1 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
3 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
4 generation for cases where splatting a value is not useful.
5 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6 across a vec_duplicate and a paradoxical subreg forming a vector
9 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
11 * combine.c (simplify_set): Do not transform subregs to zero_extends
12 if the destination is not a scalar int mode.
14 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
17 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
18 shared zero if the input has overflowed.
20 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
23 support for Cortex-A55 and Cortex-A75.
25 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27 * common/config/arm/arm-common.c (compare_opt_names): Add function
28 comment. Use strcmp instead of manual loop.
30 2017-12-21 Martin Liska <mliska@suse.cz>
33 * gcov-dump.c (dump_gcov_file): Do not read info about
34 support_unexecuted_blocks for gcda files.
36 2017-12-21 Jakub Jelinek <jakub@redhat.com>
38 PR rtl-optimization/82973
39 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
40 (valid_for_const_vector_p): ... this.
41 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
42 (valid_for_const_vector_p): ... this. Adjust function comment.
43 (gen_vec_duplicate): Adjust caller.
44 * optabs.c (expand_vector_broadcast): Likewise.
45 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
46 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
48 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
49 instead of GET_CODE == CONST_FIXED.
50 (simplify_subreg): Use CONST_FIXED_P macro instead of
51 GET_CODE == CONST_FIXED.
54 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
55 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
56 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
57 (ix86_option_override_internal): Adjust for
58 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
59 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
60 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
61 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
62 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
63 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
64 def_builtin2 for CET builtins.
65 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
66 up formatting in IX86_BUILTIN_RDPID code.
67 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
68 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
69 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
70 ix86_isa_flags2 to ix86_isa_flags.
71 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
73 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
74 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
75 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
77 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
78 Or in OPTION_MASK_ISA_AVX512F_SET.
79 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
80 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
81 (ix86_handle_option): Adjust for
82 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
83 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
85 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
86 Alan Hayward <alan.hayward@arm.com>
87 David Sherwood <david.sherwood@arm.com>
89 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
90 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
91 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
92 Track polynomial offsets.
94 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
95 Alan Hayward <alan.hayward@arm.com>
96 David Sherwood <david.sherwood@arm.com>
98 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
99 Treat drb->init as a poly_int. Fail if its misalignment wrt
100 vector_alignment isn't known.
102 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
103 Alan Hayward <alan.hayward@arm.com>
104 David Sherwood <david.sherwood@arm.com>
106 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
108 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
109 (vect_loop_versioning): Take the loop versioning threshold as a
111 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
112 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
113 versioning_threshold.
114 (vect_analyze_loop_2): Compute the loop versioning threshold
115 whenever loop versioning is needed, and store it in the new
116 field rather than combining it with the cost model threshold.
117 (vect_transform_loop): Update call to vect_loop_versioning.
118 Try to combine the loop versioning and cost thresholds here.
120 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
121 Alan Hayward <alan.hayward@arm.com>
122 David Sherwood <david.sherwood@arm.com>
124 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
125 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
126 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
127 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
128 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
129 (group_compare_offset): Update accordingly.
130 (split_small_address_groups_p): Likewise.
131 (record_use): Take addr_offset as a poly_uint64 rather than
132 an unsigned HOST_WIDE_INT.
133 (strip_offset): Return the offset as a poly_uint64 rather than
134 an unsigned HOST_WIDE_INT.
135 (record_group_use, split_address_groups): Track polynomial offsets.
136 (add_iv_candidate_for_use): Likewise.
137 (addr_offset_valid_p): Take the offset as a poly_int64 rather
138 than a HOST_WIDE_INT.
139 (strip_offset_1): Return the offset as a poly_int64 rather than
142 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
143 Alan Hayward <alan.hayward@arm.com>
144 David Sherwood <david.sherwood@arm.com>
146 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
147 rather than a HOST_WIDE_INT.
148 * tree.c (get_binfo_at_offset): Likewise.
150 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
151 Alan Hayward <alan.hayward@arm.com>
152 David Sherwood <david.sherwood@arm.com>
154 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
155 rather than a HOST_WIDE_INT.
156 * tree-sra.c (build_ref_for_offset): Likewise.
158 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
159 Alan Hayward <alan.hayward@arm.com>
160 David Sherwood <david.sherwood@arm.com>
162 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
164 * tree.c (mem_ref_offset): Likewise.
165 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
166 * builtins.c (get_object_alignment_2): Likewise.
167 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
168 * gimple-fold.c (get_base_constructor): Likewise.
169 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
170 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
172 * ipa-polymorphic-call.c
173 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
174 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
175 (get_ancestor_addr_info): Likewise.
176 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
177 * match.pd: Likewise.
178 * tree-data-ref.c (dr_analyze_innermost): Likewise.
179 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
180 * tree-eh.c (tree_could_trap_p): Likewise.
181 * tree-object-size.c (addr_object_size): Likewise.
182 * tree-ssa-address.c (copy_ref_info): Likewise.
183 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
184 (indirect_refs_may_alias_p): Likewise.
185 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
186 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
187 (non_rewritable_mem_ref_base): Likewise.
188 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
189 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
190 * varasm.c (decode_addr_const): Likewise.
192 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
193 Alan Hayward <alan.hayward@arm.com>
194 David Sherwood <david.sherwood@arm.com>
196 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
197 offsets for MEM_REFs.
199 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
204 as a poly_uint64 rather than an unsigned int.
205 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
207 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
208 Alan Hayward <alan.hayward@arm.com>
209 David Sherwood <david.sherwood@arm.com>
211 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
213 (decode_addr_const): Update accordingly.
215 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
216 Alan Hayward <alan.hayward@arm.com>
217 David Sherwood <david.sherwood@arm.com>
219 * tree.h (bit_field_size, bit_field_offset): New functions.
220 * hsa-gen.c (gen_hsa_addr): Use them.
221 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
222 (simplify_vector_constructor): Likewise.
223 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
224 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
225 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
226 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
229 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
230 Alan Hayward <alan.hayward@arm.com>
231 David Sherwood <david.sherwood@arm.com>
233 * expr.h (emit_group_load, emit_group_load_into_temps)
234 (emit_group_store): Take the size as a poly_int64 rather than an int.
235 * expr.c (emit_group_load_1, emit_group_load): Likewise.
236 (emit_group_load_into_temp, emit_group_store): Likewise.
238 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
239 Alan Hayward <alan.hayward@arm.com>
240 David Sherwood <david.sherwood@arm.com>
242 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
243 an unsigned int to a poly_uint64.
244 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
245 sizes as poly_uint64s rather than unsigned ints.
246 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
249 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayward <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * lra-constraints.c (emit_inc): Change inc_amount from an int
256 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
257 Alan Hayward <alan.hayward@arm.com>
258 David Sherwood <david.sherwood@arm.com>
260 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
262 (add_stack_var, stack_var_cmp, partition_stack_vars)
263 (dump_stack_var_partition): Update accordingly.
264 (alloc_stack_frame_space): Take the size as a poly_int64 rather
265 than a HOST_WIDE_INT.
266 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
267 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
268 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
269 rather than a HOST_WIDE_INT.
271 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
272 Alan Hayward <alan.hayward@arm.com>
273 David Sherwood <david.sherwood@arm.com>
275 * target.def (return_pops_args): Treat both the input and output
276 sizes as poly_int64s rather than HOST_WIDE_INTS.
277 * targhooks.h (default_return_pops_args): Update accordingly.
278 * targhooks.c (default_return_pops_args): Likewise.
279 * doc/tm.texi: Regenerate.
280 * emit-rtl.h (incoming_args): Change pops_args, size and
281 outgoing_args_size from int to poly_int64_pod.
282 * function.h (expr_status): Change x_pending_stack_adjust and
283 x_stack_pointer_delta from int to poly_int64.
284 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
285 (ARGS_SIZE_RTX): Update accordingly.
286 * calls.c (highest_outgoing_arg_in_use): Change from int to
288 (stack_usage_watermark, stored_args_watermark): New variables.
289 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
290 (emit_call_1): Change the stack_size and rounded_stack_size
291 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
293 (save_fixed_argument_area): Check stack_usage_watermark.
294 (initialize_argument_information): Change old_pending_adj from
295 a HOST_WIDE_INT * to a poly_int64_pod *.
296 (compute_argument_block_size): Return the size as a poly_int64
298 (finalize_must_preallocate): Track polynomial argument sizes.
299 (compute_argument_addresses): Likewise.
300 (internal_arg_pointer_based_exp): Track polynomial offsets.
301 (mem_overlaps_already_clobbered_arg_p): Rename to...
302 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
303 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
304 Check stored_args_used_watermark.
305 (load_register_parameters): Update accordingly.
306 (check_sibcall_argument_overlap_1): Likewise.
307 (combine_pending_stack_adjustment_and_call): Take the unadjusted
308 args size as a poly_int64 rather than an int. Return a bool
309 indicating whether the optimization was possible and return
310 the new adjustment by reference.
311 (check_sibcall_argument_overlap): Track polynomail argument sizes.
312 Update stored_args_watermark.
313 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
314 (expand_call): Likewise. Maintain stack_usage_watermark and
315 stored_args_watermark. Update calls to
316 combine_pending_stack_adjustment_and_call.
317 (emit_library_call_value_1): Handle polynomial argument sizes.
318 Call stack_region_maybe_used_p and mark_stack_region_used.
319 Maintain stack_usage_watermark.
320 (store_one_arg): Likewise. Update call to
321 mem_overlaps_already_clobbered_arg_p.
322 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
324 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
325 * config/microblaze/microblaze.c (microblaze_function_prologue):
327 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
328 TARGET_RETURN_POPS_ARGS interface.
329 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
331 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
332 * config/i386/i386.c (ix86_return_pops_args): Update for new
333 TARGET_RETURN_POPS_ARGS interface.
334 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
335 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
336 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
337 TARGET_RETURN_POPS_ARGS interface.
338 * config/vax/vax.c (vax_return_pops_args): Likewise.
339 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
340 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
341 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
342 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
343 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
344 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
345 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
346 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
347 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
348 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
349 and x_stack_pointer_delta from int to poly_int64.
350 * dojump.c (do_pending_stack_adjust): Update accordingly.
351 * explow.c (allocate_dynamic_stack_space): Handle polynomial
352 stack_pointer_deltas.
353 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
354 (pad_to_arg_alignment): Track polynomial offsets.
355 (assign_parm_find_stack_rtl): Likewise.
356 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
357 * toplev.c (output_stack_usage): Update reference to
358 current_function_pushed_stack_size.
360 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
361 Alan Hayward <alan.hayward@arm.com>
362 David Sherwood <david.sherwood@arm.com>
364 * function.c (in_arg_offset, var_offset, dynamic_offset)
365 (out_arg_offset, cfa_offset): Change from int to poly_int64.
366 (instantiate_new_reg): Return the new offset as a poly_int64_pod
367 rather than a HOST_WIDE_INT.
368 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
369 (instantiate_virtual_regs_in_insn): Likewise.
371 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
372 Alan Hayward <alan.hayward@arm.com>
373 David Sherwood <david.sherwood@arm.com>
375 * rtl.h (get_args_size, add_args_size_note): New functions.
376 (find_args_size_adjust): Return a poly_int64 rather than a
378 (fixup_args_size_notes): Likewise. Make the same change to the
379 end_args_size parameter.
380 * rtlanal.c (get_args_size, add_args_size_note): New functions.
381 * builtins.c (expand_builtin_trap): Use add_args_size_note.
382 * calls.c (emit_call_1): Likewise.
383 * explow.c (adjust_stack_1): Likewise.
384 * cfgcleanup.c (old_insns_match_p): Update use of
385 find_args_size_adjust.
386 * combine.c (distribute_notes): Track polynomial arg sizes.
387 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
388 end_true_args_size, beg_delay_args_size and end_delay_args_size
389 from HOST_WIDE_INT to poly_int64.
390 (add_cfi_args_size): Take the args_size as a poly_int64 rather
391 than a HOST_WIDE_INT.
392 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
393 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
394 polynomial arg sizes.
395 * emit-rtl.c (try_split): Use get_args_size.
396 * recog.c (peep2_attempt): Likewise.
397 * reload1.c (reload_as_needed): Likewise.
398 * expr.c (find_args_size_adjust): Return the adjustment as a
399 poly_int64 rather than a HOST_WIDE_INT.
400 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
401 to a poly_int64 and change the return type in the same way.
402 (emit_single_push_insn): Track polynomial arg sizes.
404 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
405 Alan Hayward <alan.hayward@arm.com>
406 David Sherwood <david.sherwood@arm.com>
408 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
409 from HOST_WIDE_INT to poly_int64.
410 * expr.c (push_block, emit_push_insn): Likewise.
412 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
413 Alan Hayward <alan.hayward@arm.com>
414 David Sherwood <david.sherwood@arm.com>
416 * function.h (frame_space): Change start and length from HOST_WIDE_INT
418 (get_frame_size): Return the size as a poly_int64 rather than a
420 (frame_offset_overflow): Take the offset as a poly_int64 rather
421 than a HOST_WIDE_INT.
422 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
423 (assign_stack_temp): Likewise for the size.
424 * function.c (get_frame_size): Return a poly_int64 rather than
426 (frame_offset_overflow): Take the offset as a poly_int64 rather
427 than a HOST_WIDE_INT.
428 (try_fit_stack_local): Take the start, length and size as poly_int64s
429 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
430 rather than a HOST_WIDE_INT.
431 (add_frame_space): Take the start and end as poly_int64s rather than
433 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
434 (assign_stack_temp): Likewise for the size.
435 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
437 (find_temp_slot_from_address): Handle polynomial offsets.
438 (combine_temp_slots): Likewise.
439 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
441 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
442 poly_int64 rather than a HOST_WIDE_INT.
443 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
444 than a HOST_WIDE_INT.
445 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
446 polynomial frame offsets.
447 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
448 as a poly_int64 rather than an int.
449 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
450 * config/v850/v850-protos.h (compute_frame_size): Likewise.
451 * config/v850/v850.c (compute_frame_size): Likewise.
452 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
453 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
454 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
455 * config/pa/pa.c (pa_compute_frame_size): Likewise.
456 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
457 rather than a HOST_WIDE_INT.
458 * explow.c (get_dynamic_stack_base): Likewise.
459 * final.c (final_start_function): Use the constant lower bound
460 of the frame size for -Wframe-larger-than.
461 * ira.c (do_reload): Adjust for new get_frame_size return type.
462 * lra.c (lra): Likewise.
463 * reload1.c (reload): Likewise.
464 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
465 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
466 * rtlanal.c (get_initial_register_offset): Return the offset as
467 a poly_int64 rather than a HOST_WIDE_INT.
469 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
470 Alan Hayward <alan.hayward@arm.com>
471 David Sherwood <david.sherwood@arm.com>
473 * reload1.c (elim_table): Change initial_offset, offset and
474 previous_offset from HOST_WIDE_INT to poly_int64_pod.
475 (offsets_at): Change the target array's element type from
476 HOST_WIDE_INT to poly_int64_pod.
477 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
478 (elimination_costs_in_insn, update_eliminable_offsets)
479 (verify_initial_elim_offsets, set_offsets_for_label)
480 (init_eliminable_invariants): Update after above changes.
482 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
483 Alan Hayward <alan.hayward@arm.com>
484 David Sherwood <david.sherwood@arm.com>
486 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
487 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
488 (decomposition): Change start and end from HOST_WIDE_INT
490 (decompose, immune_p): Update accordingly.
491 (find_inc_amount): Return a poly_int64 rather than an int.
492 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
495 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
496 Alan Hayward <alan.hayward@arm.com>
497 David Sherwood <david.sherwood@arm.com>
499 * tree.h (get_inner_reference): Return the bitsize and bitpos
500 as poly_int64_pods rather than HOST_WIDE_INT.
501 * fold-const.h (ptr_difference_const): Return the pointer difference
502 as a poly_int64_pod rather than a HOST_WIDE_INT.
503 * expr.c (get_inner_reference): Return the bitsize and bitpos
504 as poly_int64_pods rather than HOST_WIDE_INT.
505 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
507 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
508 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
509 (optimize_bit_field_compare): Update call to get_inner_reference.
510 (decode_field_reference): Likewise.
511 (fold_unary_loc): Track polynomial offsets and sizes.
512 (split_address_to_core_and_offset): Return the bitpos as a
513 poly_int64_pod rather than a HOST_WIDE_INT.
514 (ptr_difference_const): Likewise for the pointer difference.
515 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
516 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
517 * dbxout.c (dbxout_expand_expr): Likewise.
518 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
519 (loc_list_from_tree_1, fortran_common): Likewise.
520 * gimple-laddress.c (pass_laddress::execute): Likewise.
521 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
522 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
523 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
524 * tree-affine.c (tree_to_aff_combination): Likewise.
525 (get_inner_reference_aff): Likewise.
526 * tree-data-ref.c (split_constant_offset_1): Likewise.
527 (dr_analyze_innermost): Likewise.
528 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
529 * tree-sra.c (ipa_sra_check_caller): Likewise.
530 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
531 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
532 (instrument_bool_enum_load, instrument_object_size): Likewise.
533 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
534 to get_inner_reference.
535 * hsa-gen.c (gen_hsa_addr): Likewise.
536 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
537 * tsan.c (instrument_expr): Likewise.
538 * match.pd: Update call to ptr_difference_const.
540 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * fold-const.c (fold_comparison): Track sizes and offsets as
545 poly_int64s rather than HOST_WIDE_INTs when folding address
548 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
549 Alan Hayward <alan.hayward@arm.com>
550 David Sherwood <david.sherwood@arm.com>
552 * expr.h (get_bit_range): Return the bitstart and bitend as
553 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
554 as a poly_int64 rather than a HOST_WIDE_INT.
555 * expr.c (get_bit_range): Likewise.
556 (expand_assignment): Update call accordingly.
557 * fold-const.c (optimize_bit_field_compare): Likewise.
559 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
560 Alan Hayward <alan.hayward@arm.com>
561 David Sherwood <david.sherwood@arm.com>
563 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
564 (aarch64_legitimate_address_p): Use it instead of an rtx code,
565 as an optional final parameter.
566 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
567 (aarch64_legitimate_address_p): Likewise.
568 (aarch64_print_address_internal): Take an aarch64_addr_query_type
569 instead of an rtx code.
570 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
571 (aarch64_legitimate_address_hook_p): Likewise.
572 (aarch64_print_ldpstp_address): Likewise.
573 (aarch64_print_operand_address): Likewise.
574 (aarch64_address_cost): Likewise.
575 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
576 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
577 (aarch64_mem_pair_lanes_operand): Likewise.
579 2017-12-20 Richard Biener <rguenther@suse.de>
581 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
582 update_stmt_if_modified.
584 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
586 PR tree-optimization/83491
587 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
588 before walking uses. Improve coding style and comments.
590 2017-12-20 Tom de Vries <tom@codesourcery.com>
592 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
594 2017-12-20 Jakub Jelinek <jakub@redhat.com>
597 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
599 (pass_ipa_free_fn_summary): Add small_p private data member,
600 initialize to false in the ctor.
601 (pass_ipa_free_fn_summary::clone,
602 pass_ipa_free_fn_summary::set_pass_param,
603 pass_ipa_free_fn_summary::gate): New methods.
604 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
605 | TODO_dump_symtab if small_p.
606 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
607 entry and add another instance of the pass with false parm after
609 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
610 ipa_free_fn_summary here.
612 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
614 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
616 2017-12-20 Martin Sebor <msebor@redhat.com>
619 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
620 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
621 (expand_builtin_strncmp): Same.
623 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
626 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
627 allowed debug stmts before labels.
628 (expand_gimple_basic_block): Likewise.
629 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
630 * gimple-iterator.h (gsi_after_labels): Likewise.
631 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
632 rename reused variable, and simplify using gsi_move_before.
633 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
634 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
635 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
636 (gimple_verify_flow_info, gimple_block_label): Likewise.
637 (make_blocks): Move debug markers after adjacent labels.
638 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
639 allowed debug insns outside blocks.
640 * df-scan.c (df_insn_delete): Likewise.
641 * lra-constraints.c (update_ebb_live_info): Likewise.
642 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
643 (vt_initialize, delete_vta_debug_insns): Likewise.
644 (reemit_marker_as_note): Drop BB parm. Adjust callers.
646 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
647 Alan Hayward <alan.hayward@arm.com>
648 David Sherwood <david.sherwood@arm.com>
650 * poly-int-types.h (round_down_to_byte_boundary): New macro.
651 (round_up_to_byte_boundary): Likewise.
652 * expr.h (get_bit_range): Add temporary shim.
653 * gimple-ssa-store-merging.c (store_operand_info): Change the
654 bitsize, bitpos, bitregion_start and bitregion_end fields from
655 unsigned HOST_WIDE_INT to poly_uint64.
656 (merged_store_group): Likewise load_align_base.
657 (compatible_load_p, compatible_load_p): Update accordingly.
658 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
659 (split_group, imm_store_chain_info::output_merged_store): Likewise.
660 (mem_valid_for_store_merging): Return the bitsize, bitpos,
661 bitregion_start and bitregion_end as poly_uint64s rather than
662 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
663 (handled_load): Take the bitsize, bitpos,
664 bitregion_start and bitregion_end as poly_uint64s rather than
665 unsigned HOST_WIDE_INTs.
666 (pass_store_merging::process_store): Update call to
667 mem_valid_for_store_merging.
669 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
670 Alan Hayward <alan.hayward@arm.com>
671 David Sherwood <david.sherwood@arm.com>
673 * builtins.c (get_object_alignment_2): Track polynomial offsets
674 and sizes. Update the alignment handling.
676 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
677 Alan Hayward <alan.hayward@arm.com>
678 David Sherwood <david.sherwood@arm.com>
680 * tree.h (get_inner_reference): Add a version that returns the
681 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
682 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
683 the case in which bitpos is not associated with the first byte.
685 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
686 Alan Hayward <alan.hayward@arm.com>
687 David Sherwood <david.sherwood@arm.com>
689 * tree-affine.h (get_inner_reference_aff): Return the size as a
691 * tree-affine.c (get_inner_reference_aff): Likewise.
692 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
693 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
695 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
696 Alan Hayward <alan.hayward@arm.com>
697 David Sherwood <david.sherwood@arm.com>
699 * fold-const.c (pointer_may_wrap_p): Take the offset as a
700 HOST_WIDE_INT rather than a poly_int64.
702 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
703 Alan Hayward <alan.hayward@arm.com>
704 David Sherwood <david.sherwood@arm.com>
706 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
707 HOST_WIDE_INT to poly_int64_pod.
708 (perform_symbolic_merge): Update accordingly.
709 (bswap_replace): Likewise.
711 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
712 Alan Hayward <alan.hayward@arm.com>
713 David Sherwood <david.sherwood@arm.com>
715 * tree-affine.h (aff_tree::offset): Change from widest_int
717 (wide_int_ext_for_comb): Delete.
718 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
719 constants as poly_widest_int rather than widest_int.
720 (aff_combination_constant_multiple_p): Return the multiplier
721 as a poly_widest_int.
722 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
724 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
725 version static and add an overload for poly_widest_int.
726 (aff_combination_const, aff_combination_add_cst)
727 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
728 the constants as poly_widest_int rather than widest_int.
729 (tree_to_aff_combination): Generalize INTEGER_CST case to
731 (aff_combination_to_tree): Track offsets as poly_widest_ints.
732 (aff_combination_add_product, aff_combination_mult): Handle
734 (aff_combination_constant_multiple_p): Return the multiplier
735 as a poly_widest_int.
736 * tree-predcom.c (determine_offset): Return the offset as a
738 (split_data_refs_to_components, suitable_component_p): Update
740 (valid_initializer_p): Update call to
741 aff_combination_constant_multiple_p.
742 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
743 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
744 as a poly_int64 rather than a HOST_WIDE_INT.
745 (get_address_cost): Handle polynomial offsets.
746 (iv_elimination_compare_lt): Likewise.
747 (rewrite_use_nonlinear_expr): Likewise.
749 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
754 as a poly_int64_pod rather than a HOST_WIDE_INT.
755 (get_addr_base_and_unit_offset): Likewise.
756 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
757 (get_addr_base_and_unit_offset): Likewise.
758 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
759 to poly_int64 in example.
760 * fold-const.c (fold_binary_loc): Update call to
761 get_addr_base_and_unit_offset.
762 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
763 (maybe_canonicalize_mem_ref_addr): Likewise.
764 (gimple_fold_stmt_to_constant_1): Likewise.
765 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
767 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
768 * match.pd: Likewise.
769 * omp-low.c (lower_omp_target): Likewise.
770 * tree-sra.c (build_ref_for_offset): Likewise.
771 (build_debug_ref_for_model): Likewise.
772 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
773 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
774 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
775 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
776 (constant_pointer_difference): Likewise.
777 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
778 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
779 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
780 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
781 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
782 (set_ssa_val_to): Likewise.
783 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
784 (maybe_diag_stxncpy_trunc): Likewise.
785 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
786 * tree.c (build_simple_mem_ref_loc): Likewise.
787 (array_at_struct_end_p): Likewise.
789 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
790 Alan Hayward <alan.hayward@arm.com>
791 David Sherwood <david.sherwood@arm.com>
793 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
794 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
795 (get_ref_base_and_extent_hwi): Declare.
796 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
797 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
798 (get_ref_base_and_extent_hwi): New function.
799 * cfgexpand.c (expand_debug_expr): Update call to
800 get_ref_base_and_extent.
801 * dwarf2out.c (add_var_loc_to_decl): Likewise.
802 * gimple-fold.c (get_base_constructor): Return the offset as a
803 poly_int64_pod rather than a HOST_WIDE_INT.
804 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
805 * ipa-polymorphic-call.c
806 (ipa_polymorphic_call_context::set_by_invariant)
807 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
808 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
809 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
810 rather than get_ref_base_and_extent.
811 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
812 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
813 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
815 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
816 call to get_ref_base_and_extent.
817 * tree-sra.c (create_access, get_access_for_expr): Likewise.
818 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
819 (stmt_kills_ref_p): Likewise.
820 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
821 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
823 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
824 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
825 when calling native_encode_expr.
826 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
827 call to get_ref_base_and_extent.
828 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
829 get_ref_base_and_extent.
830 * var-tracking.c (track_expr_p): Likewise.
832 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
833 Alan Hayward <alan.hayward@arm.com>
834 David Sherwood <david.sherwood@arm.com>
836 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
837 HOST_WIDE_INT to poly_int64_pod.
838 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
839 polynomail parameter offsets.
841 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
842 Alan Hayward <alan.hayward@arm.com>
843 David Sherwood <david.sherwood@arm.com>
845 * gengtype.c (main): Handle poly_int64_pod.
846 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
847 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
848 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
849 from HOST_WIDE_INT to poly_int64_pod.
850 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
851 (copy_cfa): New function.
852 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
853 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
854 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
855 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
856 (initial_return_save): Treat offsets as poly_ints.
857 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
858 offset is nonconstant.
859 (reg_save): Take the offset as a poly_int64. Fall back to
860 DW_CFA_expression for nonconstant offsets.
861 (queue_reg_save): Take the offset as a poly_int64.
862 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
864 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
865 Alan Hayward <alan.hayward@arm.com>
866 David Sherwood <david.sherwood@arm.com>
868 * rtl.h (operand_subword, operand_subword_force): Take the offset
869 as a poly_uint64 an unsigned int.
870 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
872 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
873 Alan Hayward <alan.hayward@arm.com>
874 David Sherwood <david.sherwood@arm.com>
876 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
877 'p' format code. Use INT_LIST rather than SUBREG as the example of
878 a code with an XINT and an XEXP. Remove the implication that
879 accessing an rtx field using XINT is expected to work.
880 * rtl.def (SUBREG): Change format from "ei" to "ep".
881 * rtl.h (rtunion::rt_subreg): New field.
882 (XCSUBREG): New macro.
883 (SUBREG_BYTE): Use it.
884 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
885 Update constructor accordingly.
886 (subreg_shape::operator ==): Update accordingly.
887 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
888 than an unsigned int.
889 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
890 a poly_uint64 rather than an unsigned int.
891 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
892 than an unsigned int.
893 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
894 (subreg_size_highpart_offset): Return a poly_uint64 rather than
895 an unsigned int. Take the sizes as poly_uint64s.
896 (subreg_offset_from_lsb): Return a poly_uint64 rather than
897 an unsigned int. Take the shift as a poly_uint64 rather than
899 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
900 as a poly_uint64 rather than an unsigned int.
901 (simplify_subreg_regno): Likewise.
902 (byte_lowpart_offset): Return the memory offset as a poly_int64
904 (subreg_memory_offset): Likewise. Take the subreg offset as a
905 poly_uint64 rather than an unsigned int.
906 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
907 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
908 poly_uint64 rather than an unsigned int.
909 * rtl.c (rtx_format): Describe 'p' in comment.
910 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
911 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
912 offset as a poly_uint64 rather than an unsigned int.
