.c (function_and_variable_visibility): Reformat
40835 comments and long lines. Remove extrneous if.
40836 * symtab.c (symtab_node::make_decl_local): Fix code format.
40837 (symtab_node::set_section_for_node): Fix comment typo.
40839 2017-01-06 Martin Liska <mliska@suse.cz>
40842 * lra-constraints.c: Rename invariant to lra_invariant.
40843 * predict.c (set_even_probabilities): Initialize e to NULL.
40845 2017-01-05 Martin Sebor <msebor@redhat.com>
40847 PR tree-optimization/78910
40848 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40849 (format_integer): Correct off-by-one error in the handling
40850 of precision with negative numbers in signed conversions..
40852 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40854 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40856 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40858 PR tree-optimization/71016
40859 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40860 factor_out_conditional_conversion. Formatting fix.
40861 (factor_out_conditional_conversion): Add cond_stmt argument.
40862 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40863 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40866 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40868 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40869 read-rtl-function.o, and selftest-rtl.o.
40870 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40871 (selftest::aarch64_test_loading_full_dump): New function.
40872 (selftest::aarch64_run_selftests): New function.
40873 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40874 selftest::aarch64_run_selftests.
40875 * config/i386/i386.c
40876 (selftest::ix86_test_loading_dump_fragment_1): New function.
40877 (selftest::ix86_test_loading_call_insn): New function.
40878 (selftest::ix86_test_loading_full_dump): New function.
40879 (selftest::ix86_test_loading_unspec): New function.
40880 (selftest::ix86_run_selftests): Call the new functions.
40881 * emit-rtl.c (maybe_set_max_label_num): New function.
40882 * emit-rtl.h (maybe_set_max_label_num): New decl.
40883 * function.c (instantiate_decls): Guard call to
40884 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40885 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40887 * gensupport.c (gen_reader::gen_reader): Pass "false"
40888 for new "compact" param of rtx_reader.
40889 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40890 rather than an empty string for NULL strings.
40891 * read-md.c: Potentially include config.h rather than bconfig.h.
40892 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40893 (have_error): New global, copied from errors.c.
40894 (md_reader::read_name): Rename to...
40895 (md_reader::read_name_1): ...this, adding "out_loc" param,
40896 and converting "missing name or number" to returning false, rather
40898 (md_reader::read_name): Reimplement in terms of read_name_1.
40899 (md_reader::read_name_or_nil): New function.
40900 (md_reader::read_string): Handle "(nil)" by returning NULL.
40901 (md_reader::md_reader): Add new param "compact".
40902 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40903 (md_reader::read_file): New method.
40904 * read-md.h (md_reader::md_reader): Add new param "compact".
40905 (md_reader::read_file): New method.
40906 (md_reader::is_compact): New accessor.
40907 (md_reader::read_name): Convert return type from void to file_location.
40908 (md_reader::read_name_or_nil): New decl.
40909 (md_reader::read_name_1): New decl.
40910 (md_reader::m_compact): New field.
40911 (noop_reader::noop_reader): Pass "false" for new "compact" param
40913 (rtx_reader::rtx_reader): Add new "compact" param.
40914 (rtx_reader::read_rtx_operand): Make virtual and convert return
40915 type from void to rtx.
40916 (rtx_reader::read_until): New decl.
40917 (rtx_reader::handle_any_trailing_information): New virtual function.
40918 (rtx_reader::postprocess): New virtual function.
40919 (rtx_reader::finalize_string): New virtual function.
40920 (rtx_reader::m_in_call_function_usage): New field.
40921 (rtx_reader::m_reuse_rtx_by_id): New field.
40922 * read-rtl-function.c: New file.
40923 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40924 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40925 (selftest::verify_three_block_rtl_cfg): New decl.
40926 * read-rtl-function.h: New file.
40927 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40928 For host, include function.h, memmodel.h, and emit-rtl.h.
40929 (one_time_initialization): New function.
40930 (struct compact_insn_name): New struct.
40931 (compact_insn_names): New array.
40932 (find_code): Handle insn codes in compact dumps.
40933 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40934 (bind_subst_iter_and_attr): Likewise.
40935 (add_condition_to_string): Likewise.
40936 (add_condition_to_rtx): Likewise.
40937 (apply_attribute_uses): Likewise.
40938 (add_current_iterators): Likewise.
40939 (apply_iterators): Likewise.
40940 (initialize_iterators): Guard usage of apply_subst_iterator with
40941 #ifdef GENERATOR_FILE.
40942 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40943 (md_reader::read_mapping): Likewise.
40944 (add_define_attr_for_define_subst): Likewise.
40945 (add_define_subst_attr): Likewise.
40946 (read_subst_mapping): Likewise.
40947 (check_code_iterator): Likewise.