913 (byte_lowpart_offset): Return the memory offset as a poly_int64
915 (subreg_memory_offset): Likewise. Take the subreg offset as a
916 poly_uint64 rather than an unsigned int.
917 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
918 mode sizes as poly_uint64s rather than unsigned ints. Return a
919 poly_uint64 rather than an unsigned int.
920 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
921 (copy_insn_1): Handle 'p'.
922 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
923 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
924 an unsigned int. Return the shift in the same way.
925 (subreg_lsb): Return the shift as a poly_uint64 rather than an
927 (subreg_size_offset_from_lsb): Take the sizes and shift as
928 poly_uint64s rather than unsigned ints. Return the offset as
930 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
931 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
933 * rtlhash.c (add_rtx): Handle 'p'.
934 * genemit.c (gen_exp): Likewise.
935 * gengenrtl.c (type_from_format, gendef): Likewise.
936 * gensupport.c (subst_pattern_match, get_alternatives_number)
937 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
938 (subst_dup): Likewise.
939 * gengtype.c (adjust_field_rtx_def): Likewise.
940 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
941 (match_pattern_2): Likewise.
942 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
943 (rtx_test::subreg_field): New function.
944 (operator ==, safe_to_hoist_p, transition_parameter_type)
945 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
946 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
947 * genpeep.c (match_rtx): Likewise.
948 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
949 (rtx_writer::print_rtx_operand): Handle 'p'.
950 (print_value): Handle SUBREG.
951 * read-rtl.c (apply_int_iterator): Likewise.
952 (rtx_reader::read_rtx_operand): Handle 'p'.
953 * alias.c (rtx_equal_for_memref_p): Likewise.
954 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
955 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
957 * calls.c (expand_call): Likewise.
958 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
959 (make_extraction, gen_lowpart_for_combine): Likewise.
960 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
962 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
963 rather than an unsigned int. Treat subreg offsets as poly_ints.
964 (exp_equiv_p): Handle 'p'.
965 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
966 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
967 * dse.c (find_shift_sequence): Likewise.
968 * dwarf2out.c (rtl_for_decl_location): Likewise.
969 * expmed.c (extract_low_bits): Likewise.
970 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
971 (expand_expr_real_2): Likewise.
972 * final.c (alter_subreg): Likewise.
973 (leaf_renumber_regs_insn): Handle 'p'.
974 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
975 Treat subreg offsets as poly_ints.
976 * fwprop.c (forward_propagate_and_simplify): Likewise.
977 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
978 * ira.c (get_subreg_tracking_sizes): Likewise.
979 * ira-conflicts.c (go_through_subreg): Likewise.
980 * ira-lives.c (process_single_reg_class_operands): Likewise.
981 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
982 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
983 as a poly_uint64 rather than an unsigned int.
984 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
985 subreg offsets as poly_ints.
986 * lra-constraints.c (operands_match_p): Handle 'p'.
987 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
988 * lra-spills.c (assign_mem_slot): Likewise.
989 * postreload.c (move2add_valid_value_p): Likewise.
990 * recog.c (general_operand, indirect_operand): Likewise.
991 * regcprop.c (copy_value, maybe_mode_change): Likewise.
992 (copyprop_hardreg_forward_1): Likewise.
993 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
994 (record_subregs_of_mode): Likewise.
995 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
996 * reload.c (operands_match_p): Handle 'p'.
997 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
998 * reload1.c (alter_reg, choose_reload_regs): Likewise.
999 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1000 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1001 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1002 (simplify_const_poly_int_tests<N>::run): Likewise.
1003 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1004 a poly_uint64 rather than an unsigned int.
1005 * valtrack.c (debug_lowpart_subreg): Likewise.
1006 * var-tracking.c (var_lowpart): Likewise.
1007 (loc_cmp): Handle 'p'.
1009 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1010 Alan Hayward <alan.hayward@arm.com>
1011 David Sherwood <david.sherwood@arm.com>
1013 * ira.c (get_subreg_tracking_sizes): New function.
1014 (init_live_subregs): Take an integer size rather than a register.
1015 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1016 to init_live_subregs.
1018 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1019 Alan Hayward <alan.hayward@arm.com>
1020 David Sherwood <david.sherwood@arm.com>
1022 * expr.c (store_constructor_field): Change bitsize from a
1023 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1024 HOST_WIDE_INT to a poly_int64.
1025 (store_constructor): Change size from a HOST_WIDE_INT to
1027 (store_field): Likewise bitsize and bitpos.
1029 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1030 Alan Hayward <alan.hayward@arm.com>
1031 David Sherwood <david.sherwood@arm.com>
1033 * expmed.h (store_bit_field): Change bitregion_start and
1034 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1035 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1036 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1037 (store_fixed_bit_field, store_split_bit_field): Likewise.
1038 * expr.c (store_constructor_field, store_field): Likewise.
1039 (optimize_bitfield_assignment_op): Likewise. Make the same change
1040 to bitsize and bitpos.
1041 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1042 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1043 same change in the constructor arguments.
1044 (get_best_mode): Change bitregion_start and bitregion_end from
1045 unsigned HOST_WIDE_INT to poly_uint64.
1046 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1047 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1049 (bit_field_mode_iterator::next_mode): Update for new types
1050 of m_bitregion_start and m_bitregion_end.
1051 (get_best_mode): Change bitregion_start and bitregion_end from
1052 unsigned HOST_WIDE_INT to poly_uint64.
1054 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1055 Alan Hayward <alan.hayward@arm.com>
1056 David Sherwood <david.sherwood@arm.com>
1058 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1059 accepts poly_uint64 offsets.
1060 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1061 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1062 * expmed.c (lowpart_bit_field_p): Likewise.
1063 (extract_bit_field_as_subreg): New function, split out from...
1064 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1065 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1066 extractions, check that BITSIZE matches the size of the extracted
1067 value and that BITNUM is an exact multiple of that size.
1068 If all else fails, try forcing the value into memory if
1069 BITNUM is variable, and adjusting the address so that the
1070 offset is constant. Split the part that can only handle constant
1071 bitsize and bitnum out into...
1072 (extract_integral_bit_field): ...this new function.
1073 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1074 rather than unsigned HOST_WIDE_INTs.
1076 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1077 Alan Hayward <alan.hayward@arm.com>
1078 David Sherwood <david.sherwood@arm.com>
1080 * expmed.h (store_bit_field): Take bitsize and bitnum as
1081 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1082 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1083 that returns the byte size.
1084 (store_bit_field_1): Take bitsize and bitnum as
1085 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1086 to simple_mem_bitfield_p. Split the part that can only handle
1087 constant bitsize and bitnum out into...
1088 (store_integral_bit_field): ...this new function.
1089 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1090 than unsigned HOST_WIDE_INTs.
1091 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1093 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1094 Alan Hayward <alan.hayward@arm.com>
1095 David Sherwood <david.sherwood@arm.com>
1097 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1098 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1100 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1101 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1102 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1103 offset as a poly_int64 rather than an int.
1104 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1105 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1106 * lra-constraints.c (equiv_address_substitution): Track offsets
1108 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1109 (curr_insn_transform): Handle the new form of sp_offset.
1110 * lra-eliminations.c (lra_elim_table): Change previous_offset
1111 and offset from HOST_WIDE_INT to poly_int64.
1112 (print_elim_table, update_reg_eliminate): Update accordingly.
1113 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1114 (get_elimination): Update accordingly.
1115 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1116 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1117 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1118 poly_int64 offsets generally.
1119 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1120 (mark_not_eliminable, init_elimination): Update accordingly.
1121 (remove_reg_equal_offset_note): Return a bool and pass the new
1122 offset back by pointer as a poly_int64.
1123 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1124 rather than a HOST_WIDE_INT.
1125 (do_remat): Track offsets poly_int64s.
1126 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1128 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1129 Alan Hayward <alan.hayward@arm.com>
1130 David Sherwood <david.sherwood@arm.com>
1132 * rtl.h (mem_attrs): Add a default constructor. Change size and
1133 offset from HOST_WIDE_INT to poly_int64.
1134 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1135 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1136 (widen_memory_access): Take the sizes and offsets as poly_int64s
1137 rather than HOST_WIDE_INTs.
1138 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1139 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1141 (adjust_offset_for_component_ref): Change the offset from a
1142 HOST_WIDE_INT to a poly_int64.
1143 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1144 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1145 * dce.c (find_call_stack_args): Likewise.
1146 * dse.c (record_store): Likewise.
1147 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1148 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1149 * read-rtl-function.c (test_loading_mem): Likewise.
1150 * rtlanal.c (may_trap_p_1): Likewise.
1151 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1152 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1153 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1154 (mem_attrs::mem_attrs): New function.
1155 (set_mem_attributes_minus_bitpos): Change bitpos from a
1156 HOST_WIDE_INT to poly_int64.
1157 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1158 (clear_mem_offset, clear_mem_size, change_address)
1159 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1160 initialize mem_attrs.
1161 (set_mem_offset, set_mem_size, adjust_address_1)
1162 (adjust_automodify_address_1, offset_address, widen_memory_access):
1163 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1165 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1166 Alan Hayward <alan.hayward@arm.com>
1167 David Sherwood <david.sherwood@arm.com>
1169 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1170 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1171 rather than 0 to represent an unknown size. Assert that the size
1172 is known when the mode isn't BLKmode.
1173 (may_trap_p_1): Use -1 for unknown sizes.
1174 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1176 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1177 Alan Hayward <alan.hayward@arm.com>
1178 David Sherwood <david.sherwood@arm.com>
1180 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1181 to poly_int64. Update commentary for positions_needed.large.
1182 (read_info_type): Change offset and width from HOST_WIDE_INT
1184 (set_usage_bits): Likewise.
1185 (canon_address): Return the offset as a poly_int64 rather than
1186 a HOST_WIDE_INT. Use strip_offset_and_add.
1187 (set_all_positions_unneeded, any_positions_needed_p): Use
1188 positions_needed.large to track stores with non-constant widths.
1189 (all_positions_needed_p): Likewise. Take the offset and width
1190 as poly_int64s rather than ints. Assert that rhs is nonnull.
1191 (record_store): Cope with non-constant offsets and widths.
1192 Nullify the rhs of an earlier store if we can't tell which bytes
1194 (find_shift_sequence): Take the access_size and shift as poly_int64s
1196 (get_stored_val): Take the read_offset and read_width as poly_int64s
1197 rather than HOST_WIDE_INTs.
1198 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1199 non-constant offsets and widths.
1201 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * inchash.h (inchash::hash::add_poly_int): New function.
1206 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1207 Use poly_int64 rather than HOST_WIDE_INT.
1208 (ao_ref::max_size_known_p): New function.
1209 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1210 rather than HOST_WIDE_INT.
1211 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1212 to temporaries until its interface is adjusted to match.
1213 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1214 (aliasing_component_refs_p, decl_refs_may_alias_p)
1215 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1216 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1217 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1219 * alias.c (ao_ref_from_mem): Likewise.
1220 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1221 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1222 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1223 (maybe_trim_complex_store, maybe_trim_constructor_store)
1224 (live_bytes_read, dse_classify_store): Likewise.
1225 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1226 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1227 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1228 (vn_reference_lookup_3): Likewise.
1229 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1231 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1232 Alan Hayward <alan.hayward@arm.com>
1233 David Sherwood <david.sherwood@arm.com>
1235 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1236 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1237 instead of ranges_overlap_p.
1239 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1244 sizes as poly_int64s rather than HOST_WIDE_INTs.
1246 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1247 Alan Hayward <alan.hayward@arm.com>
1248 David Sherwood <david.sherwood@arm.com>
1250 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1251 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1252 * gimple-fold.c (fold_ctor_reference): Likewise.
1254 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1255 Alan Hayward <alan.hayward@arm.com>
1256 David Sherwood <david.sherwood@arm.com>
1258 * target.def (dwarf_poly_indeterminate_value): New hook.
1259 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1260 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1261 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1262 * doc/tm.texi: Regenerate.
1263 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1264 offset as a poly_int64.
1265 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1266 offset as a poly_int64.
1267 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1268 Take the offset as a poly_int64.
1269 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1270 (frame_pointer_fb_offset): Change to a poly_int64.
1271 (int_loc_descriptor): Take the offset as a poly_int64. Use
1272 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1273 (based_loc_descr): Take the offset as a poly_int64.
1274 Use strip_offset_and_add to handle (plus X (const)).
1275 Use new_reg_loc_descr instead of an open-coded version of the
1276 previous implementation.
1277 (mem_loc_descriptor): Handle CONST_POLY_INT.
1278 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1279 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1281 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1282 Alan Hayward <alan.hayward@arm.com>
1283 David Sherwood <david.sherwood@arm.com>
1285 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1286 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1287 * inchash.h (inchash::hash::add_poly_hwi): New function.
1288 * gengtype.c (main): Register poly_int64.
1289 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1290 offset as a poly_int.
1291 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1292 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1293 offset as a poly_int64.
1294 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1295 * print-rtl.c (print_poly_int): New function.
1296 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1298 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1300 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1301 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1302 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1303 Enforce track_offset_p for parts of a PARALLEL.
1304 (vt_add_function_parameter): Use const_offset for the final
1305 offset to track. Use get_tracked_reg_offset for the parts
1308 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1309 Alan Hayward <alan.hayward@arm.com>
1310 David Sherwood <david.sherwood@arm.com>
1312 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1313 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1314 * doc/tm.texi: Regenerate.
1315 * hooks.h (hook_bool_uint_uint_true): Delete.
1316 (hook_bool_puint64_puint64_true): Declare.
1317 * hooks.c (hook_bool_uint_uint_true): Delete.
1318 (hook_bool_puint64_puint64_true): New function.
1319 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1320 instead of unsigned ints.
1321 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1322 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1324 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1325 Alan Hayward <alan.hayward@arm.com>
1326 David Sherwood <david.sherwood@arm.com>
1328 * optabs.h (expand_operand): Add an int_value field.
1329 (create_expand_operand): Add an int_value parameter and use it
1330 to initialize the new expand_operand field.
1331 (create_integer_operand): Replace with a declaration of a function
1332 that accepts poly_int64s. Move the implementation to...
1333 * optabs.c (create_integer_operand): ...here.
1334 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1335 the mode preserves the value of int_value, instead of calling
1336 const_int_operand on the rtx. Use gen_int_mode to generate
1339 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1340 Alan Hayward <alan.hayward@arm.com>
1341 David Sherwood <david.sherwood@arm.com>
1343 * dumpfile.h (dump_dec): Declare.
1344 * dumpfile.c (dump_dec): New function.
1345 * pretty-print.h (pp_wide_integer): Turn into a function and
1346 declare a poly_int version.
1347 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1349 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1350 Alan Hayward <alan.hayward@arm.com>
1351 David Sherwood <david.sherwood@arm.com>
1353 * doc/generic.texi (POLY_INT_CST): Document.
1354 * tree.def (POLY_INT_CST): New tree code.
1355 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1356 * tree-core.h (tree_poly_int_cst): New struct.
1357 (tree_node): Add a poly_int_cst field.
1358 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1359 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1360 instead of a wide_int_ref.
1361 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1363 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1364 instead of an unsigned HOST_WIDE_INT.
1365 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1366 (ptrdiff_tree_p): Declare.
1367 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1368 extern inline implementations if the target doesn't use POLY_INT_CST.
1369 (poly_int_tree_p): New function.
1370 (wi::unextended_tree): New class.
1371 (wi::int_traits <unextended_tree>): New override.
1372 (wi::extended_tree): Add a default constructor.
1373 (wi::extended_tree::get_tree): New function.
1374 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1375 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1376 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1377 (wi::tree_to_poly_wide_ref): New typedefs.
1378 (wi::ints_for): Provide overloads for extended_tree and
1380 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1381 (wi::to_wide): New functions.
1382 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1383 * tree.c (poly_int_cst_hasher): New struct.
1384 (poly_int_cst_hash_table): New variable.
1385 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1386 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1388 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1389 (init_ttree): Initialize poly_int_cst_hash_table.
1390 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1391 a poly_int64 instead of a HOST_WIDE_INT.
1392 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1393 instead of an unsigned HOST_WIDE_INT.
1394 (wide_int_to_tree): Rename to...
1395 (wide_int_to_tree_1): ...this.
1396 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1397 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1398 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1399 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1400 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1401 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1403 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1404 (streamer_read_tree_body): Likewise.
1405 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1406 (streamer_write_tree_body): Likewise.
1407 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1408 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1409 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1410 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1411 * gimple-expr.h (is_gimple_constant): Likewise.
1412 * gimplify.c (maybe_with_size_expr): Likewise.
1413 * print-tree.c (print_node): Likewise.
1414 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1415 * tree-pretty-print.c (dump_generic_node): Likewise.
1416 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1417 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1418 * tree-vrp.c (compare_values_warnv): Likewise.
1419 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1420 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1421 (force_expr_to_var_cost): Likewise.
1422 * tree-ssa-loop.c (for_each_index): Likewise.
1423 * fold-const.h (build_invariant_address, size_int_kind): Take a
1424 poly_int64 instead of a HOST_WIDE_INT.
1425 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1426 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1428 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1429 (int_const_binop_2): New function, split out from...
1430 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1431 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1432 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1433 * gimple-ssa-strength-reduction.c (slsr_process_add)
1434 (slsr_process_mul): Check for INTEGER_CSTs before using them
1436 * stor-layout.c (bits_from_bytes): New function.
1437 (bit_from_pos): Use it.
1438 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1439 by BITS_PER_UNIT to get the TYPE_SIZE.
1440 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1441 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1443 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1444 Alan Hayward <alan.hayward@arm.com>
1445 David Sherwood <david.sherwood@arm.com>
1447 * doc/rtl.texi (const_poly_int): Document. Also document the
1448 rtl sharing behavior.
1449 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1450 * rtl.h (const_poly_int_def): New struct.
1451 (rtx_def::u): Add a cpi field.
1452 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1453 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1454 (wi::rtx_to_poly_wide_ref): New typedef
1455 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1456 (poly_int_rtx_p): New functions.
1457 (trunc_int_for_mode): Declare a poly_int64 version.
1458 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1459 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1461 (strip_offset): Declare.
1462 (strip_offset_and_add): New function.
1463 * rtl.def (CONST_POLY_INT): New rtx code.
1464 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1465 (shared_const_p): Use poly_int_rtx_p.
1466 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1468 (gen_int_shift_amount): Likewise.
1469 * emit-rtl.c (const_poly_int_hasher): New class.
1470 (const_poly_int_htab): New variable.
1471 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1472 (const_poly_int_hasher::hash): New function.
1473 (const_poly_int_hasher::equal): Likewise.
1474 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1475 (immed_wide_int_const): Rename to...
1476 (immed_wide_int_const_1): ...this and make static.
1477 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1478 instead of a wide_int_ref.
1479 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1480 (gen_lowpart_common): Handle CONST_POLY_INT.
1481 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1482 * cselib.c (cselib_hash_rtx): Likewise.
1483 * dwarf2out.c (const_ok_for_output_1): Likewise.
1484 * expr.c (convert_modes): Likewise.
1485 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1486 * rtlhash.c (add_rtx): Likewise.
1487 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1488 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1489 Handle existing CONST_POLY_INT rtxes.
1490 * expmed.h (expand_shift): Take a poly_int64 instead of a
1492 * expmed.c (expand_shift): Likewise.
1493 * rtlanal.c (strip_offset): New function.
1494 (commutative_operand_precedence): Give CONST_POLY_INT the same
1495 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1497 * rtl-tests.c (const_poly_int_tests): New struct.
1498 (rtl_tests_c_tests): Use it.
1499 * simplify-rtx.c (simplify_const_unary_operation): Handle
1501 (simplify_const_binary_operation): Likewise.
1502 (simplify_binary_operation_1): Fold additions of symbolic constants
1503 and CONST_POLY_INTs.
1504 (simplify_subreg): Handle extensions and truncations of
1506 (simplify_const_poly_int_tests): New struct.
1507 (simplify_rtx_c_tests): Use it.
1508 * wide-int.h (storage_ref): Add default constructor.
1509 (wide_int_ref_storage): Likewise.
1510 (trailing_wide_ints): Use GTY((user)).
1511 (trailing_wide_ints::operator[]): Add a const version.
1512 (trailing_wide_ints::get_precision): New function.
1513 (trailing_wide_ints::extra_size): Likewise.
1515 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * emit-rtl.h (gen_int_shift_amount): Declare.
1520 * emit-rtl.c (gen_int_shift_amount): New function.
1521 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1523 * calls.c (shift_return_value): Likewise.
1524 * cse.c (fold_rtx): Likewise.
1525 * dse.c (find_shift_sequence): Likewise.
1526 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1527 (expand_shift, expand_smod_pow2): Likewise.
1528 * lower-subreg.c (shift_cost): Likewise.
1529 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1530 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1531 (expand_vec_perm_var): Likewise.
1532 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1533 (simplify_binary_operation_1): Likewise.
1534 * combine.c (try_combine, find_split_point, force_int_to_mode)
1535 (simplify_shift_const_1, simplify_shift_const): Likewise.
1536 (change_zero_ext): Likewise. Use simplify_gen_binary.
1538 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1540 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1542 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1544 * doc/invoke.texi (ARM Options): Document accepted extension options
1545 for -march=armv8.3-a.
1547 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1550 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1551 or --with-float=softfp, set the default CPU to arm10e.
1553 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1555 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1556 * config/visium/predicates.md (const_shift_operand): Likewise.
1557 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1558 (visium_legitimize_reload_address): Likewise.
1560 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1563 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1565 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1568 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1570 2017-12-20 Julia Koval <julia.koval@intel.com>
1572 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1573 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1574 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1575 * config.gcc: Include vpclmulqdqintrin.h.
1576 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1577 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1579 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1580 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1581 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1582 * config/i386/i386.c (isa2_opts): Add -mcx16.
1583 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1584 (ix86_option_override_internal): Move mcx16 to flags2.
1585 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1586 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1587 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1588 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1589 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1590 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1591 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1592 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1593 * doc/invoke.texi: Add -mvpclmulqdq.
1595 2017-12-20 Tom de Vries <tom@codesourcery.com>
1598 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1599 * calls.c (rtx_for_static_chain): ... here. New function.
1600 * calls.h (rtx_for_static_chain): Declare.
1601 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1602 instead of targetm.calls.static_chain.
1603 * df-scan.c (df_get_entry_block_def_set): Same.
1605 2017-12-19 Tom de Vries <tom@codesourcery.com>
1607 PR tree-optimization/83493
1608 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1611 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1613 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1614 inform with hardcoded english plural handling.
1616 2017-12-18 Jeff Law <law@redhat.com>
1618 PR tree-optimization/83477
1619 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1620 a non-virtual PHI, always push a new range.
1622 2017-12-19 Martin Sebor <msebor@redhat.com>
1625 * builtins.c (compute_objsize): Handle non-constant offsets.
1627 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1629 PR tree-optimization/83444
1630 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1631 character load case, if get_stridx on MEM_REF's operand doesn't
1632 look usable, retry with get_addr_stridx.
1634 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1637 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1638 (variable_tracking_main_1): Keep markers even when VTA fails.
1641 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1642 even if there are markers before it.
1643 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1645 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1647 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1648 typical order conditions.
1649 * sel-sched.c (extract_new_fences_from): Likewise.
1650 * config/visium/constraints.md (J, K, L): Likewise.
1651 * config/visium/predicates.md (const_shift_operand): Likewise.
1652 * config/visium/visium.c (visium_legitimize_address,
1653 visium_legitimize_reload_address): Likewise.
1654 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1655 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1656 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1657 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1658 SET_DUMP_DETAIL): Likewise.
1659 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1660 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1661 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1662 avr_set_core_architecture, avr_set_current_function,
1663 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1664 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1665 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1666 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1667 avr_map_decompose, avr_fold_builtin): Likewise.
1668 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1669 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1670 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1671 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1672 * config/m32c/m32c.c (m32c_conditional_register_usage,
1673 m32c_address_cost): Likewise.
1674 * config/m32c/predicates.md (shiftcount_operand,
1675 longshiftcount_operand): Likewise.
1676 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1677 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1678 can_use_cdx_ldstw): Likewise.
1679 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1680 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1682 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1683 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1684 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1685 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1686 * config/vax/vax.c (adjacent_operands_p): Likewise.
1687 * config/ft32/constraints.md (L, b, KA): Likewise.
1688 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1690 * cfgexpand.c (expand_stack_alignment): Likewise.
1691 * gcse.c (insert_expr_in_table): Likewise.
1692 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1693 * cgraphunit.c (cgraph_node::expand): Likewise.
1694 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1695 * emit-rtl.c (add_insn): Likewise.
1696 * input.c (dump_location_info): Likewise.
1697 * passes.c (NEXT_PASS): Likewise.
1698 * read-rtl-function.c (parse_note_insn_name,
1699 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1701 * sched-rgn.c (sched_rgn_init): Likewise.
1702 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1703 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1704 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1705 * explow.c (eliminate_constant_term): Likewise.
1706 * final.c (leaf_renumber_regs_insn): Likewise.
1707 * cfgrtl.c (print_rtl_with_bb): Likewise.
1708 * genhooks.c (emit_init_macros): Likewise.
1709 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1710 * tree-data-ref.c (conflict_fn): Likewise.
1711 * selftest.c (assert_streq): Likewise.
1712 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1713 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1714 fold_binary_loc, multiple_of_p): Likewise.
1715 * reload.c (push_reload, find_equiv_reg): Likewise.
1716 * et-forest.c (et_nca, et_below): Likewise.
1717 * dbxout.c (dbxout_symbol_location): Likewise.
1718 * reorg.c (relax_delay_slots): Likewise.
1719 * dojump.c (do_compare_rtx_and_jump): Likewise.
1720 * gengtype-parse.c (type): Likewise.
1721 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1722 simplify_const_relational_operation): Likewise.
1723 * reload1.c (do_output_reload): Likewise.
1724 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1725 * gengtype.c (type_for_name): Likewise.
1726 * gimple-ssa-sprintf.c (format_directive): Likewise.
1728 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1731 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1732 accessing it. Adjust comment.
1734 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1737 * predict.c (zero_one_minusone): New function.
1738 (apply_return_prediction): Avoid return prediction for functions
1739 returning only -1, 0 and 1 values, unless they only return -1 and 0
1742 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1744 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1745 fall-through warning.
1746 (arc_compute_frame_size): Remove unused variables.
1747 (arc_print_operand): Fix fprintif format.
1748 (arc_can_follow_jump): Clean fall-through warning.
1750 2017-12-19 Marek Polacek <polacek@redhat.com>
1753 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1756 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1758 * config/arc/arc.c (overriderregs): New variable.
1759 (arc_override_options): Track fixed/call saved/call options.
1760 (arc_conditional_register_usage): Check against overrideregs
1761 variable whenever we change register properties.
1763 2017-12-19 Nathan Sidwell <nathan@acm.org>
1765 * opts.c (finish_options): Don't prefix dump_base_name if it
1766 already contains directories.
1768 2017-12-19 Martin Liska <mliska@suse.cz>
1770 PR rtl-optimization/82675
1771 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1772 more element in sbitmap.
1774 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1776 * gimplify.c (gimplify_expr): Use error_operand_p.
1778 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1781 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1784 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1786 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1787 the jump is to a label.
1789 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1791 PR tree-optimization/83444
1792 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1797 * ipa-inline.c (flatten_remove_node_hook): New function.
1798 (ipa_inline): Keep only nodes with flatten attribute at the end of
1799 the array in the order from ipa_reverse_postorder, only walk that
1800 portion of array for flattening, if there is more than one such
1801 node, temporarily register a removal hook and ignore removed nodes.
1803 PR tree-optimization/80631
1804 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1805 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1806 IFN_REDUC_MAX or IFN_REDUC_MIN.
1808 2017-12-18 Jeff Law <law@redhat.com>
1810 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1811 record symbolic equivalences from backedges in the CFG.
1814 2017-11-19 Jeff Law <law@redhat.com>
1816 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1817 of degenerates resulting from ignoring an edge.
1819 2017-12-18 Martin Sebor <msebor@redhat.com>
1822 PR tree-optimization/78450
1823 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1824 (handle_builtin_strlen): Call it.
1826 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1828 PR rtl-optimization/83424
1829 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1831 2017-12-18 Marek Polacek <polacek@redhat.com>
1834 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1835 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1838 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1840 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1841 (umaddsidi4, umaddsidi_split): Likewise.
1843 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1845 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1848 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1851 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1853 2017-12-18 Richard Biener <rguenther@suse.de>
1855 PR tree-optimization/81877
1856 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1857 (outermost_indep_loop): Adjust.
1858 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1859 (can_sm_ref_p): Adjust.
1861 2017-12-18 Richard Biener <rguenther@suse.de>
1864 * tree.c (array_at_struct_end_p): Return true if the underlying
1865 object has space for at least one element in excess of what
1866 the array domain specifies.
1868 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1870 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1871 example to avoid overfull hbox.
1872 * doc/invoke.texi (Option Summary): Add missing @gol.
1873 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1876 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1877 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1879 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1880 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1881 (AArch64 Options): Document -mverbose-cost-dump.
1882 (ARM Options): Likewise, plus -mflip-thumb.
1884 2017-12-17 Martin Sebor <msebor@redhat.com>
1887 * gimple-ssa-warn-restrict.c
1888 (builtin_memref::offset_out_of_bounds): Correct the handling of
1891 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1893 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1895 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1897 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1899 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1901 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1904 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1906 * doc/invoke.texi: Fix some typos.
1908 2017-12-16 Martin Sebor <msebor@redhat.com>
1910 PR tree-optimization/78918
1911 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1912 * builtins.c (check_sizes): Rename...
1913 (check_access): ...to this. Rename function arguments for clarity.
1914 (check_memop_sizes): Adjust names.
1915 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1916 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1917 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1918 (check_strncat_sizes, expand_builtin_strncat): Same.
1919 (expand_builtin_strncpy, expand_builtin_memset): Same.
1920 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1921 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1922 (maybe_emit_sprintf_chk_warning): Same.
1923 (expand_builtin_strcpy): Adjust.
1924 (expand_builtin_stpcpy): Same.
1925 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1926 in pointer-checking forms of memcpy, memmove, and mempcpy.
1927 (gcall_to_tree_minimal, max_object_size): Define new functions.
1928 * builtins.h (max_object_size): Declare.
1929 * calls.c (alloc_max_size): Call max_object_size instead of
1930 hardcoding ssizetype limit.
1931 (get_size_range): Handle new argument.
1932 * calls.h (get_size_range): Add a new argument.
1933 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1934 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1935 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1937 (gimple_fold_builtin_memory_chk): Same.
1938 (gimple_fold_builtin_stxcpy_chk): New function.
1939 * gimple-ssa-warn-restrict.c: New source.
1940 * gimple-ssa-warn-restrict.h: New header.
1941 * gimple.c (gimple_build_call_from_tree): Propagate location.
1942 * passes.def (pass_warn_restrict): Add new pass.
1943 * tree-pass.h (make_pass_warn_restrict): Declare.
1944 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1946 (handle_builtin_strcat): Same.
1947 (strlen_optimize_stmt): Rename...
1948 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1949 stpncpy, strncpy, and their checking forms.
1951 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1953 PR rtl-optimization/82849
1954 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1955 and get_max_loop_iterations_int.
1957 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1958 Alan Hayward <alan.hayward@arm.com>
1959 David Sherwood <david.sherwood@arm.com>
1961 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1962 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1963 size as a poly_uint64.
1964 (mode_for_vector, mode_for_int_vector): Take the number of vector
1965 elements as a poly_uint64.
1966 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1967 size as a poly_uint64.
1968 (mode_for_vector, mode_for_int_vector): Take the number of vector
1969 elements as a poly_uint64.
1971 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1972 Alan Hayward <alan.hayward@arm.com>
1973 David Sherwood <david.sherwood@arm.com>
1975 * machmode.h (MACRO_MODE): New macro.
1976 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1977 * targhooks.c (default_libcall_value, default_secondary_reload)
1978 (default_memory_move_cost, default_register_move_cost)
1979 (default_class_max_nregs): Likewise.
1981 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1982 Alan Hayward <alan.hayward@arm.com>
1983 David Sherwood <david.sherwood@arm.com>
1985 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1986 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1988 * genautomata.c (main): Likewise.
1989 * genconditions.c (write_header): Likewise.
1990 * genemit.c (main): Likewise.
1991 * genextract.c (print_header): Likewise.
1992 * genopinit.c (main): Likewise.
1993 * genoutput.c (output_prologue): Likewise.
1994 * genpeep.c (main): Likewise.
1995 * genpreds.c (write_insn_preds_c): Likewise.
1996 * genrecog.c (writer_header): Likewise.
1997 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1998 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1999 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2000 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2001 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2002 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2003 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2004 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2005 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2006 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2007 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2008 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2009 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2010 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2011 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2012 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2013 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2014 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2015 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2016 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2017 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2018 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2019 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2020 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2021 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2022 * config/darwin.c (IN_TARGET_CODE): Likewise.
2023 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2024 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2025 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2026 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2027 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2028 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2029 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2030 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2031 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2032 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2033 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2034 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2035 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2036 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2037 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2038 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2039 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2040 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2041 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2042 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2043 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2044 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2045 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2046 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2047 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2048 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2049 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2050 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2051 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2052 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2053 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2054 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2055 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2056 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2057 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2058 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2059 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2060 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2061 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2062 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2063 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2064 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2065 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2066 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2067 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2068 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2069 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2070 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2071 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2072 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2073 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2074 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2075 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2076 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2077 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2078 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2079 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2080 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2081 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2082 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2083 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2084 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2085 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2086 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2087 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2088 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2089 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2090 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2091 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2092 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2093 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2094 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2095 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2096 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2097 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2098 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2099 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2100 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2101 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2102 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2103 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2104 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2105 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2106 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2107 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2108 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2109 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2110 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2111 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2112 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2113 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2114 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2115 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2116 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2117 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2118 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2119 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2120 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2121 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2122 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2123 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2124 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2125 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2126 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2127 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2128 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2130 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2131 Alan Hayward <alan.hayward@arm.com>
2132 David Sherwood <david.sherwood@arm.com>
2134 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2137 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2138 Alan Hayward <alan.hayward@arm.com>
2139 David Sherwood <david.sherwood@arm.com>
2141 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2142 * doc/md.texi (vec_series@var{m}): Document.
2143 * tree.def (VEC_SERIES_EXPR): New tree code.
2144 * tree.h (build_vec_series): Declare.
2145 * tree.c (build_vec_series): New function.
2146 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2147 * tree-pretty-print.c (dump_generic_node): Likewise.
2148 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2149 * tree-inline.c (estimate_operator_cost): Likewise.
2150 * expr.c (expand_expr_real_2): Likewise.
2151 * optabs-tree.c (optab_for_tree_code): Likewise.
2152 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2153 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2154 * expmed.c (make_tree): Handle VEC_SERIES.
2155 * optabs.def (vec_series_optab): New optab.
2156 * optabs.h (expand_vec_series_expr): Declare.
2157 * optabs.c (expand_vec_series_expr): New function.
2158 * tree-vect-generic.c (expand_vector_operations_1): Check that
2159 the operands also have vector type.
2161 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2162 Alan Hayward <alan.hawyard@arm.com>
2163 David Sherwood <david.sherwood@arm.com>
2165 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2166 (VEC_COND_EXPR): Add missing @tindex.
2167 * doc/md.texi (vec_duplicate@var{m}): Document.
2168 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2169 * tree.c (build_vector_from_val): Add stubbed-out handling of
2170 variable-length vectors, using VEC_DUPLICATE_EXPR.
2171 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2172 * cfgexpand.c (expand_debug_expr): Likewise.
2173 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2174 * tree-inline.c (estimate_operator_cost): Likewise.
2175 * tree-pretty-print.c (dump_generic_node): Likewise.
2176 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2177 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2178 (test_vec_duplicate_folding): New function.
2179 (fold_const_c_tests): Call it.
2180 * optabs.def (vec_duplicate_optab): New optab.
2181 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2182 * optabs.h (expand_vector_broadcast): Declare.
2183 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2184 vec_duplicate_optab.
2185 * expr.c (store_constructor): Try using vec_duplicate_optab for
2187 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2189 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2192 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2193 div/mod latencies a bit.
2195 2017-12-15 Jeff Law <law@redhat.com>
2197 PR tree-optimization/36550
2198 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2199 (mark_threaded_blocks): Rewrite code to avoid block copying when
2200 optimizing for size. Don't pessimize blocks which will be
2201 copied, but all the statements will be dead.
2203 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2205 PR tree-optimization/81165
2206 * tree-ssa-threadupdate.c (uses_in_bb): New.
2207 (estimate_threading_killed_stmts): New.
2208 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2209 * tree-ssa-threadedge.c
2210 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2213 2017-12-15 Jeff Law <law@redhat.com>
2215 PR tree-optimization/83410
2216 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2217 threads when parallelizing loops.
2219 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2221 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2223 * config/alpha/alpha.c (vms_attribute_table): Swap
2224 affects_type_identity and handler fields, adjust comments.
2225 * config/mips/mips.c (mips_attribute_table): Likewise.
2226 * config/visium/visium.c (visium_attribute_table): Likewise.
2227 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2228 * config/microblaze/microblaze.c (microblaze_attribute_table):
2230 * config/spu/spu.c (spu_attribute_table): Likewise.
2231 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2232 * config/arc/arc.c (arc_attribute_table): Likewise.
2233 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2234 * config/v850/v850.c (v850_handle_interrupt_attribute,
2235 v850_handle_data_area_attribute): Formatting fixes.
2236 (v850_attribute_table): Swap affects_type_identity and handler
2237 fields, adjust comments.
2238 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2239 * config/arm/arm.c (arm_attribute_table): Likewise.
2240 * config/avr/avr.c (avr_attribute_table): Likewise.
2241 * config/s390/s390.c (s390_attribute_table): Likewise.
2242 * config/sh/sh.c (sh_attribute_table): Likewise.
2243 * config/i386/i386.c (ix86_handle_cconv_attribute,
2244 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2245 (ix86_attribute_table): Swap affects_type_identity and handler
2246 fields, adjust comments.
2247 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2248 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2249 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2250 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2251 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2252 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2253 * config/rx/rx.c (rx_attribute_table): Likewise.
2254 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2255 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2256 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2257 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2258 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2259 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2260 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2261 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2262 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2263 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2264 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2266 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2268 (register_scoped_attributes, decl_attributes): Formatting fixes.
2270 PR tree-optimization/83269
2271 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2272 subtraction in arg0's type if type is signed and arg0 is unsigned.
2276 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2277 convert? on @0 instead of convert. Check type of @1, not @0.
2278 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2279 convert? on @0 instead of convert. Check type of @1, not @0.
2280 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2281 only optimize if either both @1 and @2 types are narrower
2282 precision, or both are wider or equal precision, and in the former
2283 case only if both have undefined overflow.
2285 2017-12-15 Richard Biener <rguenther@suse.de>
2288 * internal-fn.def (IFN_NOP): Add.
2289 * internal-fn.c (expand_NOP): Do nothing.
2290 * lto-streamer-in.c (input_function): Instead of removing
2291 sanitizer calls replace them with IFN_NOP calls.
2293 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * dse.c (store_info, read_info_type): Replace begin and end with
2299 (print_range): New function.
2300 (set_all_positions_unneeded, any_positions_needed_p)
2301 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2303 (record_store): Likewise. Optimize the case in which all positions
2305 (get_stored_val): Replace read_begin and read_end with read_offset
2307 (replace_read): Update call accordingly.
2309 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2311 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2312 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2313 (loop_cand::loop_cand): Initialize above members.
2314 (loop_cand::supported_operations): Delete.
2315 (loop_cand::can_interchange_p): Inline above function.
2316 (loop_cand::classify_simple_reduction): Record number of constant
2317 initialized simple reductions.
2318 (should_interchange_loops): New parameters. Check stmt cost of loops
2320 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2321 Update call to should_interchange_loops.
2322 (should_interchange_loop_nest): Update call to
2323 should_interchange_loops.
2325 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2328 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2329 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2330 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2332 2017-12-15 Julia Koval <julia.koval@intel.com>
2334 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2335 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2336 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2337 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2338 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2340 2017-12-15 Julia Koval <julia.koval@intel.com>
2342 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2343 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2344 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2345 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2346 _mm_aesenc_epi128): New intrinsics.
2348 2017-12-15 Julia Koval <julia.koval@intel.com>
2350 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2351 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2352 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2353 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2354 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2356 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2358 * gimple-ssa-strength-reduction.c (analyze_increments):
2359 Distinguish replacement costs for constant strides from those for
2362 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2364 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2366 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2368 * doc/invoke.texi: Document -Wcast-function-type.
2369 * recog.h (stored_funcptr): Change signature.
2370 * tree-dump.c (dump_node): Avoid warning.
2371 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2373 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2376 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2379 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2381 PR tree-optimization/83312
2382 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2383 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2384 "val" param, and to cope with arbitrary basic blocks.
2385 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2386 handle NULL_TREE for "val", dropping "bb" param.
2387 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2388 drop "bb" param. Handle NULL_TREE for "val".
2389 (find_case_label_for_value): Make "switch_stmt" param const.
2390 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2392 (vrp_prop::check_all_array_refs): Reimplement as...
2393 (check_array_bounds_dom_walker::before_dom_children): ...this new
2394 vfunc. Replace linear search through BB block list, excluding
2395 those with non-executable in-edges via dominator walk.
2397 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2399 * config/arm/arm.opt (mverbose-cost-dump): New option.
2400 * config/arm/arm.c (arm_rtx_costs): Use it.
2402 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2405 * reload1.c (emit_input_reload_insns): Skip debug markers.
2407 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2409 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2413 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2414 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2415 * config/sh/sh.c: Skip debug insns besides notes.
2416 * config/sh/sh.md: Likewise.
2417 * config/sh/sh_treg_combine.cc: Likewise.
2418 * config/sh/sync.md: Likewise.
2420 2017-12-14 Tom de Vries <tom@codesourcery.com>
2422 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2423 Add item for weak_undefined.
2425 2017-12-14 Richard Biener <rguenther@suse.de>
2427 PR tree-optimization/67842
2428 * sese.h (bb_in_region): Remove #if 0'ed code.
2430 2017-12-14 Richard Biener <rguenther@suse.de>
2432 PR tree-optimization/83326
2433 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2434 may_be_zero parameter and handle it by not marking the first
2435 peeled copy as not exiting the loop.
2436 (try_peel_loop): Likewise.
2437 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2438 to handle the case of constant or zero iterations and perform
2439 loop header copying on-the-fly.
2441 2017-12-14 Richard Biener <rguenther@suse.de>
2443 PR tree-optimization/83418
2444 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2445 Instead of asserting we don't get unfolded comparisons deal with
2448 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2451 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2452 basic blocks. Assert debug bind insns don't appear outside of bbs,
2453 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2456 PR tree-optimization/83198
2457 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2458 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2459 value if arg is a REAL_CST with incompatible type.
2461 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2462 Bin Cheng <bin.cheng@arm.com>
2465 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2467 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2470 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2472 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2473 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2475 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2476 Alan Hayward <alan.hayward@arm.com>
2477 David Sherwood <david.sherwood@arm.com>
2479 * poly-int.h: New file.
2480 * poly-int-types.h: Likewise.
2481 * coretypes.h: Include them.
2482 (POLY_INT_CONVERSION): Define.
2483 * target.def (estimated_poly_value): New hook.
2484 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2485 * doc/tm.texi: Regenerate.
2486 * doc/poly-int.texi: New file.
2487 * doc/gccint.texi: Include it.
2488 * doc/rtl.texi: Describe restrictions on subreg modes.
2489 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2490 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2491 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2492 * targhooks.h (default_estimated_poly_value): Declare.
2493 * targhooks.c (default_estimated_poly_value): New function.
2494 * target.h (estimated_poly_value): Likewise.
2495 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2496 (wi::unary_traits): Delete.
2497 (wi::binary_traits::signed_shift_result_type): Define for
2498 offset_int << HOST_WIDE_INT, etc.
2499 (generic_wide_int::operator <<=): Define for all types and use
2500 wi::lshift instead of <<.
2501 (wi::hwi_with_prec): Add a default constructor.
2502 (wi::ints_for): New class.
2503 (operator <<): Define for all wide-int types.
2504 (operator /): New function.
2505 (operator %): Likewise.
2506 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2507 (ASSERT_MAYBE_NE_AT): New macros.
2509 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2510 Dominik Vogt <vogt@linux.vnet.ibm.com>
2513 * emit-rtl.c (init_emit): Remove ??? comment.
2514 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2515 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2516 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2517 alignment of 3 virtual registers to BITS_PER_WORD.
2519 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2521 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2523 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2524 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2526 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2529 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2531 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2533 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2534 with the exception of debug begin stmt markers appear before
2538 * final.c (rest_of_handle_final): Call variable_tracking_main only
2539 if !flag_var_tracking.
2541 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2542 Jakub Jelinek <jakub@redhat.com>
2546 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2547 labels debug stmts that can only appear after labels.
2549 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2551 PR rtl-optimization/82398
2552 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2553 EXPR_USEFULNESS in priority comparison.
2555 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2557 PR rtl-optimization/83393
2558 * combine.c (move_deaths): If reg_stat points to a too new insn in
2559 last_death, do not use it: find the proper insn instead.
2561 2017-12-12 Jeff Law <law@redhat.com>
2563 PR tree-optimization/83298
2564 PR tree-optimization/83362
2565 PR tree-optimization/83383
2566 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2567 push_value_range a public interface. Add new argument to
2568 record_ranges_from_stmt.
2569 * gimple-ssa-evrp-analyze.c
2570 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2571 Update comments. Handle recording temporary equivalences.
2572 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2573 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2574 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2575 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2576 gimple-ssa-evrp-analyze.h.
2577 (record_temporary_equivalences_from_phis): Add new argument. When
2578 the PHI arg is an SSA_NAME, set the result's range to the range
2580 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2581 from statements too.
2582 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2583 Pass it down to children as needed.
2584 (thread_outgoing_edges): Likewise.
2585 (thread_across_edge): Likewise. Push/pop range state as needed.
2586 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2588 2017-12-12 Julia Koval <julia.koval@intel.com>
2590 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2591 (PTA_CANNONLAKE): Remove PTA_CLWB.
2593 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2596 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2597 (vcondv2div2df): Likewise.
2598 (vconduv2dfv2di): Likewise.
2600 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2602 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2603 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2604 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2605 ATTR_MATHFN_FPROUNDING_ERRNO.
2607 2017-12-12 Richard Biener <rguenther@suse.de>
2609 PR tree-optimization/83385
2610 * graphite-scop-detection.c (get_order, order): Remove.
2611 (bb_to_rpo): New global.
2613 (build_scops): Sort pbbs in RPO order.
2615 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2617 * combine.c (simplify_set): Do not transform subregs to zero_extends
2618 if the destination mode is a vector mode.
2620 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2622 PR tree-optimization/83359
2623 * tree-cfg.h (fold_loop_internal_call): Declare.
2624 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2625 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2626 (find_loop_dist_alias): New function.
2627 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2628 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2631 PR tree-optimization/80631
2632 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2633 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2634 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2635 hardcoding zero as the value if COND_EXPR is never true. For
2636 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2637 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2638 hardcoding MAX_EXPR as the reduction operation.
2639 (is_nonwrapping_integer_induction): Allow negative step.
2640 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2641 vect_create_epilog_for_reduction, if no value is suitable, don't
2642 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2644 2017-12-12 Richard Biener <rguenther@suse.de>
2646 PR tree-optimization/81889
2647 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2648 range info from the non-wrapping IV instead of just the range
2651 2017-12-12 Julia Koval <julia.koval@intel.com>
2653 * config.gcc: Add vaesintrin.h.
2654 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2656 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2657 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2659 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2660 * config/i386/immintrin.h: Include vaesintrin.h.
2661 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2662 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2663 _mm_aesdec_epi128): New intrinsics.
2665 2017-12-12 Julia Koval <julia.koval@intel.com>
2667 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2668 OPTION_MASK_ISA_VAES_UNSET): New.
2669 (ix86_handle_option): Handle -mvaes.
2670 * config/i386/cpuid.h: Define bit_VAES.
2671 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2672 * config/i386/i386-c.c (__VAES__): New.
2673 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2674 (ix86_valid_target_attribute_inner_p): Ditto.
2675 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2676 * config/i386/i386.opt: Add -mvaes.
2677 * doc/invoke.texi: Ditto.
2679 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2681 * debug.h (gcc_debug_hooks): Add inline_entry.
2682 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2683 * debug.c (do_nothing_debug_hooks): Likewise.
2684 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2685 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2686 (dwarf2_lineno_debug_hooks): Likewise.
2688 * common.opt (gstatement-frontiers): New, setting
2689 debug_nonbind_markers_p.
2690 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2691 * toplev.c (process_options): Autodetect value for debug statement
2693 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2694 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2696 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2697 markers. Integrate source bind into debug stmt expand loop.
2698 (pass_expand::execute): Check debug marker limit. Avoid deep
2699 TER and expand debug locations for debug bind insns only.
2700 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2702 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2703 * final.c (reemit_insn_block_notes): Take current block from
2704 nonbind markers. Declare note where it's first set.
2705 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2706 begin stmt markers if enabled.
2707 (notice_source_line): Handle nonbind markers. Fail if their
2708 location is unknown or that of builtins.
2709 (rest_of_handle_final): Convert begin stmt markers to notes if
2710 var-tracking didn't run.
2711 (rest_of_clean_state): Skip begin stmt markers.
2712 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2714 * function.c (allocate_struct_function): Set begin_stmt_markers.
2715 * function.h (struct function): Add debug_marker_count counter
2716 and debug_nonbind_markers flag.
2717 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2718 * gimple-low.c (lower_function_body): Adjust
2719 debug_nonbind_markers.
2720 (lower_stmt): Drop or skip gimple debug stmts.
2721 (lower_try_catch): Skip debug stmts.
2722 * gimple.c (gimple_build_debug_begin_stmt): New.
2723 (gimple_copy): Increment debug_marker_count if copying one.
2724 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2725 * gimplify.c (rexpr_location): New.
2726 (rexpr_has_location): New.
2727 (warn_switch_unreachable_r): Handle gimple debug stmts.
2728 (shortcut_cond_r): Call expr_location.
2730 (find_goto_label): New.
2731 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2733 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2735 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2736 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2737 (LANG_HOOKS_INITIALIZER): ... this.
2738 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2739 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2741 (update_ebb_live_info): Skip debug insn markers.
2742 * lra.c (debug_insn_static_data): Rename to...
2743 (debug_bind_static_data): ... this.
2744 (debug_marker_static_data): New.
2745 (lra_set_insn_recog_data): Select one of the above depending
2747 (lra_update_isn_regno_info): Don't assume debug insns have
2749 (push_insns): Skip debug insns.
2750 * lto-streamer-in.c (input_function): Drop debug stmts
2751 depending on active options. Adjust debug_nonbind_markers.
2752 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2753 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2754 begin stmt marker notes.
2755 (print_insn): Likewise.
2756 * recog.c (extract_insn): Recognize rtl for debug markers.
2757 * rtl.def (DEBUG_MARKER): New.
2758 * tree-inline.c: Include params.h.
2759 (remap_gimple_stmt): Handle nonbind markers.
2760 (maybe_move_debug_stmts_to_successors): Likewise.
2761 (copy_debug_stmt): Likewise.
2762 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2763 markers regardless of no side effects.
2764 (tsi_link_before): Don't update container's side effects when adding
2765 a begin stmt marker.
2766 (tsi_link_after): Likewise.
2767 (expr_first): Skip begin stmt markers.
2768 (expr_last): Likewise.
2769 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2770 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2771 Disregard nonbind markers.
2772 * tree.c (make_node_stat): Don't set side effects for begin stmt
2774 (build1_stat): Likewise.
2775 * tree.def (DEBUG_BEGIN_STMT): New.
2776 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2777 * var-tracking.c (delete_debug_insns): Renamed to...
2778 (delete_vta_debug_insns): ... this.
2779 (reemit_marker_as_note): New.
2780 (vt_initialize): Reemit markers.
2781 (delete_vta_debug_insns): Likewise.
2782 (vt_debug_insns_local): Reemit or delete markers.
2783 (variable_tracking_main_1): Likewise.
2784 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2785 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2786 (gimple_debug_nonbind_marker_p): New.
2787 (gimple_build_debug_bind): Adjust.
2788 (gimple_build_debug_begin_stmt): New.
2789 * doc/invoke.texi (max-debug-marker-count): New param.
2790 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2791 (debug_parameter_ref, debug_marker): New.
2792 (NOTE_INSN_BEGIN_STMT): New.
2793 (DEBUG_INSN): Describe begin stmt markers.
2795 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2796 without debug insns, we wouldn't, but clean up debug insns
2797 after a control flow insn nevertheless.
2799 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2800 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2801 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2802 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2803 (rtl_tidy_fallthru_edge): Likewise.
2804 (rtl_verify_fallthru): Likewise.
2805 (rtl_verify_bb_layout): Likewise.
2806 (skip_insns_after_block): Likewise.
2807 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2808 * dwarf2out.c: Include print-rtl.h.
2809 (dwarf2out_next_real_insn): New.
2810 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2811 Dump debug binds in asm comments.
2812 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2813 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2814 callers to use gsi_start_nondebug_bb instead.
2815 (gsi_after_labels): Skip gimple debug stmts.
2816 (gsi_start_nondebug): New.
2817 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2818 (proper_loop_form_for_interchange): Adjust.
2819 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2820 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2821 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2822 (collect_fallthrough_labels): Likewise.
2823 (should_warn_for_implicit_fallthrough): Likewise.
2824 (warn_implicit_fallthrough_r): Likewise.
2825 (expand_FALLTHROUGH_r): Likewise.
2826 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2827 (graphite_copy_stmts_from_block): Skip nonbind markers.
2828 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2829 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2830 * jump.c (clean_barriers): Skip debug insns.
2831 * omp-expand.c (expand_parallel_call): Skip debug insns.
2832 (expand_task_call): Likewise.
2833 (remove_exit_barrier): Likewise.
2834 (expand_omp_taskreg): Likewise.
2835 (expand_omp_for_init_counts): Likewise.
2836 (expand_omp_for_generic): Likewise.
2837 (expand_omp_for_static_nochunk): Likewise.
2838 (expand_omp_for_static_chunk): Likewise.
2839 (expand_omp_simd): Likewise.
2840 (expand_omp_taskloop_for_outer): Likewise.
2841 (expand_omp_taskloop_for_inner): Likewise.
2842 (expand_oacc_for): Likewise.
2843 (expand_omp_sections): Likewise.
2844 (expand_omp_single): Likewise.
2845 (expand_omp_synch): Likewise.
2846 (expand_omp_atomic_load): Likewise.
2847 (expand_omp_atomic_store): Likewise.
2848 (expand_omp_atomic_fetch_op): Likewise.
2849 (expand_omp_atomic_pipeline): Likewise.
2850 (expand_omp_atomic_mutex): Likewise.
2851 (expand_omp_target): Likewise.
2852 (grid_expand_omp_for_loop): Likewise.
2853 (grid_expand_target_grid_body): Likewise.
2854 (build_omp_regions_1): Likewise.
2855 * omp-low.c (check_combined_parallel): Skip debug stmts.
2856 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2857 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2858 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2860 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2861 (make_edges): Likewise.
2862 (cleanup_dead_labels): Likewise.
2863 (gimple_can_merge_blocks_p): Likewise.
2864 (stmt_starts_bb_p): Likewise.
2865 (gimple_block_label): Likewise.
2866 (gimple_redirect_edge_and_branch): Likewise.
2867 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2869 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2871 * tree-pretty-print (print_declaration): Omit initializer in slim
2873 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2875 (eliminate_unnecessary_stmts): Stabilize block removal order.
2876 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2877 * var-tracking.c (get_first_insn): New.
2878 (vt_emit_notes): Call it.
2879 (vt_initialize): Walk any insns before the first BB.
2880 (delete_debug_insns): Likewise.
2882 * gimple.h (enum gimple_debug_subcode): Add
2883 GIMPLE_DEBUG_BEGIN_STMT.
2884 (gimple_debug_begin_stmt_p): New.
2885 (gimple_debug_nonbind_marker_p): New.
2886 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2887 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2888 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2889 * insn-notes.def (BEGIN_STMT): New.
2890 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2891 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2892 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2893 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2894 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2895 (INSN_DEBUG_MARKER_KIND): New.
2896 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2897 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2898 (INSN_VAR_LOCATION_PTR): New.
2899 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2901 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2902 deps only if debug bind insns are enabled.
2903 (pass_expand::execute): Avoid deep TER and expand
2904 debug locations for debug bind insns only.
2905 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2906 debug stmts special handling down to debug bind stmts.
2907 * combine.c (try_combine): Narrow debug insns special handling
2908 down to debug bind insns.
2909 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2910 Narrow debug insns preexisting special handling down to debug
2912 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2913 handling down to debug bind insns.
2914 * function.c (instantiate_virtual_regs): Skip debug markers,
2915 adjust handling of debug binds.
2916 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2917 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2918 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2919 insns to debug bind insns.
2920 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2921 special handling of debug stmts to debug bind stmts.
2922 * ipa-split.c (split_function): Likewise.
2923 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2924 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2926 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2927 * regrename.c (build_def_use): Likewise.
2928 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2929 (pass_cprop_hardreg): Narrow special casing of debug insns to
2931 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2932 * reload1.c (reload): Likewise.
2933 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2934 casing of debug stmts to debug bind stmts.
2935 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2936 * ssa-iterators.h (num_imm_uses): Likewise.
2937 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2938 debug stmts to debug bind stmts.