40948 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40950 (one_time_initialization): New function.
40951 (rtx_reader::read_until): New method.
40952 (read_flags): New function.
40953 (parse_reg_note_name): New function.
40954 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40955 Handle reuse_rtx ids.
40956 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40957 Add parsing support for RTL dumps, mirroring the special-cases in
40958 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40959 values, and calling handle_any_trailing_information.
40960 (rtx_reader::read_rtx_operand): Convert return type from void
40961 to rtx, returning return_rtx. Handle case 'e'. Call
40962 finalize_string on XSTR and XTMPL fields.
40963 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40964 "(nil)" values were omitted. Call the postprocess vfunc on the
40966 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40967 class ctor. Initialize m_in_call_function_usage. Call
40968 one_time_initialization.
40969 * rtl-tests.c (selftest::test_uncond_jump): Call
40970 set_new_first_and_last_insn.
40971 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40972 * selftest-rtl.c: New file.
40973 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40974 (selftest::get_insn_by_uid): New decl.
40975 * selftest-run-tests.c (selftest::run_tests): Call
40976 read_rtl_function_c_tests.
40977 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40978 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40981 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40983 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40984 operands in a special way. Assert that pos+len <= mode precision.
40986 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40988 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40989 3 argument Alias with unlimited for the negative form.
40990 (fno-vect-cost-model): Removed.
40992 2017-01-05 Martin Liska <mliska@suse.cz>
40994 * hsa-gen.c (gen_hsa_divmod): New function.
40995 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40997 2017-01-05 Martin Liska <mliska@suse.cz>
41000 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41003 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41005 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41006 small constant length operands.
41008 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41010 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41011 between loop iterations.
41013 2017-01-05 Martin Liska <mliska@suse.cz>
41016 * gimplify.c (gimplify_decl_expr): Compare to
41017 asan_poisoned_variables instread of checking flags.
41018 (gimplify_target_expr): Likewise.
41019 (gimplify_expr): Likewise.
41020 (gimplify_function_tree): Conditionally initialize
41021 asan_poisoned_variables.
41023 2017-01-04 Jeff Law <law@redhat.com>
41025 PR tree-optimizatin/78812
41026 * rtl.h (contains_mem_rtx_p): Prototype.
41027 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41028 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41029 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41030 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41031 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41033 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41035 * input.c (assert_char_at_range): Default-initialize actual_range.
41037 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41039 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41040 to match the caller.
41042 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41044 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41045 insns after final jump in test to emit dummy move.
41047 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41049 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41050 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41054 * multiple_target.c (create_dispatcher_calls): Init e_next.
41055 * tree-ssa-loop-split.c (split_loop): Init border.
41056 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41059 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41064 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41065 (altivec_register_operand): Do not return true if the operand
41066 contains a SUBREG mixing SImode and SFmode.
41067 (vsx_register_operand): Likewise.
41068 (vsx_reg_sfsubreg_ok): New predicate.
41069 (vfloat_operand): Do not return true if the operand contains a
41070 SUBREG mixing SImode and SFmode.
41071 (vint_operand): Likewise.
41072 (vlogical_operand): Likewise.
41073 (gpc_reg_operand): Likewise.
41074 (int_reg_operand): Likewise.
41075 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41076 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41077 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41079 (rs6000_emit_move_si_sf_subreg): New helper function.
41080 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41081 fixup SUBREGs involving SImode and SFmode.
41082 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41083 numbers for the new peephole2 optimization.
41084 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41085 the GLIBC math library that do AND/IOR/XOR operations on single
41086 precision floating point.
41087 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41088 target macros to say whether we need to avoid SUBREGs mixing
41090 (TARGET_ALLOW_SF_SUBREG): Likewise.
41091 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41092 (UNSPEC_SI_FROM_SF): Likewise.
41093 (iorxor): Change spacing.
41094 (and_ior_xor): New iterator for AND, IOR, and XOR.
41095 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41096 (movdi_from_sf_zero_ext): Likewise.
41097 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41098 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41099 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41100 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41101 (fms<mode>4): Likewise.
41102 (fnma<mode>4): Likewise.
41103 (fnms<mode>4): Likewise.
41104 (nfma<mode>4): Likewise.
41105 (nfms<mode>4): Likewise.
41107 2017-01-04 Marek Polacek <polacek@redhat.com>
41110 * doc/invoke.texi: Document -Wpointer-compare.
41112 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41114 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41117 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41118 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41121 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41124 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41125 documentation of the powerpc_popcntb_ok attribute.
41126 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41127 code to issue warning messages if a requested CPU configuration is
41128 not supported by the binary (assembler and loader) toolchain.
41129 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41130 made to define a built-in function that has been disabled.
41131 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41132 made to define a built-in function that has been disabled.