2939 * tree-inline.c (tree_function_versioning): Narrow special casing
2940 of debug stmts to debug bind stmts.
2941 * tree-loop-distribution.c (generate_loops_for_partition):
2942 Narrow special casing of debug stmts to debug bind stmts.
2943 * tree-sra.c (analyze_access_subtree): Narrow special casing
2944 of debug stmts to debug bind stmts.
2945 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2946 stmts to debug bind stmts.
2947 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2948 casing of debug stmts to debug bind stmts.
2949 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2950 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2951 casing of debug stmts to debug bind stmts.
2952 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2954 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2955 debug stmts to debug bind stmts.
2956 (gimple_replace_ssa_lhs): Likewise.
2957 (insert_debug_temp_for_var_def): Likewise.
2958 (insert_debug_temps_for_defs): Likewise.
2959 (reset_debug_uses): Likewise.
2960 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2961 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2962 (adjust_debug_stmts): Likewise.
2963 (adjust_phi_and_debug_stmts): Likewise.
2964 (vect_do_peeling): Likewise.
2965 * tree-vect-loop.c (vect_transform_loop): Likewise.
2966 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2967 * var-tracking.c (adjust_mems): Narrow special casing of debug
2968 insns to debug bind insns.
2969 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2970 (compute_bb_dataflow, vt_find_locations): Likewise.
2971 (vt_expand_loc, emit_notes_for_changes): Likewise.
2972 (vt_init_cfa_base): Likewise.
2973 (vt_emit_notes): Likewise.
2974 (vt_initialize): Likewise.
2975 (vt_finalize): Likewise.
2977 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2978 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2979 (next_nonnote_nondebug_insn_bb): New.
2980 (prev_nonnote_nondebug_insn_bb): New.
2981 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2982 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2983 (prev_nonnote_nondebug_insn_bb): Declare.
2984 (next_nonnote_nondebug_insn_bb): Declare.
2985 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2986 * cfgrtl.c (get_last_bb_insn): Likewise.
2987 * lra.c (push_insns): Likewise.
2989 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2992 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2994 * selftest.c (selftest::test_runner::test_runner): New ctor.
2995 (selftest::test_runner::~test_runner): New dtor.
2996 * selftest.h (class selftest::test_runner): New class.
2998 2017-12-11 Carl Love <cel@us.ibm.com>
3000 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3001 vec_extract_fp32_from_shortl]): Add #defines.
3002 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3003 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3004 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3005 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3006 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3007 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3008 * doc/extend.texi: Add documentation for the added builtins.
3010 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3012 PR rtl-optimization/80693
3013 PR rtl-optimization/81019
3014 PR rtl-optimization/81020
3015 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3016 are not mentioned in i3. Place the REG_UNUSED note on i2,
3017 possibly modified to REG_DEAD, if it did not originate in i3.
3019 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3021 * recog.c (store_data_bypass_p_1): New function.
3022 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3023 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3025 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3027 PR rtl-optimization/83361
3028 * ifcvt.c (if_convert): Call fixup_partitions.
3030 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3033 early folding of splat_u{8,16,32}.
3035 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3037 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3038 output_operand_lossage first argument with capital letter.
3039 (aarch64_override_options): Don't start error and sorry first argument
3040 with capital letter.
3042 2017-12-11 Andi Kleen <ak@linux.intel.com>
3044 PR gcov-profile/83355
3045 * auto-profile.c (string_table::get_index_by_decl): Don't
3046 recurse when abstract origin points to itself.
3048 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3050 PR tree-optimization/83320
3051 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3052 (prune_datarefs_not_in_loop): Ditto.
3054 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3056 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3058 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3060 PR tree-optimization/83337
3061 * gimple-loop-interchange.cc (compute_access_stride): Handle
3062 bitfield DRs properly.
3064 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3066 PR tree-optimization/83338
3067 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3068 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3069 vector of unsigned integers to vector of signed integers.
3071 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3073 PR rtl-optimization/83317
3074 * lra-constraints.c (process_address_1): Add insn code check.
3076 2017-12-08 Michael Matz <matz@suse.de>
3078 Fix PR tree-optimization/83323
3079 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3080 head-controlled loops and loop BBs.
3081 * common.opt (funroll-and-jam): Remove, instead ...
3082 (floop-unroll-and-jam): ... reuse this option.
3083 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3084 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3085 (-floop-unroll-and-jam): ... this option.
3087 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3089 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3090 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3091 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3093 2017-12-08 Richard Biener <rguenther@suse.de>
3096 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3097 handle accesses outside of zero-sized vars.
3099 2017-12-08 Martin Jambor <mjambor@suse.cz>
3101 PR tree-optimization/83141
3102 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3103 test for MEM_REFs implicitely changing types with padding. Remove
3105 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3107 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3109 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3111 * config/avr/avr.c (avr_attribute_table): Likewise.
3112 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3113 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3114 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3115 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3116 of struct attribute_spec.
3117 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3119 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3122 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3124 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3127 * profile-count.c (profile_count::from_gcov_type): Move from
3128 profile-count.h; handle overflow.
3129 * profile-count.h (profile_count::from_gcov_type): Move offline.
3131 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3133 PR rtl-optimization/83304
3134 * combine.c (move_deaths): If we do not know where a register died,
3137 2017-12-08 Richard Biener <rguenther@suse.de>
3139 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3140 Provide -fopt-info-loop feedback when we interchange in a nest.
3142 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3144 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3145 for armv6 ARM CPU IDs.
3147 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3149 * common/config/arm/arm-common.c: Include <algorithm>.
3150 (INCLUDE_VECTOR): Define.
3151 (compare_opt_names): New function.
3152 (arm_rewrite_selected_arch): Only strip out extensions that can be
3153 expressed through -mfpu. Sort the remaining extensions
3156 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3158 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3159 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3160 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3161 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3163 2017-06-08 Tristan Gingold <gindold@adacore.com>
3166 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3167 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3169 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3171 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3172 of struct attribute_spec.
3174 2017-12-08 Julia Koval <julia.koval@intel.com>
3176 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3177 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3178 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3179 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3180 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3181 _mm_maskz_dpwssds_epi32): Ditto.
3183 2017-12-08 Richard Biener <rguenther@suse.de>
3185 PR tree-optimization/81303
3186 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3187 conditions try to create a comparison vector type matching
3188 the data vector type.
3189 (vectorizable_condition): Adjust.
3190 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3191 Leave invariant conditions alone in case we can vectorize those.
3193 2017-12-08 Julia Koval <julia.koval@intel.com>
3195 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3196 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3197 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3198 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3199 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3201 2017-12-08 Julia Koval <julia.koval@intel.com>
3203 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3204 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3205 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3206 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3207 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3208 _mm_maskz_dpbusds_epi32): New intrinsics.
3210 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3212 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3214 (ld<bh>io_signed): Likewise.
3215 (st<bhw>io): Likewise.
3216 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3217 SMALL_INT12 constant integer operand.
3219 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3221 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3223 (nios2_large_constant_allowed): ...this. Adjust uses.
3224 (nios2_plus_symbolic_constant_p): Rename to...
3225 (nios2_plus_large_constant_p): ...this. Adjust uses.
3226 (nios2_legitimate_address_p): Correct CONST_INT handling.
3227 (nios2_symbolic_memory_operand_p): Rename to...
3228 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3229 (nios2_large_constant_p): Check for large constant integers too.
3230 (nios2_split_large_constant): Handle constant integers.
3231 (nios2_split_symbolic_memory_operand): Rename to...
3232 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3233 (nios2_legitimize_constant_address): Handle constant integers.
3234 (r0rel_constant_p): Handle small constant integers.
3235 (nios2_print_operand_address): Handle r0-relative integer addresses.
3236 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3237 * config/nios2/nios2.md: Adjust for renamed functions.
3239 2017-12-07 Andrew Waterman <andrew@sifive.com>
3241 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3242 (TARGET_HAVE_SRODATA_SECTION): New define.
3243 (riscv_select_section): New function.
3245 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3246 Alexander Monakov <amonakov@ispras.ru>
3247 Jakub Jelinek <jakub@redhat.com>
3250 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3252 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3254 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3255 a bool success value. Don't call output_operand_lossage here.
3256 (aarch64_print_ldpstp_address): Return a bool success value.
3257 (aarch64_print_operand_address): Call output_addr_const if
3258 aarch64_print_address_internal fails.
3259 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3260 'y'; call output_operand_lossage instead. Call output_operand_lossage
3261 if aarch64_print_ldpstp_address fails.
3263 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3265 * tree-vector-builder.h
3266 (tree_vector_builder::binary_encoded_nelts): Declare.
3267 * tree-vector-builder.c
3268 (tree_vector_builder::binary_encoded_nelts): New function.
3269 * fold-const.c (negate_expr_p): Likewise.
3270 (operand_equal_p, fold_checksum_tree): Likewise.
3271 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3272 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3273 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3274 (uniform_vector_p): Likewise.
3275 * varasm.c (const_hash_1, compare_constant): Likewise.
3276 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3277 (valid_lattice_transition): Operate directly on the VECTOR_CST
3279 * ipa-icf.c: Include tree-vector-builder.h.
3280 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3281 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3283 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3285 * tree.c (build_vector): Delete.
3286 * tree.h (build_vector): Make static and move into the self-testing
3289 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3291 * vector-builder.h (vector_builder::derived): New const overload.
3292 (vector_builder::elt): New function.
3293 * tree-vector-builder.h (tree_vector_builder::type): New function.
3294 (tree_vector_builder::apply_step): Declare.
3295 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3297 * gimple-fold.h (tree_vector_builder): Declare.
3298 (gimple_build_vector): Take a tree_vector_builder instead of a
3299 type and vector of elements.
3300 * gimple-fold.c (gimple_build_vector): Likewise.
3301 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3303 (get_initial_defs_for_reduction): Likewise.
3304 (vectorizable_induction): Likewise.
3306 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3308 * tree-vector-builder.h
3309 (tree_vector_builder::new_binary_operation): Declare.
3310 * tree-vector-builder.c
3311 (tree_vector_builder::new_binary_operation): New function.
3312 * fold-const.c (fold_relational_const): Use it.
3313 (const_binop): Likewise. Check that both input vectors have
3314 the same number of elements, thus excluding things like WIDEN_SUM.
3315 Check whether it is possible to operate directly on the encodings
3318 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3320 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3321 new_unary_operation, operating only on the encoded elements.
3322 (const_unop): Likewise.
3323 (exact_inverse): Likewise.
3324 (distributes_over_addition_p): New function.
3325 (const_binop): Use tree_vector_builder and new_unary_operation
3326 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3327 on the encoded elements unless the encoding is strided and the
3328 operation does not distribute over addition.
3329 (fold_convert_const): Use tree_vector_builder and
3330 new_unary_operation. Operate only on the encoded elements
3331 for truncating integer conversions, or for non-stepped encodings.
3333 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3335 * config/sparc/sparc.c: Include tree-vector-builder.h.
3336 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3337 * expmed.c: Include tree-vector-builder.h.
3338 (make_tree): Use tree_vector_builder instead of build_vector.
3339 * fold-const.c: Include tree-vector-builder.h.
3340 (const_binop): Use tree_vector_builder instead of build_vector.
3341 (const_unop): Likewise.
3342 (native_interpret_vector): Likewise.
3343 (fold_vec_perm): Likewise.
3344 (fold_ternary_loc): Likewise.
3345 * gimple-fold.c: Include tree-vector-builder.h.
3346 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3348 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3349 (simplify_vector_constructor): Use tree_vector_builder instead
3351 * tree-vect-generic.c: Include tree-vector-builder.h.
3352 (add_rshift): Use tree_vector_builder instead of build_vector.
3353 (expand_vector_divmod): Likewise.
3354 (optimize_vector_constructor): Likewise.
3355 * tree-vect-loop.c: Include tree-vector-builder.h.
3356 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3357 of build_vector. Explicitly use a stepped encoding for
3359 * tree-vect-slp.c: Include tree-vector-builder.h.
3360 (vect_get_constant_vectors): Use tree_vector_builder instead
3362 (vect_transform_slp_perm_load): Likewise.
3363 (vect_schedule_slp_instance): Likewise.
3364 * tree-vect-stmts.c: Include tree-vector-builder.h.
3365 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3366 (vect_gen_perm_mask_any): Likewise.
3367 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3368 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3370 (build_vector_from_val): Likewise. Explicitly use a duplicate
3373 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3375 * doc/generic.texi (VECTOR_CST): Describe new representation of
3377 * vector-builder.h: New file.
3378 * tree-vector-builder.h: Likewise.
3379 * tree-vector-builder.c: Likewise.
3380 * Makefile.in (OBJS): Add tree-vector-builder.o.
3381 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3382 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3383 (tree_vector): Change the number of elements to
3384 vector_cst_encoded_nelts.
3385 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3386 (VECTOR_CST_ELTS): Delete.
3387 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3388 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3389 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3390 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3391 (VECTOR_CST_ENCODED_ELT): Likewise.
3392 (vector_cst_encoded_nelts): New function.
3393 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3394 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3395 (vector_cst_int_elt, vector_cst_elt): Declare.
3396 * tree.c: Include tree-vector-builder.h.
3397 (tree_code_size): Abort if passed VECTOR_CST.
3398 (tree_size): Update for new VECTOR_CST layout.
3399 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3400 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3401 (build_vector): Use tree_vector_builder.
3402 (vector_cst_int_elt, vector_cst_elt): New functions.
3403 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3404 encoded elements and then create the vector in the canonical form.
3405 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3406 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3407 (tree_c_tests): Call test_vector_cst_patterns.
3408 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3410 (hash_tree): Likewise.
3411 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3412 (streamer_write_tree_header): Likewise.
3413 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3414 (streamer_alloc_tree): Likewise. Update call to make_vector.
3415 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3417 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3419 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3420 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3421 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3423 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3424 Richard Biener <rguenther@suse.de>
3426 PR tree-optimization/81303
3427 * Makefile.in (gimple-loop-interchange.o): New object file.
3428 * common.opt (floop-interchange): Reuse the option from graphite.
3429 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3430 -floop-interchange and mention it for -O3.
3431 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3432 * gimple-loop-interchange.cc: New file.
3433 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3434 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3435 * passes.def (pass_linterchange): New pass.
3436 * timevar.def (TV_LINTERCHANGE): New time var.
3437 * tree-pass.h (make_pass_linterchange): New declaration.
3438 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3439 interchange. Record IV before/after increment in new parameters.
3440 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3441 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3443 (check_reduction_path): ...New function here.
3444 * tree-vectorizer.h (check_reduction_path): New declaration.
3446 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3449 PR rtl-optimization/80818
3450 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3451 always early clobbered.
3452 * lra-lives.c (process_bb_lives): Check input hard regs for early
3453 clobbered non-operand hard reg.
3455 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3458 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3459 types_compatible_p, just that TYPE_MODE is the same.
3461 2017-12-07 Martin Sebor <msebor@redhat.com>
3464 * attribs.c (empty_attribute_table): Initialize new member of
3465 struct attribute_spec.
3466 (decl_attributes): Add argument. Handle mutually exclusive
3467 combinations of attributes.
3468 (selftests::test_attribute_exclusions): New function.
3469 (selftests::attribute_c_tests): Ditto.
3470 * attribs.h (decl_attributes): Add default argument.
3471 * selftest.h (attribute_c_tests): Declare.
3472 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3473 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3475 * doc/extend.texi (Common Function Attributes): Update const and pure.
3476 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3477 of struct attribute_spec.
3478 * config/arc/arc.c (arc_attribute_table): Same.
3479 * config/arm/arm.c (arm_attribute_table): Same.
3480 * config/avr/avr.c ( avr_attribute_table): Same.
3481 * config/bfin/bfin.c (bfin_attribute_table): Same.
3482 * config/cr16/cr16.c (cr16_attribute_table): Same.
3483 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3484 * config/h8300/h8300.c (h8300_attribute_table): Same.
3485 * config/i386/i386.c (ix86_attribute_table): Same.
3486 * config/ia64/ia64.c (ia64_attribute_table): Same.
3487 * config/m32c/m32c.c (m32c_attribute_table): Same.
3488 * config/m32r/m32r.c (m32r_attribute_table): Same.
3489 * config/m68k/m68k.c (m68k_attribute_table): Same.
3490 * config/mcore/mcore.c (mcore_attribute_table): Same.
3491 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3492 * config/mips/mips.c (mips_attribute_table): Same.
3493 * config/msp430/msp430.c (msp430_attribute_table): Same.
3494 * config/nds32/nds32.c (nds32_attribute_table): Same.
3495 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3496 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3497 * config/rl78/rl78.c (rl78__attribute_table): Same.
3498 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3499 * onfig/rx/rx.c (rx_attribute_table): Same.
3500 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3501 * config/sh/sh.c (sh_attribute_table): Same.
3502 * config/sparc/sparc.c (sparc_attribute_table): Same.
3503 * config/spu/spu.c (spu_attribute_table): Same.
3504 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3505 * config/v850/v850.c (v850_attribute_table): Same.
3506 * config/visium/visium.c (visium_attribute_table): Same.
3508 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3511 * config/arm/arm.c (INCLUDE_STRING): Define.
3512 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3513 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3516 2017-12-07 Michael Matz <matz@suse.de>
3518 Add unroll and jam pass
3520 * gimple-loop-jam.c: New file.
3521 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3522 * common.opt (funroll-and-jam): New option.
3523 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3524 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3525 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3526 * passes.def: Add pass_loop_jam.
3527 * timevar.def (TV_LOOP_JAM): Add.
3528 * tree-pass.h (make_pass_loop_jam): Declare.
3529 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3530 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3531 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3532 to flow_loop_tree_node_add.
3533 (duplicate_subloops, copy_loops_to): Append to sibling list.
3534 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3535 * doc/invoke.texi (-funroll-and-jam): Document new option.
3536 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3538 2017-12-07 Richard Biener <rguenther@suse.de>
3540 PR tree-optimization/83296
3541 PR tree-optimization/67769
3542 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3543 flow sensitive info in an unrelated BB.
3544 (value_replacement): Use reset_flow_sensitive_info.
3545 (minmax_replacement): Reset flow sensitive info on the def
3546 we move. Do not reset flow sensitive info in the whole BB
3547 we move the stmt to.
3548 (abs_replacement): Likewise.
3550 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3553 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3554 rs6000_cpu to the given -mcpu=, or to the default processor.
3556 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3558 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3559 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3560 instead of rs6000_cpu_attr.
3561 (rs6000_adjust_cost): Ditto.
3562 (is_microcoded_insn): Ditto.
3563 (rs6000_adjust_priority): Ditto.
3564 (rs6000_issue_rate): Ditto.
3565 (rs6000_use_sched_lookahead): Ditto.
3566 (rs6000_use_sched_lookahead_guard): Ditto.
3567 (rs6000_sched_reorder): Ditto.
3568 (force_new_group): Ditto.
3569 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3570 (group_ending_nop): Ditto.
3572 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3574 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3575 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3576 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3577 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3579 (rs6000_reassociation_width): Ditto.
3580 (rs6000_emit_epilogue): Ditto.
3581 (rs6000_adjust_cost): Ditto.
3582 (is_microcoded_insn): Ditto.
3583 (is_cracked_insn): Ditto.
3584 (rs6000_adjust_priority): Ditto.
3585 (rs6000_sched_reorder): Ditto.
3586 (rs6000_sched_reorder2): Ditto.
3587 (insn_must_be_first_in_group): Ditto.
3588 (insn_must_be_last_in_group): Ditto.
3589 (rs6000_register_move_cost): Ditto.
3590 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3593 2017-12-07 Julia Koval <julia.koval@intel.com>
3595 * config.gcc: Add vaesintrin.h.
3596 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3598 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3599 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3601 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3602 * config/i386/immintrin.h: Include vaesintrin.h.
3603 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3604 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3605 _mm_aesdec_epi128): New intrinsics.
3607 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3609 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3610 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3611 c-family/c-spellcheck.cc.
3612 (best_macro_match::best_macro_match): Likewise.
3614 (struct edit_distance_traits<cpp_hashnode *>): Move to
3615 c-family/c-spellcheck.h.
3616 (class best_macro_match): Likewise.
3618 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3620 PR tree-optimization/83293
3621 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3622 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3623 might insert into empty bb.
3626 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3627 simplify for plus with :c added, and pointer_plus without that.
3628 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3629 with undefined overflow and the conversion is not widening,
3630 perform negation in utype and only convert to type afterwards.
3631 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3632 simplify for plus with :c added, and pointer_plus without that.
3633 If type is integral with undefined overflow and the conversion is
3634 not widening, perform minus in utype and only convert to type
3635 afterwards. Move the last pointer_diff_expr simplify into the
3638 2017-12-06 Martin Sebor <msebor@redhat.com>
3640 PR tree-optimization/82646
3641 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3642 strncpy, not maxlen.
3644 2017-12-06 Martin Sebor <msebor@redhat.com>
3646 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3649 PR tree-optimization/83075
3650 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3651 strncat/strncpy don't change length of source string.
3653 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3656 2017-11-29 Martin Aberg <maberg@gaisler.com>
3658 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3659 to prevent b2bst errata sequence.
3660 (sqrtdf2_fix): Likewise.
3662 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3664 PR tree-optimization/81945
3665 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3666 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3667 to dest_cfun has orig_loop_num set, either remap it to the new
3668 loop number if the loop got moved too, or clear it.
3670 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3672 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3674 (thunderx2t99_nothing): New insn reservation.
3675 (thunderx2t99_mrs): New insn reservation.
3676 (thunderx2t99_multiple): New insn reservation.
3677 (thunderx2t99_alu_basi): Add bfx to reservation.
3678 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3680 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3683 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3685 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3687 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3688 tree-ssa-pre.c::remove_dead_inserted_code.
3689 * tree-ssa-dce.h: New file.
3690 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3691 (remove_dead_inserted_code): Move and rename to function
3692 tree-ssa-dce.c::simple_dce_from_worklist.
3693 (pass_pre::execute): Update use.
3695 2017-12-05 Richard Biener <rguenther@suse.de>
3697 PR tree-optimization/83277
3698 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3699 to code-gen liveout vars.
3701 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3703 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3704 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3705 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3708 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3710 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3712 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3713 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3716 2017-12-05 Richard Biener <rguenther@suse.de>
3718 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3719 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3720 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3721 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3723 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3725 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3727 2017-12-05 Martin Liska <mliska@suse.cz>
3728 Jakub Jelinek <jakub@redhat.com>
3730 * doc/invoke.texi: Document the options.
3731 * flag-types.h (enum sanitize_code): Add
3732 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3733 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3734 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3735 * opts.c: Define new sanitizer options.
3736 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3737 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3739 2017-12-05 Julia Koval <julia.koval@intel.com>
3741 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3742 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3743 (ix86_handle_option): Handle -mavx512vnni.
3744 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3745 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3746 * config/i386/i386-c (__AVX512VNNI__): New.
3747 * config/i386/i386.c (ix86_target_string): Handle new option.
3748 (ix86_valid_target_attribute_inner_p): Handle new option.
3749 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3750 * config/i386/i386.opt (mavx512vnni): New option.
3752 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3755 * athlon.md: Disable for generic.
3756 * haswell.md: Enable for generic.
3757 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3758 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3760 (ix86_adjust_cost): Move generic to haswell path.
3762 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3764 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3765 instead of MEM_P in a couple more places. Fix formatting issues.
3767 2017-12-04 Jim Wilson <jimw@sifive.com>
3769 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3770 instead of GP_REG_LAST-1.
3771 (riscv_adjust_libcall_cfi_prologue): Likewise.
3772 (riscv_adjust_libcall_cri_epilogue): Likewise.
3773 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3776 2017-12-04 Luis Machado <luis.machado@linaro.org>
3778 * ipa-pure-const.c (check_decl): Add missing newline.
3779 (state_from_flags): Likewise.
3781 2017-12-04 Jeff Law <law@redhat.com>
3783 PR tree-optimizatin/78496
3784 * gimple-ssa-evrp-analyze.h
3785 (evrp_range_analyzer::get_vr_values): Simplify.
3786 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3787 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3788 and gimple-ssa-evrp-analyze.h.
3789 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3790 (simplify_stmt_for_jump_threading): Copy a blob of code from
3791 tree-vrp.c to use ranges to simplify statements.
3792 (dom_opt_dom_walker::before_dom_children): Call
3793 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3794 (dom_opt_dom_walker::after_dom_children): Similarly for
3795 evrp_range_analyzer::leave.
3796 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3799 * gimple-ssa-evrp-analyze.c
3800 (evrp_range_analyzer::extract_range_from_stmt): Always use
3801 vr_values::update_value_range so preexisting range info is
3802 medged with new range info, even if the new range is VR_VARYING.
3804 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3806 * combine.c: Adjust comment.
3807 (use_crosses_set_p): Delete.
3808 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3809 (try_combine): Ditto.
3811 2017-12-04 Richard Biener <rguenther@suse.de>
3813 PR tree-optimization/83255
3814 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3815 Re-add zero-iteration check.
3817 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3819 PR rtl-optimization/83245
3820 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3821 hard registers as earlyclobber, also if not in an asm.
3823 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3827 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3831 rs6000_cpu based on cpu_index, not tune_index.
3833 2017-12-04 Richard Biener <rguenther@suse.de>
3835 PR tree-optimization/83238
3836 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3837 code match comment, rejecting invalid SESE regions.
3839 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3841 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3842 require base operand is a REG_POINTER prior to reload on targets
3843 with non-equivalent space registers.
3845 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3847 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3848 (update_specialized_profile): Fix updating of counts.
3849 (perhaps_add_new_callers): Likewise.
3851 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3854 * x86-tune.def: Remove obsolette FIXMEs.
3855 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3856 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3857 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3859 (X86_TUNE_PAD_RETURNS): Disable for generic.
3861 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3863 PR tree-optimization/83170
3864 PR tree-optimization/83241
3865 * gimple-ssa-store-merging.c
3866 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3867 gimple_vuse (ins_stmt) in case it has changed.
3868 (imm_store_chain_info::output_merged_store): Likewise.
3870 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3874 * tree-cfg.c (pass_warn_function_return::execute): Handle
3875 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3876 with BUILTINS_LOCATION.
3880 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3881 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3882 instead of DECL_MODE.
3884 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3885 last argument to rtx pointer.
3886 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3887 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3888 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3889 depending on the chosen ISAs.
3890 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3891 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3892 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3894 * config/i386/sse.md (mov<mode>_internal): Likewise.
3895 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3897 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3899 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3901 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3903 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3904 parameters from prototype.
3905 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3906 parameters. Don't print a comment.
3907 (emit_fusion_gpr_load): Adjust.
3908 (emit_fusion_load_store): Adjust.
3909 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3910 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3911 comment on the second line.
3913 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3916 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3917 rs6000_cpu based on cpu_index, not tune_index.
3919 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3921 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3922 which_alternative instead of which_alternative + 1.
3923 (output_asm_insn): Print an extra tab if the template is short.
3925 2017-12-01 Jim Wilson <jimw@sifive.com>
3927 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3929 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3930 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3932 * doc/tm.texi: Regenerate.
3934 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3937 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3938 whether we can allocate pseudos before trying to fix an address.
3939 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3940 memory address is indexed or indirect.
3941 (floatuns_<mode>si2_hw2): Likewise.
3943 2017-12-01 Jason Merrill <jason@redhat.com>
3945 * Makefile.in (TAGS): Add c-family/*.cc.
3947 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3949 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3950 (call_value_insn): Likewise.
3951 (sibcall_insn): Likewise.
3952 (sibcall_value_insn): Likewise.
3953 (movsi_aarch64): Likewise.
3954 (movdi_aarch64): Likewise.
3955 (add_losym_): Likewise.
3956 (ldr_got_small_): Likewise.
3957 (ldr_got_small_sidi): Likewise.
3958 (ldr_got_small_28k_): Likewise.
3959 (ldr_got_small_28k_sidi): Likewise.
3960 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3961 Move output_addr_const to symbolic case. Add error check.
3963 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3965 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3966 * config/s390/s390.md: Disable MVC merging peephole if it would
3967 disable operand forwarding.
3968 (new peephole2): Split MVCs if it would turn them into up to 2
3971 2017-12-01 Richard Biener <rguenther@suse.de>
3973 PR tree-optimization/83232
3974 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3975 detection of same access. Instead of breaking the group here
3976 do not consider the duplicate. Add comment explaining real fix.
3978 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3980 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3982 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3984 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3986 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3988 * function.h (struct function): Remove cilk_frame_decl,
3989 is_cilk_function and calls_cilk_spawn fields.
3990 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3992 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3993 cilk_elemental field.
3994 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3995 * target.def: Adjust comment.
3996 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3997 Don't test cilk_elemental.
3999 PR tree-optimization/83233
4000 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4001 bswap_stat name for the struct.
4004 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4005 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4006 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4007 SWITCH_BREAK_LABEL_P set on the label.
4008 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4009 added for default case if it was missing and not all cases covered.
4010 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4011 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4013 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4014 to avoid -Wimplicit-fallthrough warning.
4015 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4016 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4018 PR tree-optimization/83221
4019 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4021 (init_reassoc): Formatting fix.
4024 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4025 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4029 * tree-cfg.c: Include asan.h.
4030 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4032 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4034 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4036 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4038 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4039 vpcmpeqd instruction.
4041 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4043 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4044 registers type for 128bit mode.
4046 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4048 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4050 * gengtype-state.c (read_state_pair): Likewise.
4051 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4052 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4053 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4054 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4055 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4056 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4057 * builtins.c (fold_builtin_3): Likewise.
4058 * graphite-scop-detection.c
4059 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4060 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4063 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4066 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4067 parameter. Return true if ignore_decl_rtl_set_p is true and other
4068 conditions are satisfied.
4069 * asan.h (asan_protect_global): Add new parameter.
4070 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4071 to asan_protect_global calls.
4073 2017-11-30 Jim Wilson <jimw@sifive.com>
4075 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4076 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4077 -mstrict-align, add info on default value. Delete redundant lines for
4078 -mabi. Add missing -mexplicit-relocs docs.
4080 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4082 * config/arc/arc.md (trap): New pattern.
4084 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4086 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4087 instruction to end into a delay slot.
4088 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4089 can be placed into a delay slot against reg_note.
4091 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4093 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4094 labels number of usages.
4096 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4098 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4100 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4102 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4105 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4106 multiplication by power of 2 constant into two shifts + comparison.
4108 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4111 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4113 2017-11-30 Richard Biener <rguenther@suse.de>
4115 PR tree-optimization/83202
4116 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4117 allow_peel argument and guard peeling.
4118 (canonicalize_loop_induction_variables): Likewise.
4119 (canonicalize_induction_variables): Pass false.
4120 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4121 peeling from cunrolli.
4123 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4125 * combine.c (try_combine): Print a message to dump file whenever
4126 I0, I1, or I2 cannot be combined into I3.
4128 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4130 PR rtl-optimization/83156
4131 PR rtl-optimization/82621
4132 * combine.c (try_combine): Don't split an I2 if one of the dests is
4133 set again before I3. Allow unused dests.
4135 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4137 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4139 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4141 PR rtl-optimization/80818
4142 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4143 recursively. Use insn code for clobber.
4144 (lra_set_insn_recog_data): Pass the new arg to
4145 collect_non_operand_hard_regs.
4146 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4148 (lra_update_insn_regno_info): Pass insn to
4149 add_regs_to_insn_regno_info.
4151 2017-11-29 Jim Wilson <jimw@sifive.com>
4152 Andrew Waterman <andrew@sifive.com>
4154 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4155 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4156 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4157 use SINGLE_SHIFT_COST.
4158 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4159 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4161 2017-11-29 Julia Koval <julia.koval@intel.com>
4163 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4164 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4165 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4166 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4167 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4168 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4169 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4170 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4171 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4172 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4173 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4174 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4175 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4176 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4177 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4178 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4179 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4180 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4181 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4182 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4183 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4184 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4185 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4186 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4187 __builtin_ia32_vpshldv_v2di_mask,
4188 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4189 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4190 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4192 2017-11-29 Julia Koval <julia.koval@intel.com>
4194 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4195 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4196 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4197 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4198 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4199 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4200 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4201 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4202 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4203 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4204 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4205 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4206 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4207 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4208 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4209 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4210 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4211 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4212 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4213 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4214 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4216 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4218 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4219 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4222 2017-11-29 Martin Aberg <maberg@gaisler.com>
4224 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4225 to prevent b2bst errata sequence.
4226 (sqrtdf2_fix): Likewise.
4228 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4230 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4231 (div_sqrt_insn_p): New function.
4232 (sparc_do_work_around_errata): Insert NOP instructions to
4233 prevent sequences that could trigger the TN-0013 errata for
4234 certain LEON3 processors.
4235 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4236 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4237 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4238 (in_branch_delay): Prevent div and sqrt in delay slot if
4240 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4242 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4244 * config/sparc/sparc.c (atomic_insn_p): New function.
4245 (sparc_do_work_around_errata): Insert NOP instructions to
4246 prevent sequences that could trigger the TN-0010 errata for
4248 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4249 instruction referable in atomic_insns_p.
4251 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4253 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4254 (atomic_compare_and_swap_leon3_1): Likewise.
4257 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4259 * config/sparc/sparc.c (fpop_insn_p): New function.
4260 (sparc_do_work_around_errata): Insert NOP instructions to
4261 prevent sequences that could trigger the TN-0012 errata for
4263 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4264 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4265 (in_branch_annul_delay): Prevent floating-point instructions
4266 in delay slot of annulled integer branch.
4268 2017-11-29 Richard Biener <rguenther@suse.de>
4270 PR tree-optimization/83202
4271 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4273 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4274 nodes vectorized to the same stmts multiple times.
4275 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4276 (vect_analyze_slp_instance): Adjust.
4277 (scalar_stmts_to_slp_tree_map_t): New typedef.
4278 (vect_schedule_slp_instance): Add a map recording the SLP node
4279 representing the vectorized stmts for a set of scalar stmts.
4280 Avoid code-generating redundancies.
4281 (vect_schedule_slp): Allocate map and pass it down.
4283 2017-11-29 Nathan Sidwell <nathan@acm.org>
4286 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4289 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4291 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4292 cases for printing LDP/STP memory addresses.
4293 (aarch64_print_address_internal): Renamed from
4294 aarch64_print_operand_address, added parameter, add Pmode check.
4295 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4296 (aarch64_print_operand_address): Indirect to
4297 aarch64_print_address_internal.
4298 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4299 'y' operand output specifier.
4301 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4304 * tree.c (build_simple_mem_ref_loc): Handle
4305 get_addr_base_and_unit_offset returning a MEM_REF.
4308 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4311 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4314 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4316 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4318 (layout::move_to_column): Likewise.
4320 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4323 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4324 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4326 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4327 Martin Jambor <mjambor@suse.cz>
4330 * tree.h (expr_type_first_operand_type_p): Declare
4331 * tree.c (expr_type_first_operand_type_p): New function.
4332 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4333 (ipa_value_from_jfunc): Adjust declaration.
4334 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4335 Use it as result type for arithmetics, unless it is NULL in which case
4336 be more conservative.
4337 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4338 ipa_get_jf_pass_through_result.
4339 (propagate_vals_across_pass_through): Likewise.
4340 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4341 is to propagate_vals_across_pass_through.
4342 (propagate_constants_across_call): Pass PARM_TYPE to
4343 propagate_scalar_across_jump_function.
4344 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4345 ipa_value_from_jfunc.
4346 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4347 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4348 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4349 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4350 target_type, pass it to ipa_value_from_jfunc.
4351 (update_indirect_edges_after_inlining): Pass parameter type to
4352 try_make_edge_direct_simple_call.
4354 2017-11-28 Jeff Law <law@redhat.com>
4356 * gimple-ssa-evrp-analyze.c
4357 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4358 refine ranges if scev_initialized_p returns true.
4359 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4361 2017-11-28 Julia Koval <julia.koval@intel.com>
4363 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4364 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4365 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4366 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4367 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4368 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4369 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4370 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4371 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4372 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4373 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4374 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4375 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4376 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4377 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4378 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4379 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4380 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4381 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4382 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4383 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4384 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4386 2017-11-28 Julia Koval <julia.koval@intel.com>
4388 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4389 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4390 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4391 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4392 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4393 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4394 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4395 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4396 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4397 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4398 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4399 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4400 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4401 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4402 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4403 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4404 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4405 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4406 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4407 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4408 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4409 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4410 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4411 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4412 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4413 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4414 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4415 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4416 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4417 __builtin_ia32_vpshld_v2di_mask): New builtins.
4418 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4419 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4421 2017-11-28 Richard Biener <rguenther@suse.de>
4423 PR tree-optimization/80776
4424 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4426 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4428 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4429 If the incoming edge is an effective fallthru because the other
4430 edge only reaches a __builtin_unreachable () then record ranges
4431 derived from the controlling condition in SSA info.
4432 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4433 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4435 2017-11-28 Olivier Hainque <hainque@adacore.com>
4437 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4438 on mingw build hosts.
4440 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4443 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4445 2017-11-28 Richard Biener <rguenther@suse.de>
4448 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4449 copies generated from memcpy use a character array as reference
4452 2017-11-28 Julia Koval <julia.koval@intel.com>
4453 Sebastian Peryt <sebastian.peryt@intel.com>
4455 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4456 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4457 cilk-common.o, cilk.h, cilk-common.c): Remove.
4459 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4460 * builtins.c (is_builtin_name): Remove cilkplus condition.
4461 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4462 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4463 cilk-builtins.def, cilkplus.def): Remove.
4464 * cif-code.def (CILK_SPAWN): Remove.
4465 * cilk-builtins.def: Delete.
4466 * cilk-common.c: Ditto.
4468 * cilkplus.def: Ditto.
4469 * config/darwin.h (fcilkplus): Delete.
4470 * cppbuiltin.c: Ditto.
4471 * doc/extend.texi: Remove cilkplus doc.
4472 * doc/generic.texi: Ditto.
4473 * doc/invoke.texi: Ditto.
4474 * doc/passes.texi: Ditto.
4475 * gcc.c (fcilkplus): Remove.
4476 * gengtype.c (cilk.h): Remove.
4477 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4479 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4481 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4483 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4484 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4485 cilkplus conditions.
4486 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4487 inline_read_section): Ditto.
4488 * ipa-inline-analysis.c (cilk.h): Remove.
4489 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4490 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4491 append_linker_options): Remove condition for fcilkplus.
4492 * lto/lto-lang.c (cilk.h): Remove.
4493 (lto_init): Remove condition for fcilkplus.
4494 * omp-expand.c (expand_cilk_for_call): Delete.
4495 (expand_omp_taskreg, expand_omp_for_static_chunk,
4496 expand_omp_for): Remove cilkplus
4498 (expand_cilk_for): Delete.
4499 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4500 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4501 execute_lower_omp, diagnose_sb_0): Ditto.
4502 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4503 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4504 * tree-nested.c: Ditto.
4505 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4506 (dump_generic_node): Ditto.
4507 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4508 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4510 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4512 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4514 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4515 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4516 for vector mode and !TARGET_SIMD.
4518 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4520 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4522 * tree.h (SWITCH_LABELS): Remove.
4523 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4524 assert SWITCH_BODY is non-NULL.
4525 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4527 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4529 PR tree-optimization/80788
4530 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4531 has TREE_OVERFLOW set, call drop_tree_overflow.
4533 2017-11-28 Richard Biener <rguenther@suse.de>
4535 PR tree-optimization/83158
4536 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4538 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4541 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4542 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4544 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4546 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4547 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4549 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4552 * varasm.c (compare_constant): Take the mode of the constants into
4553 account when comparing floating point constants.
4555 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4557 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4558 from explicit instantiation of debug_helper.
4559 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4561 2017-11-27 Richard Biener <rguenther@suse.de>
4563 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4566 2017-11-27 Richard Biener <rguenther@suse.de>
4568 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4569 (type_hash_canon): Also clear int_cst_hash_table entry for
4571 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4573 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4575 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4576 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4577 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4579 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4581 * hash-map.h (gt_cleare_cache): Avoid UB.
4583 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4585 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4586 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4587 (decide_unroll_runtime_iterations): Likewise.
4588 (decide_unroll_stupid): Likewise.
4590 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4593 * config/i386/i386.md: Add a loop with incssp.
4595 2017-11-27 Martin Jambor <mjambor@suse.cz>
4597 PR tree-optimization/81248
4598 * tree-sra.c (splice_param_accesses): Remove size check.
4599 (decide_one_param_reduction): Fix size check.
4600 * gimple-pretty-print.c (dump_profile): Silence warning.
4601 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4603 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4606 * dbxout.c (lastlineno): New variable.
4607 (dbx_debug_hooks): Use dbxout_switch_text_section as
4608 switch_text_section debug hook.
4609 (dbxout_function_end): Switch to current_function_section
4610 rather than function_section. If crtl->has_bb_partition,
4611 output just one N_FUN, depending on in_cold_section_p.
4612 (dbxout_source_line): Remember last lineno in lastlineno.
4613 (dbxout_switch_text_section): New function.
4614 (dbxout_function_decl): Adjust dbxout_block caller.
4615 (dbx_block_with_cold_children): New function.
4616 (dbxout_block): Return true if any LBRAC/RBRAC have been
4617 emitted. Use dbx_block_with_cold_children at depth == 0
4618 in second partition. Add PARENT_BLOCKNUM argument, pass
4619 it optionally adjusted to children. Output LBRAC/RBRAC
4620 around recursive call only if the block is in the current
4621 partition, if not and anything was output, emit empty
4623 * final.c (final_scan_insn): Compute cold_function_name
4624 before calling switch_text_section debug hook. Call
4625 that hook even if dwarf2out_do_frame if not emitting
4629 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4630 TREE_READONLY decls.
4632 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4634 PR rtl-optimization/82488
4635 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4637 2017-11-26 Julia Koval <julia.koval@intel.com>
4639 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4641 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4644 2017-11-26 Julia Koval <julia.koval@intel.com>
4646 * config/i386/driver-i386.c (host_detect_local_cpu):
4647 Detect skylake-avx512.
4649 2017-11-26 Julia Koval <julia.koval@intel.com>
4651 * config.gcc: Add -march=cannonlake.
4652 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4653 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4654 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4655 (PTA_CANNONLAKE): New.
4656 (processor_target_table): Add cannonlake.
4657 (ix86_option_override_internal): Ditto.
4658 (fold_builtin_cpu): Ditto.
4659 (get_builtin_code_for_version): Handle cannonlake.
4660 (M_INTEL_COREI7_CANNONLAKE): New.
4661 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4662 * doc/invoke.texi: Add -march=cannonlake.
4664 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4666 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4667 (try_init_one_plugin): Alternative implementation for MinGW.
4668 * Makefile.in (plugin_implib): New.
4669 (gengtype-lex.c): Fix broken AIX workaround.
4670 * configure: Regenerate.
4671 * doc/plugins.texi: Document support for MinGW.
4673 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4675 PR rtl-optimization/81553
4676 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4677 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4678 is a shift where C1 has different mode than the whole shift, use C1's
4679 mode for MULT rather than the shift's mode.
4682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4683 builtins not enabled in the currently selected ISA.
4685 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4687 PR tree-optimization/71026
4688 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4689 (insert_reciprocals): Change to insert reciprocals before a division
4690 by a square and to insert the square of a reciprocal.
4691 (execute_cse_reciprocals_1): Change to consider division by a square.
4692 (register_division_in): Add importance parameter.
4694 2017-11-24 Richard Biener <rguenther@suse.de>
4696 PR tree-optimization/82402
4697 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4698 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4700 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4702 * match.pd (0-ptr): New transformation.
4704 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4707 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4708 when skipping checking.
4710 2017-11-24 Richard Biener <rguenther@suse.de>
4712 PR tree-optimization/83128
4713 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4714 (vn_reference_lookup_3): Likewise.
4716 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4719 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4720 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4723 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4724 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4725 check for the latter with gimple_call_builtin_p. Do not handle
4726 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4728 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4730 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4733 2017-11-23 Julia Koval <julia.koval@intel.com>
4735 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4736 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4737 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4738 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4739 _mm512_maskz_expandloadu_epi16): New intrinsics.
4740 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4741 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4742 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4743 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4744 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4745 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4746 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4747 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4748 _mm256_maskz_expandloadu_epi8): New intrinsics.
4749 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4750 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4751 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4752 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4753 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4754 * config/i386/sse.md (VI248_VLBW): New iterator.
4755 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4757 2017-11-23 Julia Koval <julia.koval@intel.com>
4759 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4760 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4761 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4762 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4763 _mm512_mask_compressstoreu_epi16): New.
4764 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4765 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4766 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4767 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4768 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4769 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4770 _mm256_mask_compressstoreu_epi8): New.
4771 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4772 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4773 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4774 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4775 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4776 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4777 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4778 __builtin_ia32_compresshi128_mask,
4779 __builtin_ia32_compressstoreuqi512_mask,
4780 __builtin_ia32_compressstoreuhi512_mask,
4781 __builtin_ia32_compressstoreuqi256_mask,
4782 __builtin_ia32_compressstoreuqi128_mask,
4783 __builtin_ia32_compressstoreuhi256_mask,
4784 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4785 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4787 (ix86_expand_special_args_builtin): Handle new types.
4788 (s4fma_expand): Handle new builtin array.
4789 * config/i386/immintrin.h: Include new headers.
4790 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4791 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4793 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4795 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4796 fixes. Declare temp and g variables at the top in order to avoid
4797 {} in most of the cases.
4799 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4801 * match.pd (ptr-0): New transformation.
4803 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4805 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4806 (struct addr_mode_cost_table): New.
4807 (struct tune_params): Add field addr_mode_costs.
4808 * config/arm/arm.c (generic_addr_mode_costs): New.
4809 (arm_slowmul_tune): Initialise addr_mode_costs field.
4810 (arm_fastmul_tune): Likewise.
4811 (arm_strongarm_tune): Likewise.
4812 (arm_xscale_tune): Likewise.
4813 (arm_9e_tune): Likewise.
4814 (arm_marvell_pj4_tune): Likewise.
4815 (arm_v6t2_tune): Likewise.
4816 (arm_cortex_tune): Likewise.
4817 (arm_cortex_a8_tune): Likewise.
4818 (arm_cortex_a7_tune): Likewise.
4819 (arm_cortex_a15_tune): Likewise.
4820 (arm_cortex_a35_tune): Likewise.
4821 (arm_cortex_a53_tune): Likewise.
4822 (arm_cortex_a57_tune): Likewise.
4823 (arm_exynosm1_tune): Likewise.
4824 (arm_xgene1_tune): Likewise.
4825 (arm_cortex_a5_tune): Likewise.
4826 (arm_cortex_a9_tune): Likewise.
4827 (arm_cortex_a12_tune): Likewise.
4828 (arm_cortex_a73_tune): Likewise.
4829 (arm_v7m_tune): Likewise.
4830 (arm_cortex_m7_tune): Likewise.
4831 (arm_v6m_tune): Likewise.
4832 (arm_fa726te_tune): Likewise.
4833 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4835 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4837 * config/arm/arm.c (arm_mem_costs): New function.
4838 (arm_rtx_costs_internal): Use arm_mem_costs.
4840 2017-11-23 Mark Wielaard <mark@klomp.org>
4842 * dwarf2out.c (init_sections_and_labels): Use generation to create
4843 unique ranges_section_label and ranges_base_label. Return generation.
4844 (output_rnglists): Add generation argument. Use generation to create
4845 unique ranges labels.
4846 (dwarf2out_finish): Get generation from init_sections_and_labels
4847 and pass generation to output_rnglists.
4849 2017-11-23 Mike Stump <mikestump@comcast.net>
4850 Eric Botcazou <ebotcazou@adacore.com>
4852 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4853 * cfgloop.h (struct loop): Add unroll field.
4854 * function.h (struct function): Add has_unroll bitfield.
4855 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4856 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4857 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4858 (pass_rtl_unroll_loops::gate): Likewise.
4859 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4860 for which loop->unroll==1.
4861 (decide_unroll_constant_iterations): Use note for consistency and
4862 take loop->unroll into account. Return early if loop->unroll is set.
4863 Fix thinko in existing test.
4864 (decide_unroll_runtime_iterations): Use note for consistency and
4865 take loop->unroll into account.
4866 (decide_unroll_stupid): Likewise.
4867 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4868 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4869 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4871 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4872 (print_loop): Print loop->unroll if set.
4873 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4874 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4875 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4877 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4878 loop->unroll is set and smaller than the trip count. Otherwise bypass
4879 entirely the heuristics if loop->unroll is set. Remove dead note.
4880 Fix off-by-one bug in other note.
4881 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4882 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4884 (tree_unroll_loops_completely): Make static.
4885 (pass_complete_unroll::execute): Use correct type for variable.
4886 (pass_complete_unrolli::execute): Fix formatting.
4887 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4889 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4891 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4892 enable when TARGET_PREFER_AVX128 is set.
4894 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4896 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4898 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4899 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4901 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4903 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4904 frequencies. Fix estimation of aggregate parameters.
4906 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4908 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4911 2017-11-23 Tom de Vries <tom@codesourcery.com>
4913 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4915 2017-11-23 Tom de Vries <tom@codesourcery.com>
4917 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4920 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4923 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4924 sibcall_value_pcrel_fdpic): Use local variable instead of
4926 (calli_tbr_rel): Add missing operand 2.
4927 (call_valuei_tbr_rel): Add missing operand 3.
4929 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4932 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4933 bitpos/bitsize covering the whole destination, use store_expr only if
4934 the complex mode is the same. Otherwise, use expand_normal and if
4935 it returns CONCAT, subreg each part separately instead of trying to
4936 subreg the whole result.
4938 2017-11-23 Richard Biener <rguenther@suse.de>
4940 PR tree-optimization/23094
4941 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4942 come here from walking over backedges in the first iteration.
4943 (vn_reference_lookup_3): Skip clobbers that store the same value.
4945 2017-11-23 Richard Biener <rguenther@suse.de>
4947 PR tree-optimization/81403
4948 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4949 a block we need a leader relative to.
4950 (phi_translate_1): For nary processing require a leader from
4951 get_representative_for given we run expression simplification
4952 using match-and-simplify. Remove previous fix.
4954 2017-11-22 Jeff Law <law@redhat.com>
4956 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4957 Use new method allocate_value_range rather than accessing the
4958 vrp_value_range_pool data member directly.
4959 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4960 to use extract_range_from_stmt method to avoid need for
4961 extract_range_from_assignment method.
4962 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4963 method rather than setting values_propgated data member directly.
4964 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4965 and values propagated data members and extract_range_from_assignment
4966 method. Reorder private data members to conform to standards.
4967 Add new methods set_lattice_propagation_complete and
4968 allocate_value_range.
4970 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4972 PR rtl-optimization/83030
4973 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4974 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4975 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4976 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4977 CROSSING_JUMP_P flag is already set before setting it.
4978 * cfgrtl.c (fixup_partition_crossing): Likewise.
4979 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4982 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4984 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4985 Handle the case where both arguments are using gen_const_vec_series.
4987 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4990 * pretty-print.c (pp_format): Move quoting implementation to
4991 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4992 to pass address of "quote" local.
4993 (pp_begin_quote): New function.
4994 (pp_end_quote): New function.
4995 * pretty-print.h (printer_fn): Convert penultimate param from bool
4997 (pp_begin_quote): New decl.
4998 (pp_end_quote): New decl.
4999 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5000 param from bool to bool *.
5001 * tree-diagnostic.h (default_tree_printer): Likewise.
5003 2017-11-22 Jeff Law <law@redhat.com>
5005 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5006 Initialize vr_values.
5007 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5008 vr_values via vr_values class instance rather than delegators.
5009 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5010 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5011 (evrp_range_analyzer::push_value_range): Likewise.
5012 (evrp_range_analyzer::pop_value_range): Likewise.
5013 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5014 most delegators. Those remaining are exposed as public interfaces.
5015 Make vr_values a pointer and private.
5016 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5018 (evrp_range_analyzer::get_vr_value): New method.
5019 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5020 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5021 (class evrp_dom_walker): Attach evrp_folder class, initialize
5022 it in the ctor. Remove temporary delegators.
5023 (evrp_dom_walker::before_dom_children): Call methods in attached
5024 evrp_range_analyzer class via class instance pointer. Use
5025 free value_range_constant_singleton to remove need for
5026 op_with_constant_singleton_value delegator method. Do not
5027 create a vrp_prop class instance for every call! Narrow
5028 scope of a couple variables.
5029 (evrp_dom_walker::cleanup): Call methods in attached
5030 evrp_range_analyzer class via class instance pointer.
5031 * vr-values.h (class vr_values): Privatize many methods and
5034 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5036 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5037 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5038 * expr.c (expand_expr_real_2): Likewise.
5039 * fold-const.c (const_unop): Likewise.
5040 * optabs-tree.c (optab_for_tree_code): Likewise.
5041 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5042 * tree-inline.c (estimate_operator_cost): Likewise.
5043 * tree-pretty-print.c (dump_generic_node): Likewise.
5044 (op_code_prio): Likewise.
5045 (op_symbol_code): Likewise.
5046 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5047 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5048 * internal-fn.c (direct_internal_fn_optab): New function.
5049 (direct_internal_fn_array, direct_internal_fn_supported_p
5050 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5051 * fold-const-call.c (fold_const_reduction): New function.
5052 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5054 * tree-vect-loop.c: Include internal-fn.h.
5055 (reduction_code_for_scalar_code): Rename to...
5056 (reduction_fn_for_scalar_code): ...this and return an internal
5058 (vect_model_reduction_cost): Take an internal_fn rather than
5060 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5062 (vectorizable_reduction): Use internal functions rather than tree
5063 codes for the reduction operation. Update calls to the functions
5065 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5066 Use calls to internal functions rather than REDUC tree codes.
5067 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5069 2017-11-22 Olivier Hainque <hainque@adacore.com>
5071 * config/vxworks.c (vxworks_override_options): Pick default
5072 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5073 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5074 DWARF_GNAT_ENCODINGS_DEFAULT.
5075 * config/vxworksae.h: Likewise.
5077 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5079 PR tree-optimization/83104
5080 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5083 2017-11-22 Marek Polacek <polacek@redhat.com>
5084 H.J. Lu <hongjiu.lu@intel.com>
5085 Jason Merrill <jason@redhat.com>
5090 * calls.c (initialize_argument_information): Call
5091 warn_parameter_passing_abi target hook.
5092 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5093 argument onto stack.
5094 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5095 * common.opt: Update -fabi-version description.
5096 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5097 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5099 (ix86_is_empty_record): New function.
5100 (ix86_warn_parameter_passing_abi): New function.
5101 (TARGET_EMPTY_RECORD_P): Redefine.
5102 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5103 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5104 * doc/tm.texi: Regenerated.
5105 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5106 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5107 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5108 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5109 instead of int_size_in_bytes.
5110 * expr.c (copy_blkmode_to_reg): Likewise.
5111 * function.c (aggregate_value_p): Return 0 for empty types.
5112 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5114 (locate_and_pad_parm): Call arg size_in_bytes instead
5116 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5117 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5118 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5120 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5121 (std_gimplify_va_arg_expr): Skip empty records. Call
5122 arg_size_in_bytes instead size_in_bytes.
5123 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5124 * tree-core.h (tree_type_common): Add empty_flag.
5125 (tree_decl_common): Update comments.
5126 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5128 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5129 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5131 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5132 * tree.c (default_is_empty_type): New function.
5133 (default_is_empty_record): New function.
5134 (arg_int_size_in_bytes): New function.
5135 (arg_size_in_bytes): New function.
5136 (get_ultimate_context): New function.
5137 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5138 TRANSLATION_UNIT_WARN_EMPTY_P.
5139 (default_is_empty_record, arg_int_size_in_bytes,
5140 arg_size_in_bytes, get_ultimate_context): Declare.
5142 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5144 * config/arm/arm.c (cmse_clear_registers): New function.
5145 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5146 code by call to cmse_clear_registers.
5147 (cmse_nonsecure_entry_clear_before_return): Likewise.
5149 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5151 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5152 (vdot_s32, vdotq_s32): New.
5153 (vdot_lane_u32, vdotq_lane_u32): New.
5154 (vdot_lane_s32, vdotq_lane_s32): New.
5157 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5160 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5161 for unsigned values with fewer HWIs than the precision.
5162 (test_overflow): New function.
5163 (wide_int_cc_tests): Call it.
5165 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5167 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5168 has the same mode class as Pmode.
5170 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5172 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5173 padding_bits_to_clear_ptr.
5174 (cmse_nonsecure_entry_clear_before_return): Likewise.
5176 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5178 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5179 auto_sbitap instead of integer bitfield to control register needing
5182 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5184 PR tree-optimization/83044
5185 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5186 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5187 that need the upper bound. Subtract offset from
5188 get_addr_base_and_unit_offset only if positive and subtract it
5189 before division by eltsize rather than after it.
5192 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5193 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5194 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5198 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5199 expand_binop with *mul_widen_optab, make sure at least one of the
5200 operands doesn't have VOIDmode.
5203 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5205 PR rtl-optimization/82044
5206 PR tree-optimization/82042
5207 * dse.c (record_store): Check for overflow.
5208 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5209 add_wild_read instead of clear_rhs_from_active_local_stores on
5210 overflow and log it into dump_file.
5212 2017-11-22 Richard Biener <rguenther@suse.de>
5214 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5215 fake edges to exit when looking for a place to insert.
5216 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5218 (insert_into_preds_of_block): Commit edge insertion immediately,
5219 assert that doesn't require new BBs.
5220 (fini_pre): Release expressions.