41133 (altivec_init_builtins): Add comment explaining why definition
41134 of the DST built-in functions is not preceded by an assertion
41135 check. Add assertions to prevent ICE if attempts are made to
41136 define an altivec predicate or an abs* built-in function that has
41138 (htm_init_builtins): Add comment explaining why definition of the
41139 htm built-in functions is not preceded by an assertion check.
41141 2017-01-04 Jeff Law <law@redhat.com>
41143 PR tree-optimizatin/67955
41144 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41145 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41146 the points-to solution does not include pt_null. Use DECL_PT_UID
41149 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41151 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41152 Use gen_int_mode instead of gen_lopwart for const_int operands.
41154 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41156 PR tree-optimization/71563
41157 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41158 out of range value - has low bits known to be zero.
41160 2017-01-04 Alan Modra <amodra@gmail.com>
41162 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41163 * configure: Regenerate.
41164 * config.in: Regenerate.
41166 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41169 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41170 a substring of the message, but strcmp with the whole message. Ifdef
41171 ENABLE_NLS, translate the message first using dgettext.
41173 2017-01-03 Jeff Law <law@redhat.com>
41175 PR tree-optimizatin/78856
41176 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41177 (mark_threaded_blocks): Remove code to truncate thread paths that
41178 cross multiple loop headers. Instead invalidate the cached loop
41179 iteration information and handle case of a thread path walking
41180 into an irreducible region.
41182 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41185 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41186 assertions. Add support for doing the signbit if the IEEE 128-bit
41187 floating point value is in a GPR.
41188 * config/rs6000/rs6000.md (Fsignbit): Delete.
41189 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41190 Update the length attribute if the value is in a GPR.
41191 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41192 the sign or zero extension instruction, since the value is always 0/1.
41193 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41196 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41197 extracting SImode to a GPR register so that we can generate a
41198 store, limit the vector to be in a traditional Altivec register
41199 for the vextuwrx instruction.
41201 2017-01-03 Ian Lance Taylor <iant@google.com>
41203 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41205 2017-01-03 Martin Sebor <msebor@redhat.com>
41207 PR tree-optimization/78696
41208 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41209 precision. Use MPFR for %f for greater fidelity. Correct handling
41211 (pass_sprintf_length::compute_format_length): Set width and precision
41212 specified by asrerisk to void_node for vararg functions.
41213 (try_substitute_return_value): Adjust dump output.
41215 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41217 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41219 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41221 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41222 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41223 -mlra/-mno-lra was passed to the compiler.
41225 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41227 PR rtl-optimization/65618
41228 * emit-rtl.c (try_split): Move initialization of "before" and
41229 "after" to just before the call to emit_insn_after_setloc.
41231 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41233 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41235 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41237 * dwarf2out.c (gen_enumeration_type_die): When
41238 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41240 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41242 PR tree-optimization/78965
41243 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41244 Change first argument from const call_info & to call_info &. For %n
41245 set info.nowrite to false.
41247 PR middle-end/78901
41248 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41249 possibly throwing calls.
41251 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41252 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41253 and fns handling, rather than in a separate case SSA_NAME.
41255 2017-01-02 Jeff Law <law@redhat.com>
41257 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41258 fixes for first_period and second_period variables.
41260 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41263 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41264 (*insvqi_1): New insn pattern.
41265 (*insvqi_1_mem_rex64): Ditto.
41266 (*insvqi_2): Ditto.
41267 (*insvqi_3): Rename from *insvqi.
41269 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41271 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41273 * doc/cfg.texi (Edges): Remove reference to Java.
41274 (Maintaining the CFG): Ditto.
41276 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41278 PR middle-end/77674
41279 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41280 transparent aliases.
41282 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41284 PR middle-end/77484
41285 * predict.def (PRED_CALL): Update hitrate.
41286 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41287 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41288 into direct/indirect/polymorphic variants.
41290 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41292 Update copyright years.
41294 * gcc.c (process_command): Update copyright notice dates.
41295 * gcov-dump.c (print_version): Ditto.
41296 * gcov.c (print_version): Ditto.
41297 * gcov-tool.c (print_version): Ditto.
41298 * gengtype.c (create_file): Ditto.
41299 * doc/cpp.texi: Bump @copying's copyright year.
41300 * doc/cppinternals.texi: Ditto.
41301 * doc/gcc.texi: Ditto.
41302 * doc/gccint.texi: Ditto.
41303 * doc/gcov.texi: Ditto.
41304 * doc/install.texi: Ditto.
41305 * doc/invoke.texi: Ditto.
41307 Copyright (C) 2017 Free Software Foundation, Inc.
41309 Copying and distribution of this file, with or without modification,
41310 are permitted in any medium without royalty provided the copyright
41311 notice and this notice are preserved.