5221 (pass_pre::execute): Shuffle things around a bit, if the fn
5222 is too large do not compute AVAIL either as this is really the
5225 2017-11-22 Richard Biener <rguenther@suse.de>
5227 PR tree-optimization/83089
5228 * tree-if-conv.c (pass_if_conversion::execute): If anything
5229 changed reset SCEV and free the number of iteration estimates.
5231 2017-11-21 Martin Sebor <msebor@redhat.com>
5233 PR tree-optimization/82945
5234 * calls.h (warn_nonstring_bound): Remove unused function.
5236 2017-11-21 Martin Sebor <msebor@redhat.com>
5238 PR tree-optimization/82945
5239 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5240 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5241 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5243 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5244 * calls.h (get_attr_nonstring_decl): Declare new function.
5245 * doc/extend.texi (attribute nonstring): Update.
5246 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5247 get_attr_nonstring_decl and handle it.
5248 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5249 detection of nul-termination.
5250 (strlen_to_stridx): Change to a pointer.
5251 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5252 (pass_strlen::execute): Same.
5254 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5256 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5257 for the new option -mprefer-vector-width=[none|128|256|512].
5258 * config/i386/i386.c (ix86_target_string): remove old style options
5259 -mprefer-avx256 and make -mprefer-avx128 as alias.
5260 (ix86_option_override_internal): Apply defaults for the
5261 -mprefer-vector-width=[128|256] option.
5262 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5263 Implement macros to work with -mprefer-vector-width=.
5264 * config/i386/i386.opt: Implemented option
5265 -mprefer-vector-width=[none|128|256|512].
5266 * doc/invoke.texi: Documentation for
5267 -mprefer-vector-width=[none|128|256|512].
5269 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5271 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5273 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5275 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5277 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5279 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5281 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5282 * expr.c (expand_expr_real_2): Likewise.
5283 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5284 fold_binary_loc): Likewise.
5285 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5286 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5287 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5288 MINUS_EXPR transformations.
5289 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5290 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5291 * tree-inline.c (estimate_operator_cost): Likewise.
5292 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5293 op_symbol_code): Likewise.
5294 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5295 * vr-values.c (extract_range_from_binary_expr): Likewise.
5296 * varasm.c (initializer_constant_valid_p_1): Likewise.
5297 * tree.def: New tree code POINTER_DIFF_EXPR.
5299 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5301 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5302 integer suffix to movbe mnemonic.
5303 (*bswaphi2_movbe): Ditto.
5304 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5306 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5309 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5310 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5311 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5313 2017-11-21 Martin Liska <mliska@suse.cz>
5315 * tree-inline.c (expand_call_inline): Remove not needed
5318 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5319 Jakub Jelinek <jakub@redhat.com>
5322 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5323 Remove static keyword from f variable.
5325 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5327 PR tree-optimization/83086
5328 * gimple-ssa-store-merging.c
5329 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5330 rather than n.base_addr.
5332 2017-11-21 Martin Liska <mliska@suse.cz>
5334 PR rtl-optimization/82044
5335 PR tree-optimization/82042
5336 * dse.c (check_mem_read_rtx): Check for overflow.
5338 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5340 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5343 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5345 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5346 and its corresponding call arg location note.
5348 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5349 Andrew Burgess <andrew.burgess@embecosm.com>
5351 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5353 (arc_return_slot_offset): Likewise.
5354 (arc_eh_return_address_location): New declaration.
5355 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5356 (MUST_SAVE_REGISTER): Add exception handler case.
5357 (MUST_SAVE_RETURN_ADDR): Likewise.
5358 (arc_frame_pointer_required): Likewise.
5359 (arc_frame_pointer_needed): New function.
5360 (arc_compute_frame_size): Changed.
5361 (arc_expand_prologue): Likewise.
5362 (arc_expand_epilogue): Likewise.
5363 (arc_initial_elimination_offset): Likewise.
5364 (arc_return_slot_offset): Delete.
5365 (arc_eh_return_address_location): New function.
5366 (arc_builtin_setjmp_frame_value): Likewise.
5367 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5368 (EH_RETURN_STACKADJ_RTX): Define.
5369 (EH_RETURN_HANDLER_RTX): Likewise.
5370 * config/arc/arc.md (eh_return): Delete.
5372 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5374 * print-tree.h (debug_vec_tree): Remove prototype.
5375 * gdbinit.in (pvt): Remove macro.
5377 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5379 PR tree-optimization/83047
5380 * gimple-ssa-store-merging.c
5381 (imm_store_chain_info::output_merged_store): If the loads with the
5382 same vuse are in different basic blocks, for load_gsi pick a load
5383 location that is dominated by the other loads.
5386 * config/i386/i386.c (ix86_memmodel_check): Start
5387 -Winvalid-memory-model diagnostics with lowercase letter.
5390 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5391 set in_cold_section_p to the partition containing loc_list->first.
5392 When seeing loc_list->last_before_switch node, update secname and
5393 perform range_across_switch second partition handling only after that.
5396 * run-rtl-passes.c: Include debug.h.
5397 (run_rtl_passes): Call debug_hooks->assembly_start.
5398 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5402 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5403 OPTAB_DIRECT in calls to expand_simple_binop.
5405 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5408 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5410 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5413 * spellcheck.h (best_match::blithely_get_best_candidate): New
5416 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5418 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5419 with lower case letter.
5421 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5423 * config/i386/i386.md (bswaphi2): New expander.
5424 (*bswaphi2_movbe): New insn pattern.
5425 (bswaphi -> rorhi pepehole2): New peephole pattern.
5427 2017-11-20 Jeff Law <law@redhat.com>
5429 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5430 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5431 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5432 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5433 gimple-ssa-evrp-analyze.h.
5435 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5436 set BB_VISITED here.
5437 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5439 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5441 * tree-predcom.c: Add general comment on Store-Store chains.
5442 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5444 (get_chain_last_ref_at): Rename into...
5445 (get_chain_last_write_at): ...this.
5446 (get_chain_last_write_before_load): New function.
5447 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5448 CT_STORE_STORE when write reference is added.
5449 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5450 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5451 (initialize_root_vars_store_elim_1): Ditto.
5452 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5453 definition is created.
5454 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5455 chain by replacing it with dominant stored value.
5457 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5459 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5461 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5463 * vr-values.c (extract_range_from_binary_expr): Use a full range
5466 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5468 * config/arm/arm.md (R4_REGNUM): Define constant.
5469 (nonsecure_call_internal): Remove r4 clobber.
5470 (nonsecure_call_value_internal): Likewise.
5471 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5472 clobber and resequence match_operands.
5473 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5474 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5475 (nonsecure_call_value_reg_thumb2): Likewise.
5477 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5479 PR tree-optimization/78821
5480 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5481 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5482 to the address of the base rather than the base itself.
5483 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5484 (find_bswap_or_nop_finalize): New function.
5485 (find_bswap_or_nop): Use it.
5486 (bswap_replace): Return a tree rather than bool, change first
5487 argument from gimple * to gimple_stmt_iterator, allow inserting
5488 into an empty sequence, allow ins_stmt to be NULL - then emit
5489 all stmts into gsi. Fix up MEM_REF address gimplification.
5490 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5491 (struct store_immediate_info): Add N and INS_STMT non-static
5493 (store_immediate_info::store_immediate_info): Initialize them
5494 from newly added ctor args.
5495 (merged_store_group::apply_stores): Formatting fixes. Sort by
5497 (stmts_may_clobber_ref_p): For stores call also
5498 refs_anti_dependent_p.
5499 (gather_bswap_load_refs): New function.
5500 (imm_store_chain_info::try_coalesce_bswap): New method.
5501 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5502 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5503 (imm_store_chain_info::output_merged_store): Fail if number of
5504 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5505 and NOP_EXPR rhs_code.
5506 (pass_store_merging::process_store): Compute n and ins_stmt, if
5507 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5508 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5510 (pass_store_merging::execute): Calculate dominators.
5512 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5513 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5514 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5515 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5516 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5517 class pass_optimize_bswap, bswap_replace,
5518 pass_optimize_bswap::execute): Moved to ...
5519 * gimple-ssa-store-merging.c: ... this file.
5520 Include optabs-tree.h.
5521 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5522 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5523 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5524 anonymous namespace, remove static keywords.
5525 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5526 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5528 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5531 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5533 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5535 * vec.h (debug_helper): New function.
5536 (DEFINE_DEBUG_VEC): New macro.
5537 * hash-set.h (debug_helper): New function.
5538 (DEFINE_DEBUG_HASH_SET): New macro.
5539 * cfg.c (debug_slim (edge)): New function.
5540 Call DEFINE_DEBUG_VEC for edges.
5541 Call DEFINE_DEBUG_HASH_SET for edges.
5542 * cfghooks.c (debug_slim (basic_block)): New function.
5543 Call DEFINE_DEBUG_VEC for basic blocks.
5544 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5545 * print-tree.c (debug_slim): New function to handle trees.
5546 Call DEFINE_DEBUG_VEC for trees.
5547 Call DEFINE_DEBUG_HASH_SET for trees.
5548 (debug (vec<tree, va_gc>) &): Remove.
5549 (debug (<vec<tree, va_gc>) *): Remove.
5550 * print-rtl.c (debug_slim): New function to handle const_rtx.
5551 Call DEFINE_DEBUG_VEC for rtx_def.
5552 Call DEFINE_DEBUG_VEC for rtx_insn.
5553 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5554 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5555 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5556 (debug (vec<rtx_insn *> *ptr): Remove.
5557 (debug_insn_vector): Remove.
5558 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5560 2017-11-20 Tom de Vries <tom@codesourcery.com>
5562 PR rtl-optimization/82020
5563 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5564 IF_THEN_ELSE condition.
5566 2017-11-19 Jeff Law <law@redhat.com>
5568 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5569 of degenerates resulting from ignoring an edge.
5571 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5574 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5576 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5579 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5580 for uninitialied counts.
5582 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5585 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5587 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5590 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5592 2017-11-19 Tom de Vries <tom@codesourcery.com>
5594 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5597 2017-11-19 Tom de Vries <tom@codesourcery.com>
5599 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5602 2017-11-19 Tom de Vries <tom@codesourcery.com>
5604 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5607 2017-11-19 Tom de Vries <tom@codesourcery.com>
5609 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5611 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5613 2017-11-19 Tom de Vries <tom@codesourcery.com>
5615 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5617 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5618 semicolon after MCORE_EXPORT_NAME call.
5620 2017-11-19 Tom de Vries <tom@codesourcery.com>
5623 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5624 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5626 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5628 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5629 (cgraph_edge::clone): Cleanup updating of profile.
5630 * ipa-cp.c (update_profiling_info): Likewise.
5631 * ipa-inline-transform.c (inline_transform): Likewise.
5632 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5633 * ipa-split.c (execute_split_functions): Do not split when function
5635 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5636 * profile-count.c (profile_count::dump): Add global0.
5637 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5639 (profile_count::to_sreal_scale): Likewise.
5640 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5641 (profile_count::combine_with_ipa_count): New function.
5642 * profile-count.h (profile_guessed_global0adjusted): New.
5643 (profile_count::adjusted_zero): New.
5644 (profile_count::global0adjusted): New.
5645 (profile_count::combine_with_ipa_count): New.
5646 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5647 correct profile of return block of split functions.
5648 (copy_cfg_body): Remove unused profile_count.
5649 (copy_body): Likewise.
5650 (expand_call_inline): Update.
5651 (tree_function_versioning): Update.
5653 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5655 * hash-set.h (hash_set::empty): New.
5656 * tree-ssa-threadbackward.h: Delete.
5657 * tree-ssa-threadbackward.c (class thread_jumps): New.
5658 Move max_threaded_paths into class.
5659 (fsm_find_thread_path): Remove arguments that are now in class.
5660 (profitable_jump_thread_path): Rename to...
5661 (thread_jumps::profitable_jump_thread_path): ...this.
5662 (convert_and_register_jump_thread_path): Rename to...
5663 (thread_jumps::convert_and_register_current_path): ...this.
5664 (check_subpath_and_update_thread_path): Rename to...
5665 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5666 (register_jump_thread_path_if_profitable): Rename to...
5667 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5668 (handle_phi): Rename to...
5669 (thread_jumps::handle_phi): ...this.
5670 (handle_assignment): Rename to...
5671 (thread_jumps::handle_assignment): ...this.
5672 (fsm_find_control_statement_thread_paths): Rename to...
5673 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5674 (find_jump_threads_backwards): Rename to...
5675 (thread_jumps::find_jump_threads_backwards): ...this.
5676 Initialize path local data.
5677 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5678 from within thread_jumps class.
5679 (pass_early_thread_jumps::execute): Same.
5681 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5683 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5685 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5687 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5688 to canonical location.
5690 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5693 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5695 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5697 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5699 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5700 rcrt1.o%s/grcrt1.o%s for -static-pie.
5702 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5704 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5705 ix86_shift_rotate_cost): Break out from ...
5706 (ix86_rtx_costs): ... here.
5707 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5710 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5712 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5713 when entry block was promoted unlikely.
5714 (estimate_bb_frequencies): Increase frequency scale.
5715 * profile-count.h (profile_count): Export precision info.
5717 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5719 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5720 disturbin profile of entry block.
5722 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5724 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5727 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5729 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5730 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5732 2017-11-17 Jeff Law <law@redhat.com>
5734 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5735 from evrp_dom_walker class. Various methods moved into new class.
5736 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5737 (evrp_range_analyzer::enter): New method.
5738 (evrp_range_analyzer::leave): New method.
5739 (evrp_dom_walker): Remove delegators no longer needed by this class.
5740 Replace vr_values data member with evrp_range_analyzer
5742 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5743 method extracted from evrp_dom_walker::before_dom_children.
5744 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5745 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5747 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5748 Add private copy constructor and move assignment operators.
5749 Privatize methods and class data where trivially possible.
5750 (evrp_dom_walker::cleanup): New function, extracted from
5751 execute_early_vrp. Simplify access to class data.
5753 * vr-values.h (get_output_for_vrp): Prototype.
5754 * vr-values.c (get_output_for_vrp): New function extracted from
5755 vrp_visit_assignment_or_call and extract_range_from_stmt.
5756 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5758 2017-11-17 Luis Machado <luis.machado@linaro.org>
5760 * config/aarch64/aarch64.c
5761 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5762 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5763 tune_params::AUTOPREFETCHER_WEAK.
5765 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5768 * config/arm/arm.c (arm_valid_target_attribute_rec):
5769 Parse "arch=" and "+<ext>".
5770 (arm_valid_target_attribute_tree): Re-init global options.
5771 (arm_option_override): Make non-static.
5772 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5773 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5774 (__ARM_FEATURE_CRC32): Support undef.
5775 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5776 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5778 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5780 * gdbinit.in (break-on-diagnostic): New command.
5782 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5784 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5785 return to indirect jump for EH return if control-flow
5786 protection is enabled. Change explicit 'false' argument in
5787 pro_epilogue_adjust_stack with a value of
5789 * config/i386/i386.md (simple_return_indirect_internal):
5790 Remove SImode restriction to support 64-bit.
5792 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5794 * combine.c (added_notes_insn): New.
5795 (try_combine): Handle added_notes_insn like added_links_insn.
5796 Rewrite return value code.
5797 (distribute_notes): Set added_notes_insn to the earliest insn we added
5800 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5802 PR rtl-optimization/82621
5803 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5804 dest of one of those SETs is unused.
5806 2017-11-17 Richard Biener <rguenther@suse.de>
5809 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5810 * tree-pretty-print.c (dump_generic_node): Handle
5811 annot_expr_parallel_kind.
5812 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5813 * gimplify.c (gimple_boolify): Likewise.
5815 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5817 * config.gcc (extra_headers): Add cet.h for x86 targets.
5818 * config/i386/cet.h: New file.
5819 * doc/install.texi: Add --enable-cet/--disable-cet.
5821 2017-11-17 Richard Biener <rguenther@suse.de>
5823 PR tree-optimization/83017
5824 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5825 (gen_parallel_loop): Properly count iterations.
5826 (parallelize_loops): Handle loop->can_be_parallel independent
5827 of flag_loop_parallelize_all. Make static profitability test match
5829 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5830 * invoke.texi (parloops-min-per-thread): Document.
5832 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5834 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5835 upstreaming review comments.
5837 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5839 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5840 with fast unaligned access.
5841 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5843 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5845 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5846 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5848 2017-11-17 Richard Biener <rguenther@suse.de>
5850 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5851 folding of references.
5853 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5856 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5857 of replacing call to strncmp with corresponding call to strcmp when
5860 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5862 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5863 option prefer-avx256 for skylake-avx512 configuration.
5864 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5865 (get_builtin_code_for_version): Ditto.
5867 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5868 Monk Chiang <sh.chiang04@gmail.com>
5870 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5871 (FIXED_REGISTERS): Reserve more register numbers.
5872 (CALL_USED_REGISTERS): Likewise.
5873 (REG_ALLOC_ORDER): Likewise.
5874 (REG_CLASS_CONTENTS): Likewise.
5875 (REGISTER_NAMES): Likewise.
5877 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5878 Kito Cheng <kito.cheng@gmail.com>
5880 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5882 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5884 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5886 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5887 parameter list for vec_splats.
5889 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5891 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5893 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5895 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5898 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5899 with the required version of automake 1.11.6.
5901 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5903 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5905 (DU_C2_3_power9): Correct reservation combinations.
5906 (FP_DIV_power9, VEC_DIV_power9): New.
5907 (power9-alu): Split out rotate/shift...
5908 (power9-rot): ...to here, correct dispatch resource.
5909 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5911 (power9-fp): Correct latency.
5912 (power9-sdiv): Add div/sqrt resource.
5913 (power9-ddiv): Correct latency, add div/sqrt resource.
5914 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5915 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5917 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5919 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5921 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5922 switch statement mapping KF built-ins to TF built-ins if we don't
5923 have the proper ISA 3.0 assembler support.
5925 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5927 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5928 (lower_emutls_function_body): Do not compute it.
5930 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5932 * ipa-split.c (split_bb_info): Turn time to sreal.
5933 (split_point): Likewise.
5934 (dump_split_point): Likewise.
5935 (fine_split_points): Likewise.
5936 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5938 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5940 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5943 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5945 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5947 (expensive_function_p): Remove useless assert.
5948 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5950 2017-11-16 Martin Sebor <msebor@redhat.com>
5952 PR tree-optimization/82588
5953 PR tree-optimization/82583
5954 * tree-vrp.c (check_array_ref): Handle flexible array members,
5955 string literals, and inner indices.
5956 (search_for_addr_array): Add detail to diagnostics.
5958 2017-11-16 Nathan Sidwell <nathan@acm.org>
5962 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5963 (SET_DECL_ASSEMBLER_NAME): Forward to
5964 overwrite_decl_assembler_name.
5965 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5966 (overwrite_decl_assembler_name): Declare.
5967 * tree.c (overwrite_decl_assembler_name): New.
5968 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5969 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5970 (LANG_HOOKS_INITIALIZER): Add it.
5971 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5972 * langhooks.c (lhd_set_decl_assembler_name): Use
5973 SET_DECL_ASSEMBLER_NAME.
5974 (lhd_overwrite_decl_assembler_name): Default implementation.
5976 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5977 Jackson Woodruff <jackson.woodruff@arm.com>
5979 PR tree-optimization/71026
5980 * match.pd: Canonicalize constant multiplies in division.
5982 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5984 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5985 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5986 Remove OPT_fomit_frame_pointer entry.
5987 * common/config/arc/arc-common.c: Likewise.
5988 * common/config/arm/arm-common.c: Likewise.
5989 * common/config/avr/avr-common.c: Likewise.
5990 * common/config/c6x/c6x-common.c: Likewise.
5991 * common/config/cr16/cr16-common.c: Likewise.
5992 * common/config/cris/cris-common.c: Likewise.
5993 * common/config/epiphany/epiphany-common.c: Likewise.
5994 * common/config/fr30/fr30-common.c: Likewise.
5995 * common/config/frv/frv-common.c: Likewise.
5996 * common/config/ia64/ia64-common.c: Likewise.
5997 * common/config/iq2000/iq2000-common.c: Likewise.
5998 * common/config/lm32/lm32-common.c: Likewise.
5999 * common/config/m32r/m32r-common.c: Likewise.
6000 * common/config/mcore/mcore-common.c: Likewise.
6001 * common/config/microblaze/microblaze-common.c: Likewise.
6002 * common/config/mips/mips-common.c: Likewise.
6003 * common/config/mmix/mmix-common.c: Likewise.
6004 * common/config/mn10300/mn10300-common.c: Likewise.
6005 * common/config/nios2/nios2-common.c: Likewise.
6006 * common/config/pa/pa-common.c: Likewise.
6007 * common/config/pdp11/pdp11-common.c: Likewise.
6008 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6009 * common/config/riscv/riscv-common.c: Likewise.
6010 * common/config/rs6000/rs6000-common.c: Likewise.
6011 * common/config/rx/rx-common.c: Likewise.
6012 * common/config/s390/s390-common.c: Likewise.
6013 * common/config/sh/sh-common.c: Likewise.
6014 * common/config/sparc/sparc-common.c: Likewise.
6015 * common/config/tilegx/tilegx-common.c: Likewise.
6016 * common/config/tilepro/tilepro-common.c: Likewise.
6017 * common/config/v850/v850-common.c: Likewise.
6018 * common/config/visium/visium-common.c: Likewise.
6019 * common/config/xstormy16/xstormy16-common.c: Likewise.
6020 * common/config/xtensa/xtensa-common.c: Likewise.
6021 * invoke.texi (-fomit-frame-pointer): Update documentation.
6023 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6025 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6027 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6029 * cfg.c (scale_bbs_frequencies_int,
6030 cale_bbs_frequencies_gcov_type): Remove.
6031 * cfg.h (scale_bbs_frequencies_int,
6032 cale_bbs_frequencies_gcov_type): Remove.
6034 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6036 * tree-ssa-loop-manip.c
6037 (scale_dominated_blocks_in_loop): Update to profile counts.
6038 (tree_transform_and_unroll_loop): Likewise.
6040 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6042 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6043 scale_bbs_frequencies_int.
6045 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6047 * final.c (compute_alignments): Use counts rather than frequencies.
6049 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6051 * cfgloopanal.c: Include sreal.h
6052 (average_num_loop_insns): Use counts and sreal for accounting.
6054 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6056 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6059 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6060 Kito Cheng <kito.cheng@gmail.com>
6062 * config/nds32/constraints.md: Provide more constraints.
6063 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6064 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6065 support constraints usage.
6067 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6069 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6071 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6072 Kito Cheng <kito.cheng@gmail.com>
6074 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6075 * config/nds32/nds32.opt: Refine the layout.
6076 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6077 TARGET_EXT_STRING): Support new options.
6078 * config/nds32/nds32.h: Likewise.
6079 * config/nds32/nds32.md: Likewise.
6080 * config/nds32/nds32-predicates.c: Likewise.
6081 * config/nds32/constraints.md: Likewise.
6082 * common/config/nds32/nds32-common.c: Likewise.
6084 2017-11-16 Julia Koval <julia.koval@intel.com>
6087 * config/i386/gfniintrin.h: Add sse check.
6088 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6090 2017-11-16 Julia Koval <julia.koval@intel.com>
6092 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6093 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6094 (ix86_handle_option): Handle -mavx512vbmi2.
6095 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6096 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6097 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6098 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6099 (ix86_valid_target_attribute_inner_p): Ditto.
6100 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6101 * config/i386/i386.opt (mavx512vbmi2): New option.
6102 * doc/invoke.texi: Add new option.
6104 2017-11-16 Julia Koval <julia.koval@intel.com>
6106 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6107 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6108 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6109 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6110 _mm512_gf2p8mul_epi8): New intrinsics.
6111 * config/i386/i386-builtin-types.def
6112 (V64QI_FTYPE_V64QI_V64QI): New type.
6113 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6114 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6115 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6116 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6117 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6118 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6120 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6122 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6123 explicitly as a stream of bytes.
6125 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6127 * config/rs6000/altivec.h (vec_xst_be): New #define.
6128 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6129 and externalize from *altivec_vperm_<mode>_internal.
6130 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6132 (XL_BE_V8HI): Likewise.
6133 (XL_BE_V4SI): Likewise.
6134 (XL_BE_V4SI): Likewise.
6135 (XL_BE_V2DI): Likewise.
6136 (XL_BE_V4SF): Likewise.
6137 (XL_BE_V2DF): Likewise.
6138 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6139 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6140 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6141 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6142 VSX_BUILTIN_VEC_XST_BE.
6143 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6144 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6146 (altivec_init_builtins): Replace conditional calls to def_builtin
6147 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6148 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6149 with unconditional calls. Remove calls to def_builtin for
6150 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6151 __builtin_vec_xst_be.
6152 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6153 to define_expand, and add alternate RTL generation for P8.
6154 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6155 vsx_ld_elemrev_v8hi.
6156 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6157 add alternate RTL generation for P8.
6158 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6159 vsx_ld_elemrev_v16qi.
6160 (vsx_st_elemrev_v8hi): Convert define_insn
6161 to define_expand, and add alternate RTL generation for P8.
6162 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6163 vsx_st_elemrev_v8hi.
6164 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6165 add alternate RTL generation for P8.
6166 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6167 vsx_st_elemrev_v16qi.
6169 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6172 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6173 TARGET_AVX512ER check.
6174 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6175 neither -mzeroupper nor -mno-zeroupper is used and
6176 TARGET_EMIT_VZEROUPPER is set.
6177 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6178 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6180 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6182 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6183 folding of vector compares.
6184 (fold_build_vec_cmp): New helper function.
6185 (fold_compare_helper): New helper function.
6186 (builtin_function_type): Add compare builtins to the list of functions
6187 having unsigned arguments. Cosmetic updates to comment indentation.
6188 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6189 the not+eq combination.
6191 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6193 PR tree-optimization/82726
6194 PR tree-optimization/70754
6195 * tree-predcom.c (order_drefs_by_pos): New function.
6196 (combine_chains): Move code setting has_max_use_after to...
6197 (try_combine_chains): ...here. New parameter. Sort combined chains
6198 according to position information.
6199 (tree_predictive_commoning_loop): Update call to above function.
6200 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6202 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6204 PR tree-optimization/82726
6206 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6208 PR tree-optimization/70754
6209 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6210 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6211 combined stmt before it if not NULL.
6212 (combine_chains): Process refs reversely and compute dominance point
6216 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6218 PR tree-optimization/79663
6219 * tree-predcom.c (combine_chains): Process refs in reverse order
6220 only for ZERO length chains, and add explaining comment.
6222 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6224 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6225 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6226 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6228 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6230 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6232 2017-11-15 Martin Liska <mliska@suse.cz>
6234 * tree-cfg.c (pass_warn_function_return::execute):
6235 Compare warn_return_type for greater than zero.
6237 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6241 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6242 to return true on Xeon and not on Xeon Phi.
6243 (ix86_check_avx256_register): Changed to ...
6244 (ix86_check_avx_upper_register): ... this. Add extra check for
6245 VALID_AVX512F_REG_OR_XI_MODE.
6246 (ix86_avx_u128_mode_needed): Changed
6247 ix86_check_avx256_register to ix86_check_avx_upper_register.
6248 (ix86_check_avx256_stores): Changed to ...
6249 (ix86_check_avx_upper_stores): ... this. Changed
6250 ix86_check_avx256_register to ix86_check_avx_upper_register.
6251 (ix86_avx_u128_mode_after): Changed
6252 avx_reg256_found to avx_upper_reg_found. Changed
6253 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6254 (ix86_avx_u128_mode_entry): Changed
6255 ix86_check_avx256_register to ix86_check_avx_upper_register.
6256 (ix86_avx_u128_mode_exit): Ditto.
6257 * config/i386/i386.h: (host_detect_local_cpu): New define.
6259 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6261 * config/arm/xgene1.md (xgene1): Split into automatons
6262 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6263 (xgene1_f_load): Adjust reservations and/or types.
6264 (xgene1_f_store): Likewise.
6265 (xgene1_load_pair): Likewise.
6266 (xgene1_store_pair): Likewise.
6267 (xgene1_fp_load1): Likewise.
6268 (xgene1_load1): Likewise.
6269 (xgene1_store1): Likewise.
6270 (xgene1_move): Likewise.
6271 (xgene1_alu): Likewise.
6272 (xgene1_simd): Likewise.
6273 (xgene1_bfm): Likewise.
6274 (xgene1_neon_load1): Likewise.
6275 (xgene1_neon_store1): Likewise.
6276 (xgene1_neon_logic): Likewise.
6277 (xgene1_neon_st1): Likewise.
6278 (xgene1_neon_ld1r): Likewise.
6279 (xgene1_alu_cond): Added.
6280 (xgene1_shift_reg): Likwise.
6281 (xgene1_bfx): Likewise.
6282 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6284 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6287 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6289 (can_widen_mult_without_libcall): New function.
6290 (expand_mul_overflow): If only checking unsigned mul overflow,
6291 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6292 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6293 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6294 (expand_DIVMOD): Formatting fix.
6295 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6296 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6297 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6299 PR tree-optimization/82977
6300 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6301 constructed temporary to strlen_to_stridx.put.
6303 2017-11-15 Martin Liska <mliska@suse.cz>
6305 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6306 * configure: Regenerate.
6308 2017-11-15 Martin Liska <mliska@suse.cz>
6311 * config/sh/sh-mem.cc: Use proper probability for
6314 2017-11-14 Jeff Law <law@redhat.com>
6316 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6317 the red zone for stack_clash_protection_final_dynamic_probe targets
6318 when the total dynamic stack size is zero bytes.
6320 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6321 blocks is post order.
6323 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6325 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6326 * final.c (rest_of_clean_state): Set it for the
6327 -fcompare-debug dump.
6328 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6329 class when TDF_COMPARE_DEBUG is set.
6331 * dwarf2out.c (gen_producer_string): Discard
6334 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6337 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6338 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6339 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6340 (__TGMATH_REAL_2_3): Remove macros.
6341 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6342 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6344 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6346 (remquo): Define using __TGMATH_REAL_3.
6348 2017-11-14 Jeff Law <law@redhat.com>
6350 * vr-values.c: New file with contents extracted from tree-vrp.c.
6351 * Makefile.in (OBJS): Add vr-values.o
6352 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6353 (set_value_range, set_and_canonicalize_value_range): Likewise.
6354 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6355 (value_range_constant_singleton, symbolic_range_p): Likewise.
6356 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6357 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6358 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6359 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6360 (find_case_label_range, find_case_label_index): Likewise.
6361 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6362 (range_int_cst_singleton_p, value_inside_range): Likewise.
6363 (get_single_symbol): Likewise.
6364 (switch_update): Move structure definition here.
6365 (to_remove_edges, to_update_switch_stmts): Provide externs.
6366 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6367 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6368 (vrp_val_is_min, set_value_range): Likewise.
6369 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6370 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6371 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6372 (range_is_nonnull, range_int_cst_p): Likewwise.
6373 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6374 (get_single_symbol, operand_less_p): Likewise
6375 (compare_values_warnv, compare_values): Likewise.
6376 (value_inside_range, value_range_constant_singleton): Likewise.
6377 (zero_nonzero_bitgs_from_vr): Likewise.
6378 (extract_range_from_binary_expr_1): Likewise.
6379 (overflow_comparison_p): Likewise.
6380 (to_remove_edges, to_update_switch_stmts): Likewise.
6381 (find_case_label-index, find_case_label_range): Likewise.
6382 (switch_update, set_value_range_to_nonnegative): Remove.
6383 (set_value_range_to_truthvalue): Likewise.
6384 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6385 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6386 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6387 (find_case_label_ranges, test_for_singularity): Likewise.
6388 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6389 (x_vr_values): Move to its remaining use site.
6391 2017-11-10 Jeff Law <law@redhat.com>
6393 * vr-values.h (VR_INITIALIZER): Move #define here.
6394 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6395 * Makefile.in (OBJS): Add tree-evrp.o
6396 * tree-vrp.h (assert_info): Move structure definition here.
6397 (set_value_range_to_varying): Prototype.
6398 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6399 (infer_value_range, register_edge_assert_for): Likewise.
6400 (stmt_interesting_for_vrp): Likewise.
6401 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6402 (set_value_range_to_varying): No longer static.
6403 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6404 (infer_value_range, register_edge_assert_for): Likewise.
6406 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6408 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6409 generate the XXBRD instruction.
6411 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6412 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6413 as being compatible if -mabi=ieeelongdouble.
6414 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6415 to setup float128 built-ins with hardware support.
6416 (BU_FLOAT128_HW_2): Likewise.
6417 (BU_FLOAT128_HW_3): Likewise.
6418 (BU_FLOAT128_HW_VSX_1): Likewise.
6419 (BU_FLOAT128_HW_VSX_2): Likewise.
6420 (scalar_extract_expq): Change float128 built-in functions to
6421 accommodate having both KFmode and TFmode functions. Use the
6422 KFmode variant as the default.
6423 (scalar_extract_sigq): Likewise.
6424 (scalar_test_neg_qp): Likewise.
6425 (scalar_insert_exp_q): Likewise.
6426 (scalar_insert_exp_qp): Likewise.
6427 (scalar_test_data_class_qp): Likewise.
6428 (sqrtf128_round_to_odd): Delete processing the round to odd
6429 built-in functions as special built-in functions, and define them
6430 as float128 built-ins. Use the KFmode variant as the default.
6431 (truncf128_round_to_odd): Likewise.
6432 (addf128_round_to_odd): Likewise.
6433 (subf128_round_to_odd): Likewise.
6434 (mulf128_round_to_odd): Likewise.
6435 (divf128_round_to_odd): Likewise.
6436 (fmaf128_round_to_odd): Likewise.
6437 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6438 support for KFmode and TFmode xststdcqp calls.
6439 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6440 point, switch the built-in handlers for the get/set float128
6441 exponent, get float128 mantissa, float128 test built-ins, and the
6442 float128 round to odd built-in functions. Eliminate creating the
6443 float128 round to odd built-in functions as special built-ins.
6444 (rs6000_init_builtins): Eliminate special creation of the float128
6445 round to odd built-in functions.
6446 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6447 function insns to support both TFmode and KFmode variants.
6448 (xsxsigqp_<mode>): Likewise.
6449 (xsiexpqpf_<mode>): Likewise.
6450 (xsiexpqp_<mode>): Likewise.
6451 (xststdcqp_<mode>): Likewise.
6452 (xststdcnegqp_<mode>): Likewise.
6453 (xststdcqp_<mode>): Likewise.
6455 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6457 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6458 unused path_in_freq_ptr parameter.
6459 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6461 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6463 * ipa-inline.c (edge_badness): Dump sreal frequency.
6464 (compute_inlined_call_time): Match natural implementaiton ...
6465 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6466 forgotten division by CGRAPH_FREQ_BASE.
6468 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6470 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6471 Solaris 11. Update comment.
6472 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6474 * config/sol2.h (STARTFILE_SPEC): Likewise.
6475 * configure: Regenerate.
6477 2017-11-14 Carl Love <cel@us.ibm.com>
6479 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6480 le_ and be_ prefixes to swap* variables. Remove
6481 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6484 2017-11-14 Jason Merrill <jason@redhat.com>
6486 Support GTY((cache)) on hash_map.
6487 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6488 (ggc_cache_remove): Override it instead of ggc_mx.
6489 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6490 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6491 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6492 (simple_cache_map_traits): Override maybe_mx.
6493 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6494 (hash_map): Friend gt_cleare_cache.
6495 (gt_cleare_cache): New.
6496 * tree.h (tree_cache_traits): New hash_map traits class.
6497 (tree_cache_map): New typedef.
6499 2017-11-14 Richard Biener <rguenther@suse.de>
6501 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6502 paramter and handling.
6503 (cleanup_control_flow_bb): Likewise.
6504 (cleanup_control_flow_pre): New helper performing a DFS walk
6505 to call cleanup_control_flow_bb in PRE order.
6506 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6507 via cleanup_control_flow_pre.
6509 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6511 * config/aarch64/aarch64-simd.md
6512 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6513 (*aarch64_simd_bsl<mode>_alt): Likewise.
6514 (aarch64_simd_bsldi_internal): New.
6515 (aarch64_simd_bsldi_alt): Likewise.
6517 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6519 * tracer.c (better_p): Do not compare frequencies.
6520 * reg-stack.c (better_edge): Likewise.
6521 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6524 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6526 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6527 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6528 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6529 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6530 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6531 * predict.c (maybe_hot_frequency_p): Inline to ...
6532 (maybe_hot_count_p): ... here; rewrite to counts.
6533 (counts_to_freqs): Rename to ...
6534 (update_max_bb_count): ... this one.
6535 (expensive_function_p): Use counts.
6536 (estimate_bb_frequencies): Update.
6537 (rebuild_frequencies): Update.
6538 * predict.h (counts_to_freqs): Rename to ...
6539 (update_max_bb_count): ... this one.
6540 * profile.c (compute_branch_probabilities): Add debug info
6541 * tree-inline.c (expand_call_inline): Update debug info.
6542 (optimize_inline_calls): Use update_max_bb_count..
6543 (tree_function_versioning): Use update_max_bb_count..
6544 * value-prof.c (gimple_value_profile_transformations):
6545 Do not use update_max_bb_count.
6547 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6549 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6550 always use frequencies.
6552 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6554 * bb-reorder.c: Remove frequencies from comments.
6555 (better_edge_p): Use profile counts.
6556 (find_traces): Dump profile counts.
6557 (rotate_loop): Use profile counts.
6558 (find_traces_1_round): Likewise.
6559 (connect_better_edge_p): Use counts instead of probabilities for
6561 (copy_bb_p): Drop early check for non-0 frequency.
6562 (sanitize_hot_paths): Update comments.
6564 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6566 * ipa-split.c (struct split_point): Add count.
6567 (consider_split): Do not compute incoming frequency; compute incoming
6568 count and store it to split_point.
6569 (split_function): Set count of the call to split part correctly.
6571 2017-11-13 Carl Love <cel@us.ibm.com>
6573 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6575 2017-11-13 Tom Tromey <tom@tromey.com>
6577 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6579 2017-11-13 Carl Love <cel@us.ibm.com>
6581 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6582 Add support for builtins:
6583 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6584 vector {un,}signed {char,int,short},
6585 vector {un,}signed {char,int,short}) arguments.
6586 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6587 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6588 Add BU_P9V_AV_2 expansions for the builtins.
6589 * config/rs6000/altivec.h (vec_first_match_index,
6590 vec_first_mismatch_index, vec_first_match_or_eos_index,
6591 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6592 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6593 new extern declaration.
6594 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6595 * config/rs6000/vsx.md (first_match_index_<mode>,
6596 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6597 first_mismatch_or_eos_index_<mode>): Add define expand.
6598 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6599 * doc/extend.texi: Update the built-in documenation file for the new
6602 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6604 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6605 into the min/max operations for _Float<N> and _Float<N>X types.
6607 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6610 * dwarf2out.c (do_eh_frame): New static variable.
6611 (dwarf2out_begin_prologue): Set it.
6612 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6614 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6616 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6618 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6620 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6623 * debug.h (dwarf2out_do_eh_frame): Declare.
6624 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6625 (dwarf2out_do_frame): Use it.
6626 (dwarf2out_do_cfi_asm): Likewise.
6627 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6628 (dwarf2out_assembly_start): Likewise.
6629 (dwarf2out_begin_prologue): Fix comment.
6630 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6631 if the target needs either debug or unwind DWARF2 info.
6632 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6633 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6634 -ftrapping-math, -ftrapv and -fwrapv.
6636 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6638 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6639 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6640 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6641 (estimate_edge_size_and_time): Likewise.
6642 (ipa_merge_fn_summary_after_inlining): Likewise.
6643 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6644 (compute_uninlined_call_time): Use sreal_frequency.
6645 (compute_inlined_call_time): Likewise.
6646 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6647 * profile-count.c: Include sreal.h.
6648 (profile_count::to_sreal_scale): New.
6649 * profile-count.h: Forward declare sreal.
6650 (profile_count::to_sreal_scale): Declare.
6652 2017-11-13 Nathan Sidwell <nathan@acm.org>
6654 * diagnostic.c (maybe_line_and_column): New.
6655 (diagnostic_get_location_text): Use it.
6656 (diagnostic_report_current_module): Likewise.
6657 (test_diagnostic_get_location_text): Add tests.
6659 2017-11-13 Luis Machado <luis.machado@linaro.org>
6661 * doc/md.texi (Specifying processor pipeline description): Fix
6662 incorrect latency for the div instruction example.
6664 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6666 PR tree-optimization/78821
6667 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6668 that bit_not_p is the same.
6669 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6670 (split_group): Count precisely bit_not_p bits in each statement.
6671 (invert_op): New function.
6672 (imm_store_chain_info::output_merged_store): Use invert_op to
6673 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6674 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6676 2017-11-13 Martin Liska <mliska@suse.cz>
6678 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6679 (struct source_info): Likewise.
6680 (add_branch_counts): Likewise.
6681 (add_line_counts): Likewise.
6682 (function_summary): Likewise.
6683 (output_intermediate_line): Likewise.
6684 (generate_results): Likewise.
6686 2017-11-13 Martin Liska <mliska@suse.cz>
6688 * gcov.c (struct block_info): Remove typedef for block_t.
6689 (struct line_info): Likewise.
6690 (line_info::has_block): Likewise.
6691 (EXIT_BLOCK): Likewise.
6692 (unblock): Likewise.
6693 (circuit): Likewise.
6694 (get_cycles_count): Likewise.
6695 (process_file): Likewise.
6696 (read_graph_file): Likewise.
6697 (solve_flow_graph): Likewise.
6698 (find_exception_blocks): Likewise.
6699 (add_line_counts): Likewise.
6700 (accumulate_line_info): Likewise.
6701 (output_line_details): Likewise.
6703 2017-11-13 Martin Liska <mliska@suse.cz>
6705 * gcov.c (struct arc_info): Remove typedef for arc_t.
6706 (struct line_info): Likewise.
6707 (add_branch_counts): Likewise.
6708 (output_branch_count): Likewise.
6709 (function_info::~function_info): Likewise.
6710 (circuit): Likewise.
6711 (output_intermediate_line): Likewise.
6712 (read_graph_file): Likewise.
6713 (solve_flow_graph): Likewise.
6714 (find_exception_blocks): Likewise.
6715 (add_line_counts): Likewise.
6716 (accumulate_line_info): Likewise.
6717 (output_line_details): Likewise.
6718 (output_function_details): Likewise.
6720 2017-11-13 Martin Liska <mliska@suse.cz>
6722 * gcov.c (struct function_info): Remove typedef for function_t.
6723 (struct source_info): Likewise.
6724 (source_info::get_functions_at_location): Likewise.
6725 (solve_flow_graph): Likewise.
6726 (find_exception_blocks): Likewise.
6727 (add_line_counts): Likewise.
6728 (output_intermediate_file): Likewise.
6729 (process_file): Likewise.
6730 (generate_results): Likewise.
6731 (release_structures): Likewise.
6732 (read_graph_file): Likewise.
6733 (read_count_file): Likewise.
6734 (accumulate_line_counts): Likewise.
6735 (output_lines): Likewise.
6737 2017-11-13 Martin Liska <mliska@suse.cz>
6739 * gcov.c (function_info::function_info): Remove num_counts
6740 and add vector<gcov_type>.
6741 (function_info::~function_info): Use the vector.
6742 (process_file): Likewise.
6743 (read_graph_file): Likewise.
6744 (read_count_file): Likewise.
6745 (solve_flow_graph): Likewise.
6747 2017-11-13 Martin Liska <mliska@suse.cz>
6749 * gcov.c (function_info::is_artificial): New function.
6750 (process_file): Erase all artificial early.
6751 (generate_results): Skip as all artificial are already
6754 2017-11-13 Martin Liska <mliska@suse.cz>
6756 * gcov.c (read_graph_file): Store to global vector of functions.
6757 (read_count_file): Iterate the vector.
6758 (process_file): Likewise.
6759 (generate_results): Likewise.
6760 (release_structures): Likewise.
6762 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6764 PR tree-optimization/82954
6765 * gimple-ssa-store-merging.c
6766 (imm_store_chain_info::coalesce_immediate_stores): If
6767 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6769 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6771 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6772 Upddate call to ENDIAN_LANE_N.
6773 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6774 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6775 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6776 and use aarch64_endian_lane_rtx.
6777 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6779 2017-11-12 Tom de Vries <tom@codesourcery.com>
6781 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6783 2017-11-12 Tom de Vries <tom@codesourcery.com>
6785 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6787 2017-11-12 Tom de Vries <tom@codesourcery.com>
6789 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6791 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6792 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6793 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6795 2017-11-11 Martin Sebor <msebor@redhat.com>
6798 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6801 * prefic.c (translate_name): Replace strncpy with memcpy to
6802 avoid -Wstringop-truncation.
6804 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6806 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6808 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6810 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6812 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6814 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6817 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6819 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6820 enable generating XXBRH if the value is in a vector register.
6821 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6822 value is in a vector register.
6823 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6824 register to register bswap64's instead of doing the GPR sequence
6825 used on previous machines.
6826 (bswapdi2_xxbrd): New insn.
6827 (bswapdi2_reg): Disallow on ISA 3.0.
6828 (register to register bswap64 splitter): Do not split the insn on
6829 ISA 3.0 systems that use XXBRD.
6831 2017-11-10 Martin Sebor <msebor@redhat.com>
6834 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6836 (find_subframework_file): Same.
6838 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6840 * auto-profile.c (afdo_indirect_call): Drop frequency.
6841 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6842 (cgraph_node::create_edge): Drop frequency argument.
6843 (cgraph_node::create_indirect_edge): Drop frequency argument.
6844 (cgraph_edge::make_speculative): Drop frequency arguments.
6845 (cgraph_edge::resolve_speculation): Do not update frequencies
6846 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6847 (cgraph_node::dump): Check consistency in IPA mode.
6848 (cgraph_edge::maybe_hot_p): Use IPA counter.
6849 (cgraph_edge::verify_count_and_frequency): Rename to ...
6850 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6851 (cgraph_node::verify_node): Update.
6852 * cgraph.h (struct cgraph_edge): Drop frequency.
6853 (cgraph_edge::frequency): New function.
6854 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6856 (cgraph_edge::rebuild_edges): Likewise.
6857 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6858 (duplicate_thunk_for_node): Do not pass frequency.
6859 (cgraph_node::create_clone): Scale only counts.
6860 (cgraph_node::create_virtual_clone): Do not pass frequency.
6861 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6862 (cgraph_node::create_version_clone): Do not pass frequency.
6863 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6864 (cgraph_node::expand_thunk): Do not pass frequency.
6865 (cgraph_node::create_wrapper): Do not pass frequency.
6866 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6868 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6869 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6870 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6871 (gather_caller_stats): Use frequency function.
6872 (ipcp_cloning_candidate_p): Use frequency function.
6873 (ipcp_propagate_stage): Use frequency function.
6874 (get_info_about_necessary_edges): Use frequency function.
6875 (update_profiling_info): Update only IPA profile.
6876 (update_specialized_profile): Use frequency functoin.
6877 (perhaps_add_new_callers): Update only IPA profile.
6878 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6879 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6880 (dump_ipa_call_summary): Use frequency function.
6881 (estimate_edge_size_and_time): Use frequency function.
6882 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6883 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6884 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6885 (update_noncloned_counts): ... ths one; scale counts only.
6886 (clone_inlined_nodes): Do not scale frequency.
6887 (inline_call): Do not pass frequency.
6888 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6889 (compute_inlined_call_time): Use IPA profile.
6890 (want_inline_small_function_p): Use IPA profile.
6891 (want_inline_self_recursive_call_p): Use IPA profile.
6892 (edge_badness): Use IPA profile.
6893 (lookup_recursive_calls): Use IPA profile.
6894 (recursive_inlining): Do not pass frequency.
6895 (resolve_noninline_speculation): Do not update frequency.
6896 (inline_small_functions): Collect max of IPA profile.
6897 (dump_overall_stats): Dump IPA porfile.
6898 (dump_inline_stats): Dump IPA porfile.
6899 (ipa_inline): Collect IPA stats.
6900 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6901 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6902 (ipa_propagate_frequency): Use frequency function.
6903 (ipa_profile): Cleanup.
6904 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6905 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6906 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6907 (input_node): Do not stream frequency.
6908 (input_edge): Do not stream frequency.
6909 (merge_profile_summaries): Scale only IPA profiles.
6910 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6911 * predict.c (drop_profile): Do not recompute frequency.
6912 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6913 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6914 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6915 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6916 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6917 * tree-inline.c (copy_bb): Do not scale frequency.
6918 (expand_call_inline): Do not scale frequency.
6919 (tree_function_versioning): Do not scale frequency.
6920 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6922 2017-11-10 Julia Koval <julia.koval@intel.com>
6924 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6925 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6926 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6927 (_mm256_maskz_gf2p8affine_epi64_epi8)
6928 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6929 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6930 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6931 (__builtin_ia32_vgf2p8affineqb_v32qi)
6932 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6933 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6935 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6939 (arm_option_override): Refactor.
6940 (arm_option_reconfigure_globals): New.
6941 (arm_options_perform_arch_sanity_checks): New.
6942 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6944 (arm_options_perform_arch_sanity_checks): Likewise
6946 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6948 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6949 (power9-qpmul): New.
6950 * rs6000/rs6000.md ("type" attr): Add qmul.
6951 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6952 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6953 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6955 2017-11-10 Martin Sebor <msebor@redhat.com>
6958 * builtins.c (compute_objsize): Handle arrays that
6959 compute_builtin_object_size likes to fail for. Make extern.
6960 * builtins.h (compute_objsize): Declare.
6961 (check_strncpy_sizes): New function.
6962 (expand_builtin_strncpy): Call check_strncpy_sizes.
6963 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6964 -Wstringop-truncation.
6965 (gimple_fold_builtin_strncat): Same.
6966 * gimple.c (gimple_build_call_from_tree): Set call location.
6967 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6968 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6969 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6970 (handle_builtin_strlen): Use strlen_to_stridx.
6971 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6973 Use strlen_to_stridx.
6974 (pass_strlen::execute): Release strlen_to_stridx.
6975 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6976 (-Wstringop-truncation): Document new option.
6978 2017-11-10 Martin Liska <mliska@suse.cz>
6980 PR gcov-profile/82702
6981 * gcov.c (main): Handle intermediate files in a different
6983 (get_gcov_intermediate_filename): New function.
6984 (output_gcov_file): Remove support of intermediate files.
6985 (generate_results): Allocate intermediate file.
6986 (release_structures): Clean-up properly fn_end.
6987 (output_intermediate_file): Start iterating with line 1.
6989 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6991 PR tree-optimization/82929
6992 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6993 ops_swapped_p non-static data member.
6994 (store_immediate_info::store_immediate_info): Clear it.
6995 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6996 ops set ops_swapped_p.
6997 (count_multiple_uses): Handle ops_swapped_p.
6999 2017-11-10 Martin Liska <mliska@suse.cz>
7001 * coverage.c (coverage_init): Stream information about
7002 support of has_unexecuted_blocks.
7003 * doc/gcov.texi: Document that.
7004 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7005 * gcov.c (read_graph_file): Likewise.
7006 (output_line_beginning): Fix a small issue with
7009 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7011 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7012 reference of trivial component.
7014 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7017 * gimple-ssa-store-merging.c
7018 (pass_store_merging::terminate_all_aliasing_chains): For
7019 gimple_store_p stmts also call refs_output_dependent_p.
7021 PR rtl-optimization/82913
7022 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7025 2017-11-09 Jeff Law <law@redhat.com>
7027 * vr-values.h: New file with vr_values class.
7028 * tree-vrp.c: Include vr-values.h
7029 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7030 data objects into the vr_values class.
7031 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7032 (get_value_range): Make it a member function within vr_values class.
7033 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7034 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7035 (op_with_constant_singleton_value_range): Likewise.
7036 (extract_range_for_var_from_comparison_expr): Likewise.
7037 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7038 (extract_range_from_binary_expr): Likewise.
7039 (extract_range_from_unary_expr): Likewise.
7040 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7041 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7042 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7043 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7044 (compare_name_with_value, compare_names): Likewise.
7045 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7046 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7047 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7048 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7049 (extract_range_from_phi_node): Likewise.
7050 (simplify_truth_ops_using_ranges): Likewise.
7051 (simplify_div_or_mod_using_ranges): Likewise.
7052 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7053 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7054 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7055 (simplify_float_conversion_using_ranges): Likewise.
7056 (simplify_internal_call_using_ranges): Likewise.
7057 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7058 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7059 poitner via x_vr_values for calls into gimple folder.
7060 (vrp_initialize_lattice): Make this the vr_values ctor.
7061 (vrp_free_lattice): Make this the vr_values dtor.
7062 (set_vr_value): New function.
7063 (class vrp_prop): Add vr_values data member. Add various member
7064 functions as well as member functions that delegate to vr_values.
7065 (check_array_ref): Make a member function within vrp_prop class.
7066 (search_for_addr_array, vrp_initialize): Likewise.
7067 (vrp_finalize): Likewise. Revamp to avoid direct access to
7068 vr_value, values_propagated, etc.
7069 (check_array_bounds): Extract vrp_prop class instance pointer from
7070 walk info structure. Use it to call member functions.
7071 (check_all_array_refs): Make a member function within vrp_prop class.
7072 Smuggle class instance pointer via walk info structure.
7073 (x_vr_values): New local static.
7074 (vrp_valueize): Use x_vr_values to get class instance.
7075 (vr_valueize_1): Likewise.
7076 (class vrp_folder): Add vr_values data member. Add various member
7077 functions as well as member functions that delegate to vr_values.
7078 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7079 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7080 class instance from vr_values. Use it to call member functions.
7081 (vrp_dom_walker): Add vr_values data member.
7082 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7083 instance via x_vr_values.
7084 (identify_jump_threads): Accept vr_values as argument. Store
7085 it into the walker structure.
7086 (evrp_dom_walker): Add vr_values class data member. Add various
7088 (evrp_dom_walker::try_find_new_range): Use vr_values data
7089 member to access the memory allocator.
7090 (evrp_dom_walker::before_dom_children): Store vr_values class
7091 instance into the vrp_folder class.
7092 (evrp_dom_walker::push_value_range): Rework to avoid direct
7093 access to num_vr_values and vr_value.
7094 (evrp_dom_walker::pop_value_range): Likewise.
7095 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7096 Use vr_values to get to dump_all_value_ranges member function.
7097 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7098 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7099 Pass vr_values class instance down to identify_jump_threads.
7100 Remove call to vrp_free_lattice.
7101 (debug_all_value_ranges): Remove.
7103 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7104 (vrp_folder): Likewise.
7106 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7107 Get it from the existing bitmap instead.
7108 (vrp_intersect_ranges_1): Likewise.
7110 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7112 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7114 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7115 set bit_not_p to it.
7116 (imm_store_chain_info::coalesce_immediate_stores): Break group
7117 if bit_not_p is different.
7118 (count_multiple_uses, split_group,
7119 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7120 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7121 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7122 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7123 to store_immediate_info ctor.
7125 2017-11-09 Jim Wilson <jimw@sifive.com>
7127 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7128 (scan_prog_file): Likewise.
7130 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7132 * bb-reorder.c (max_entry_frequency): Remove.
7133 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7134 connect_traces, push_to_next_round_p): Remove prototypes.
7135 (find_traces_1_round): Use counts only.
7136 (push_to_next_round_p): Likewise.
7137 (find_traces): Likewise.
7138 (rotate_loop): Likewise.
7139 (find_traces_1_round): Likewise.
7140 (connect_traces): Likewise.
7141 (edge_order): Likewise.
7143 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7145 * config/arm/arm.c (output_return_instruction): Add comments to
7146 indicate requirement for cmse_nonsecure_entry return to account
7147 for the size of clearing instruction output here.
7148 (thumb_exit): Likewise.
7149 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7150 return in hardfloat mode.
7152 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7154 * config/rs6000/rs6000.c (machine_function): Add a bool,
7155 "toc_is_wrapped_separately".
7156 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7157 if it wasn't explicitly set or unset, we are optimizing for speed, and
7158 doing separate shrink-wrapping.
7159 (rs6000_get_separate_components): Enable the TOC component if
7160 saving the TOC register in the prologue.
7161 (rs6000_components_for_bb): Handle the TOC component.
7162 (rs6000_emit_prologue_components): Store the TOC register where needed.
7163 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7164 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7166 2017-11-09 Martin Jambor <mjambor@suse.cz>
7168 * ipa-param-manipulation.c: New file.
7169 * ipa-param-manipulation.h: Likewise.
7170 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7171 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7172 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7173 (ipa_parm_adjustment): Likewise.
7174 (ipa_parm_adjustment_vec): Likewise.
7175 (ipa_get_vector_of_formal_parms): Moved declaration to
7176 ipa-param-manipulation.h.
7177 (ipa_get_vector_of_formal_parm_types): Likewise.
7178 (ipa_modify_formal_parameters): Likewise.
7179 (ipa_modify_call_arguments): Likewise.
7180 (ipa_combine_adjustments): Likewise.
7181 (ipa_dump_param_adjustments): Likewise.
7182 (ipa_modify_expr): Likewise.
7183 (ipa_get_adjustment_candidate): Likewise.
7184 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7185 ipa-param-manipulation.c.
7186 (ipa_get_vector_of_formal_parm_types): Likewise.
7187 (ipa_modify_formal_parameters): Likewise.
7188 (ipa_modify_call_arguments): Likewise.
7189 (ipa_modify_expr): Likewise.
7190 (get_ssa_base_param): Likewise.
7191 (ipa_get_adjustment_candidate): Likewise.
7192 (index_in_adjustments_multiple_times_p): Likewise.
7193 (ipa_combine_adjustments): Likewise.
7194 (ipa_dump_param_adjustments): Likewise.
7195 * tree-sra.c: Also include ipa-param-manipulation.h
7196 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7199 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7200 Alan Hayward <alan.hayward@arm.com>
7201 David Sherwood <david.sherwood@arm.com>
7203 * doc/sourcebuild.texi (vect_masked_store): Document.
7205 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7206 Alan Hayward <alan.hayward@arm.com>
7207 David Sherwood <david.sherwood@arm.com>
7209 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7211 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7212 Alan Hayward <alan.hayward@arm.com>
7213 David Sherwood <david.sherwood@arm.com>
7215 * doc/sourcebuild.texi (vect_variable_length): Document.
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 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7223 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7224 Alan Hayward <alan.hayward@arm.com>
7225 David Sherwood <david.sherwood@arm.com>
7227 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7229 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7230 Alan Hayward <alan.hayward@arm.com>
7231 David Sherwood <david.sherwood@arm.com>
7233 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7234 previously undocumented selectors.
7235 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7237 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7239 * doc/rtl.texi (const_vector): Say that elements can be
7240 const_wide_ints too.
7241 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7242 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7243 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7244 * optabs.c (expand_vector_broadcast): Likewise.
7246 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7247 Alan Hayward <alan.hayward@arm.com>
7248 David Sherwood <david.sherwood@arm.com>
7250 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7251 scaled index even if the unscaled address was invalid.
7252 Don't increase the complexity of using a scale in that case.
7254 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7255 Alan Hayward <alan.hayward@arm.com>
7256 David Sherwood <david.sherwood@arm.com>
7258 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7259 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7260 UNITS_PER_WORD bytes.
7261 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7262 based on REGMODE_NATURAL_SIZE of the inner mode.
7263 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7264 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7265 * expmed.c (lowpart_bit_field_p): Divide the value up into
7266 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7267 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7268 whether something is likely to occupy more than one register.
7270 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7273 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7274 profile_count::adjust_for_ipa_scaling.
7275 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7276 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7278 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7280 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7282 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7283 (split_group): Add total_orig and total_new arguments, estimate the
7284 number of statements related to the store group without store merging
7285 and with store merging.
7286 (imm_store_chain_info::output_merged_store): Adjust split_group
7287 callers, punt if estimated number of statements with store merging
7288 is not smaller than estimated number of statements without it.
7290 (handled_load): Remove has_single_use checks.
7291 (pass_store_merging::process_store): Likewise.
7293 2017-11-09 Richard Biener <rguenther@suse.de>
7295 PR tree-optimization/82902
7296 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7298 2017-11-09 Martin Liska <mliska@suse.cz>
7301 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7304 2017-11-09 Martin Liska <mliska@suse.cz>
7306 PR tree-optimization/82669
7307 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7309 2017-11-09 Martin Liska <mliska@suse.cz>
7311 PR gcov-profile/48463
7312 * coverage.c (coverage_begin_function): Output also end locus
7313 of a function and information whether the function is
7315 * gcov-dump.c (tag_function): Parse and print the information.
7316 * gcov.c (INCLUDE_MAP): Add include.
7317 (INCLUDE_SET): Likewise.
7318 (struct line_info): Move earlier in the source file because
7319 of vector<line_info> in function_info structure.
7320 (line_info::line_info): Likewise.
7321 (line_info::has_block): Likewise.
7322 (struct source_info): Add new member index.
7323 (source_info::get_functions_at_location): New function.
7324 (function_info::group_line_p): New function.
7325 (output_intermediate_line): New function.
7326 (output_intermediate_file): Use the mentioned function.
7327 (struct function_start): New.
7328 (struct function_start_pair_hash): Likewise.
7329 (process_file): Add code that identifies group functions.
7330 Assign lines either to global or function scope.
7331 (generate_results): Skip artificial functions.
7332 (find_source): Assign index for each source file.
7333 (read_graph_file): Read new flag artificial and end_line.
7334 (add_line_counts): Assign it either to global of function scope.
7335 (accumulate_line_counts): Isolate core of the function to
7336 accumulate_line_info and call it for both function and global
7338 (accumulate_line_info): New function.
7339 (output_line_beginning): Fix GNU coding style.
7340 (print_source_line): New function.
7341 (output_line_details): Likewise.
7342 (output_function_details): Likewise.
7343 (output_lines): Iterate both source (global) scope and function
7345 (struct function_line_start_cmp): New class.
7346 * doc/gcov.texi: Reflect changes in documentation.
7348 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7351 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7352 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7353 and similarly for not instead of neg.
7355 2017-11-08 Andi Kleen <ak@linux.intel.com>
7357 * config/i386/i386.opt: Add -mforce-indirect-call.
7358 * config/i386/predicates.md: Check for flag_force_indirect_call.
7359 * doc/invoke.texi: Document -mforce-indirect-call
7361 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7363 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7365 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7366 config/riscv/riscv.c (predict.h): New include.
7367 (riscv_slow_unaligned_access_p): No longer static.
7368 (riscv_block_move_straight): Add require.
7369 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7371 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7374 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7375 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7376 nonimmediate_operand predicate for operand 1 instead of
7379 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7381 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7383 * config/aarch64/constraints.md (Uml): New constraint.
7384 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7387 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7389 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7390 of two vec_duplicates into a vec_concat.
7392 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7394 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7395 Simplify vec_merge of vec_duplicate and vec_concat.
7396 * config/aarch64/constraints.md (Utq): New constraint.
7397 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7400 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7402 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7403 Simplify vec_merge of vec_duplicate and const_vector.
7404 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7406 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7407 mode iterator. Update predicate on operand 1 to
7408 handle non-const_vec constants. Delete constraints.
7409 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7411 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7413 PR tree-optimization/78821
7414 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7416 (store_operand_info::store_operand_info): Initialize it to false.
7417 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7418 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7419 store in the group, and if chain_info is non-NULL, to ignore altogether
7421 (compatible_load_p): Fail if bit_not_p does not match.
7422 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7423 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7424 (pass_store_merging::process_store): Adjust
7425 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7426 call terminate_all_aliasing_chains newly when adding a store into
7427 a chain with non-NULL chain_info.
7429 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7431 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7433 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7435 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7437 (aarch64_layout_frame): Initialise emit_frame_chain.
7438 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7439 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7441 2017-11-08 Martin Liska <mliska@suse.cz>
7443 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7444 of gimple_call_internal_p.
7446 2017-11-07 Tom de Vries <tom@codesourcery.com>
7448 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7450 2017-11-07 Tom de Vries <tom@codesourcery.com>
7452 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7455 2017-11-08 Martin Liska <mliska@suse.cz>
7458 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7460 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7462 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7464 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7467 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7468 (*cmp<mode>_ccz_1): New insn with $k alternative.
7471 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7472 RTX_COMM_COMPARE as commutative as well.
7473 (ix86_binary_operator_ok): Formatting fix.
7474 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7475 *<code><mode>3<mask_name><round_saeonly_name>,
7476 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7477 *<s>mul<mode>3_highpart<mask_name>,
7478 *vec_widen_umult_even_v16si<mask_name>,
7479 *vec_widen_umult_even_v8si<mask_name>,
7480 *vec_widen_umult_even_v4si<mask_name>,
7481 *vec_widen_smult_even_v16si<mask_name>,
7482 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7483 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7484 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7485 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7486 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7487 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7488 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7489 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7490 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7491 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7492 ix86_binary_operator_ok. Formatting fixes.
7493 (*<plusminus_insn><mode>3<mask_name><round_name>,
7494 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7497 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7499 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7500 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7501 eq and ne if TARGET_ISEL.
7502 (cmp): New code_iterator.
7503 (UNS, UNSU_, UNSIK): New code_attrs.
7504 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7505 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7507 ("eq<mode>3"): ... this.
7508 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7510 ("ne<mode>3"): ... this.
7512 2017-11-07 Julia Koval <julia.koval@intel.com>
7515 * common/config/i386/i386-common.c
7516 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7517 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7518 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7519 * config/i386/i386.opt: Ditto.
7520 * config/i386/i386.c (ix86_target_string): Ditto.
7521 (ix86_option_override_internal): Ditto.
7522 (ix86_init_mpx_builtins): Move MPX to args2.
7523 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7524 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7525 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7526 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7527 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7528 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7529 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7531 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7534 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7535 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7536 (zero-extendsidi peephole2): Remove peephole.
7538 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7541 * stor-layout.c: Include attribs.h.
7542 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7543 explicit lookup of "aligned" attribute.
7545 2017-11-07 Andrew Waterman <andrew@sifive.com>
7547 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7548 (riscv_expand_block_move): Likewise.
7549 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7551 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7552 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7553 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7554 (riscv_adjust_block_mem): Likewise.
7555 (riscv_block_move_loop): Likewise.
7556 (riscv_expand_block_move): Likewise.
7557 * config/riscv/riscv.md (movmemsi): New pattern.
7559 2017-11-07 Michael Clark <michaeljclark@mac.com>
7561 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7562 (MUSL_DYNAMIC_LINKER): Likewise.
7564 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7566 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7567 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7570 2017-11-07 Richard Biener <rguenther@suse.de>
7572 * match.pd: Fix build.
7574 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7575 Jackson Woodruff <jackson.woodruff@arm.com>
7577 PR tree-optimization/71026
7578 * match.pd: Canonicalize negate in division.
7580 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7583 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7585 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7587 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7588 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7590 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7592 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7593 non-scalar integral types.
7594 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7595 (-(A-B), -(~A)): New transformations.
7597 2017-11-07 Tom de Vries <tom@codesourcery.com>
7599 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7600 semicolon after "do {} while (0)".
7601 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7602 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7603 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7604 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7605 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7607 2017-11-07 Tom de Vries <tom@codesourcery.com>
7609 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7610 after "do {} while (0)".
7611 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7612 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7613 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7614 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7615 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7617 2017-11-07 Tom de Vries <tom@codesourcery.com>
7620 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7622 (arm_rtx_costs_internal): Add missing semicolon after
7623 HANDLE_NARROW_SHIFT_ARITH call.
7625 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7627 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7628 disable isel if it was not set explicitly.
7630 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7632 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7633 (add_type_attribute) likewise.
7635 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7637 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7638 of struct ix86_frame.
7639 (ix86_initial_elimination_offset): Likewise.
7640 (ix86_expand_split_stack_prologue): Likewise.
7642 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7644 * tree-vrp.h (enum value_range_type): Update stale comment.
7646 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7647 Alan Hayward <alan.hayward@arm.com>
7648 David Sherwood <david.sherwood@arm.com>
7650 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7651 (aarch64_expand_vec_perm_const): Take the number of units too.
7652 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7653 (aarch64_expand_vec_perm_const): Likewise.
7654 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7655 (vec_perm<mode>): Update accordingly.
7657 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7658 Alan Hayward <alan.hayward@arm.com>
7659 David Sherwood <david.sherwood@arm.com>
7661 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7662 Take the number of units too.
7663 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7664 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7665 but check for a vector mode before rather than after the call.
7666 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7667 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7668 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7669 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7670 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7671 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7672 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7673 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7674 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7675 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7676 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7677 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7678 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7679 (aarch64_sqdmull2_n<mode>): Update accordingly.
7681 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7682 Alan Hayward <alan.hayward@arm.com>
7683 David Sherwood <david.sherwood@arm.com>
7685 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7686 the number of units too.
7687 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7688 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7689 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7690 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7691 (vec_store_lanesxi<mode>): Update accordingly.
7693 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7694 Alan Hayward <alan.hayward@arm.com>
7695 David Sherwood <david.sherwood@arm.com>
7697 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7698 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7699 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7700 of units rather than the mode.
7701 * config/aarch64/iterators.md (nunits): New mode attribute.
7702 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7703 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7704 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7705 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7706 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7707 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7708 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7709 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7710 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7711 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7712 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7713 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7714 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7715 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7716 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7717 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7718 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7719 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7720 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7721 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7722 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7723 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7724 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7725 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7726 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7727 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7728 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7729 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7730 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7731 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7732 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7733 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7734 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7735 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7736 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7737 (aarch64_simd_vec_setv2di): Likewise.
7739 2017-11-06 Carl Love <cel@us.ibm.com>
7741 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7743 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7744 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7746 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7748 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7749 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7750 Add power 8 macro expansions.
7751 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7752 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7753 power 8 instructions. (VSX_XXBR): Add iterator.
7755 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7757 * config/arm/arm.md (predicable_short_it): Change default to "no",
7758 improve documentation, remove uses that are identical to the default.
7759 (enabled_for_depr_it): Rename to enabled_for_short_it.
7760 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7761 Remove default uses.
7762 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7763 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7764 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7765 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7767 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7770 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7771 float128 helper macros, which are no longer used after deleting
7772 the old 'q' built-in functions, and moving the round to odd
7773 built-in functions to being special built-in functions.
7774 (BU_FLOAT128_2): Likewise.
7775 (BU_FLOAT128_1_HW): Likewise.
7776 (BU_FLOAT128_2_HW): Likewise.
7777 (BU_FLOAT128_3_HW): Likewise.
7778 (FABSQ): Delete old 'q' built-in functions.
7779 (COPYSIGNQ): Likewise.
7780 (SQRTF128_ODD): Move round to odd built-in functions to be
7781 special built-in functions, so that we can handle
7782 -mabi=ieeelongdouble.
7783 (TRUNCF128_ODD): Likewise.
7784 (ADDF128_ODD): Likewise.
7785 (SUBF128_ODD): Likewise.
7786 (MULF128_ODD): Likewise.
7787 (DIVF128_ODD): Likewise.
7788 (FMAF128_ODD): Likewise.
7789 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7790 built-in names to 'f128'.
7791 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7792 old 'q' built-in functions, as the machine independent code for
7793 'f128' built-in functions handles this.
7794 (rs6000_expand_builtin): Add expansion for float128 round to odd
7795 functions, keying off on -mabi=ieeelongdouble of whether to use
7796 the KFmode or TFmode variant.
7797 (rs6000_init_builtins): Initialize the _Float128 round to odd
7799 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7800 _Float128 'q' built-in functions are now mapped into the new
7801 'f128' built-in functions.
7803 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7805 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7806 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7808 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7811 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7812 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7813 * toplev.c: Include "ipa-fnsummary.h".
7814 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7816 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7818 PR tree-optimization/82838
7819 * gimple-ssa-store-merging.c
7820 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7821 on a separate gimple_seq which is then appended to seq.
7823 2017-11-06 Jeff Law <law@redhat.com>
7826 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7827 (get_probe_interval): New functions.
7828 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7829 (ix86_adjust_stack_and_probe): Likewise.
7830 (output_adjust_stack_and_probe): Likewise.
7831 (ix86_emit_probe_stack_range): Likewise.
7832 (ix86_expand_prologue): Likewise.
7834 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7836 PR tree-optimization/82816
7837 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7838 if the modes of the two types are the same.
7839 (convert_plusminus_to_widen): Likewise.
7841 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7843 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7845 (usadv16qi): New define_expand.
7846 (usadv8hi): New define_expand.
7848 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7851 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7853 (inline_transform): Likewise.
7854 * predict.c (counts_to_freqs): Remove useless conditional.
7855 * profile-count.h (profile_count::apply_scale): Move sanity check.
7856 * tree-inline.c (copy_bb): Always scale.
7857 (copy_cfg_body): Likewise.
7859 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7862 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7864 (*cmp_ite0): Add enabled_for_depr_it attribute.
7865 (*cmp_ite1): Likewise.
7867 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7869 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7872 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7874 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7875 return the value by pointer.
7876 (extract_range_from_multiplicative_op_1): Update accordingly.
7877 Return as soon as an operation fails.
7879 2017-11-05 Tom de Vries <tom@codesourcery.com>
7882 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7883 (DEF_SANITIZER_BUILTIN): ... here.
7884 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7885 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7887 2017-11-05 Tom de Vries <tom@codesourcery.com>
7890 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7892 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7893 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7894 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7896 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7897 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7898 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7900 2017-11-05 Tom de Vries <tom@codesourcery.com>
7903 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7906 2017-11-04 Michael Clark <michaeljclark@mac.com>
7908 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7909 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7911 (*addsi3_extended): Likewise.
7912 (*addsi3_extended2): Likewise.
7913 (<optab>si3): Likewise.
7914 (<optab>di3): Likewise.
7915 (<optab><mode>3): Likewise.
7916 (<*optabe>si3_internal): Likewise.
7917 (zero_extendqi<SUPERQI:mode>2): Likewise.
7918 (*add<mode>hi3): Likewise.
7919 (*xor<mode>hi3): Likewise.
7920 (<optab>di3): Likewise.
7921 (*<optab>si3_extend): Likewise.
7922 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7923 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7924 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7926 2017-11-04 Andrew Waterman <andrew@sifive.com>
7928 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7929 TARGET_STRICT_ALIGN based upon -mtune argument.
7931 2017-11-04 Andrew Waterman <andrew@sifive.com>
7933 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7935 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7937 * config/i386/i386.c (choose_basereg): Use optional scratch
7938 register and add assertion.
7939 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7940 needed, and don't allocate stack.
7941 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7942 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7943 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7945 2017-11-03 Jeff Law <law@redhat.com>
7947 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7948 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7949 to probe at the start of a noreturn function.
7951 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7953 PR tree-optimization/78821
7954 * gimple-ssa-store-merging.c: Update the file comment.
7955 (MAX_STORE_ALIAS_CHECKS): Define.
7956 (struct store_operand_info): New type.
7957 (store_operand_info::store_operand_info): New constructor.
7958 (struct store_immediate_info): Add rhs_code and ops data members.
7959 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7960 and op1r arguments to the ctor, initialize corresponding data members.
7961 (struct merged_store_group): Add load_align_base and load_align
7963 (merged_store_group::merged_store_group): Initialize them.
7964 (merged_store_group::do_merge): Update them.
7965 (merged_store_group::apply_stores): Pick the constant for
7966 encode_tree_to_bitpos from one of the two operands, or skip
7967 encode_tree_to_bitpos if neither operand is a constant.
7968 (class pass_store_merging): Add process_store method decl. Remove
7969 bool argument from terminate_all_aliasing_chains method decl.
7970 (pass_store_merging::terminate_all_aliasing_chains): Remove
7971 var_offset_p argument and corresponding handling.
7972 (stmts_may_clobber_ref_p): New function.
7973 (compatible_load_p): New function.
7974 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7975 if there is overlap and rhs_code is not INTEGER_CST. For
7976 non-overlapping stores terminate group if rhs is not mergeable.
7977 (get_alias_type_for_stmts): Change first argument from
7978 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7979 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7980 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7982 (get_location_for_stmts): Change first argument from
7983 auto_vec<gimple *> & to vec<gimple *> &.
7984 (struct split_store): Remove orig_stmts data member, add orig_stores.
7985 (split_store::split_store): Create orig_stores rather than orig_stmts.
7986 (find_constituent_stmts): Renamed to ...
7987 (find_constituent_stores): ... this. Change second argument from
7988 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7989 to info structures rather than the statements.
7990 (split_group): Rename ALLOW_UNALIGNED argument to
7991 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7992 it. Adjust find_constituent_stores caller.
7993 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7994 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7995 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7996 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7997 (mem_valid_for_store_merging): New function.
7998 (handled_load): New function.
7999 (pass_store_merging::process_store): New method.
8000 (pass_store_merging::execute): Use process_store method. Adjust
8001 terminate_all_aliasing_chains caller.
8003 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8005 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8006 Return true for more constants, symbols and label references.
8007 (aarch64_valid_floating_const): Remove unused function.
8009 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8012 * config/aarch64/aarch64.c (aarch64_layout_frame):
8013 Undo forcing of LR at bottom of frame.
8015 2017-11-03 Jeff Law <law@redhat.com>
8018 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8019 for int_registers_saved.
8021 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8022 extracted from tree-ssa-dom.c.
8023 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8024 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8025 (record_equivalences_from_incoming_edge): Add additional argument
8026 to single_pred_edge_ignoring_loop_edges call.
8027 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8028 (uncprop_dom_walker::before_dom_children): Add additional argument
8029 to single_pred_edge_ignoring_loop_edges call.
8030 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8031 single_pred_edge_ignoring_loop_edges rather than open coding.
8032 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8034 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8036 * match.pd (-(-A)): Rewrite.
8038 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8040 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8041 (rs6000_emit_int_cmove): New declaration.
8042 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8043 (rs6000_emit_sISEL): Delete.
8044 (rs6000_emit_int_cmove): Make non-static.
8045 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8046 instead of rs6000_emit_sISEL.
8048 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8050 * asan.c (create_cond_insert_point): Maintain profile.
8051 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8053 * basic-block.h (struct basic_block_def): Remove frequency.
8054 (EDGE_FREQUENCY): Use to_frequency
8055 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8057 (find_traces): Update to use to_frequency.
8058 (find_traces_1_round): Likewise; use only IPA counts.
8059 (bb_to_key): Likewise.
8060 (connect_traces): Use IPA counts only.
8061 (copy_bb_p): Update to use to_frequency.
8062 (fix_up_crossing_landing_pad): Likewise.
8063 (sanitize_hot_paths): Likewise.
8064 * bt-load.c (basic_block_freq): Likewise.
8065 * cfg.c (init_flow): Set count_max to uninitialized.
8066 (check_bb_profile): Remove frequencies; check counts.
8067 (dump_bb_info): Do not dump frequencies.
8068 (update_bb_profile_for_threading): Update counts only.
8069 (scale_bbs_frequencies_int): Likewise.
8070 (MAX_SAFE_MULTIPLIER): Remove.
8071 (scale_bbs_frequencies_gcov_type): Update counts only.
8072 (scale_bbs_frequencies_profile_count): Update counts only.
8073 (scale_bbs_frequencies): Update counts only.
8074 * cfg.h (struct control_flow_graph): Add count-max.
8075 (update_bb_profile_for_threading): Update prototype.
8076 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8077 (find_many_sub_basic_blocks): Likewise.
8078 * cfgcleanup.c (try_forward_edges): Likewise.
8079 (try_crossjump_to_edge): Likewise.
8080 * cfgexpand.c (expand_gimple_cond): Likewise.
8081 (expand_gimple_tailcall): Likewise.
8082 (construct_init_block): Likewise.
8083 (construct_exit_block): Likewise.
8084 * cfghooks.c (verify_flow_info): Check consistency of counts.
8085 (dump_bb_for_graph): Do not dump frequencies.
8086 (split_block_1): Do not update frequencies.
8087 (split_edge): Do not update frequencies.
8088 (make_forwarder_block): Do not update frequencies.
8089 (duplicate_block): Do not update frequencies.
8090 (account_profile_record): Do not update frequencies.
8091 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8092 for global heuristics.
8093 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8094 (expected_loop_iterations_unbounded): Use counts only.
8095 * cfgloopmanip.c (scale_loop_profile): Simplify.
8096 (create_empty_loop_on_edge): Simplify
8098 (duplicate_loop_to_header_edge): Simplify
8099 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8100 (update_br_prob_note): Take care of removing note when profile
8102 (relink_block_chain): Do not dump frequency.
8103 (rtl_account_profile_record): Use to_frequency.
8104 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8105 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8106 (cgraph_update_edges_for_call_stmt_node): Likewise.
8107 (cgraph_edge::verify_count_and_frequency): Update.
8108 (cgraph_node::verify_node): Temporarily disable frequency verification.
8109 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8110 to_cgraph_frequency.
8111 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8112 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8114 (cgraph_node::expand_thunk): Update profile.
8115 * except.c (dw2_build_landing_pads): Do not update frequency.
8116 * final.c (compute_alignments): Use to_frequency.
8117 (dump_basic_block_info): Do not dump frequency.
8118 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8119 (dump_gimple_bb_header): Do not dump frequency.
8120 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8122 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8123 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8124 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8125 (init_before_recovery): Do not update frequency.
8126 (sched_create_recovery_edges): Do not update frequency.
8127 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8128 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8129 (ipa_cp_c_finalize): Set max_count to uninitialized.
8130 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8131 (param_change_prob): Use counts.
8132 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8134 * ipa-split.c (consider_split): Use to_frequency.
8135 (split_function): Use to_frequency.
8136 * ira-build.c (loop_compare_func): Likewise.
8137 (mark_loops_for_removal): Likewise.
8138 (mark_all_loops_for_removal): Likewise.
8139 * loop-doloop.c (doloop_modify): Do not update frequency.
8140 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8142 * lto-streamer-in.c (input_function): Update count_max.
8143 * omp-expand.c (expand_omp_taskreg): Update count_max.
8144 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8145 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8146 (maybe_hot_count_p): Use ipa counts only.
8147 (maybe_hot_bb_p): Simplify.
8148 (maybe_hot_edge_p): Simplify.
8149 (probably_never_executed): Do not take frequency argument.
8150 (probably_never_executed_bb_p): Do not pass frequency.
8151 (probably_never_executed_edge_p): Likewise.
8152 (combine_predictions_for_bb): Check that profile is nonzero.
8153 (propagate_freq): Do not set frequency.
8154 (drop_profile): Simplify.
8155 (counts_to_freqs): Simplify.
8156 (expensive_function_p): Use to_frequency.
8157 (propagate_unlikely_bbs_forward): Simplify.
8158 (determine_unlikely_bbs): Simplify.
8159 (estimate_bb_frequencies): Add hack to silence graphite issues.
8160 (compute_function_frequency): Use ipa counts.
8161 (pass_profile::execute): Update.
8162 (rebuild_frequencies): Use counts only.
8163 (force_edge_cold): Use counts only.
8164 * profile-count.c (profile_count::dump): Dump new count types.
8165 (profile_count::differs_from_p): Check compatiblity.
8166 (profile_count::to_frequency): New function.
8167 (profile_count::to_cgraph_frequency): New function.
8168 * profile-count.h (struct function): Declare.
8169 (enum profile_quality): Add profile_guessed_local and
8170 profile_guessed_global0.
8171 (class profile_proability): Decrease number of bits to 29;
8172 update from_reg_br_prob_note and to_reg_br_prob_note.
8173 (class profile_count: Update comment; decrease number of bits
8174 to 61. Check compatibility.
8175 (profile_count::compatible_p): New private member function.
8176 (profile_count::ipa_p): New member function.
8177 (profile_count::operator<): Handle global zero correctly.
8178 (profile_count::operator>): Handle global zero correctly.
8179 (profile_count::operator<=): Handle global zero correctly.
8180 (profile_count::operator>=): Handle global zero correctly.
8181 (profile_count::nonzero_p): New member function.
8182 (profile_count::force_nonzero): New member function.
8183 (profile_count::max): New member function.
8184 (profile_count::apply_scale): Handle IPA scalling.
8185 (profile_count::guessed_local): New member function.
8186 (profile_count::global0): New member function.
8187 (profile_count::ipa): New member function.
8188 (profile_count::to_frequency): Declare.
8189 (profile_count::to_cgraph_frequency): Declare.
8190 * profile.c (OVERLAP_BASE): Delete.
8191 (compute_frequency_overlap): Delete.
8192 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8193 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8194 * sched-ebb.c (rank): Use counts only.
8195 * shrink-wrap.c (handle_simple_exit): Use counts only.
8196 (try_shrink_wrapping): Use counts only.
8197 (place_prologue_for_one_component): Use counts only.
8198 * tracer.c (find_best_predecessor): Use to_frequency.
8199 (find_trace): Use to_frequency.
8200 (tail_duplicate): Use to_frequency.
8201 * trans-mem.c (expand_transaction): Do not update frequency.
8202 * tree-call-cdce.c: Do not update frequency.
8203 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8204 (gimple_merge_blocks): Likewise.
8205 (gimple_split_edge): Likewise.
8206 (gimple_duplicate_sese_region): Likewise.
8207 (gimple_duplicate_sese_tail): Likewise.
8208 (move_sese_region_to_fn): Likewise.
8209 (gimple_account_profile_record): Likewise.
8210 (insert_cond_bb): Likewise.
8211 * tree-complex.c (expand_complex_div_wide): Likewise.
8212 * tree-eh.c (lower_resx): Update profile.
8213 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8215 (initialize_cfun): Do not initialize frequencies
8216 (freqs_to_counts): Delete.
8217 (copy_cfg_body): Ignore count parameter.
8218 (copy_body): Update.
8219 (expand_call_inline): Update count_max.
8220 (optimize_inline_calls): Update count_max.
8221 (tree_function_versioning): Update count_max.
8222 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8223 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8225 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8226 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8227 (try_peel_loop): Likewise.
8228 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8230 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8231 (tree_transform_and_unroll_loop): Do not use frequencies
8232 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8233 Use reliable prediction only.
8234 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8235 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8236 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8237 probability scaling.
8238 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8239 not update frequency
8240 (any_remaining_duplicated_blocks): Likewise.
8241 (update_profile): Likewise.
8242 (estimated_freqs_path): Delete.
8243 (freqs_to_counts_path): Delete.
8244 (clear_counts_path): Delete.
8245 (ssa_fix_duplicate_block_edges): Likewise.
8246 (duplicate_thread_path): Likewise.
8247 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8248 * tree-tailcall.c (decrease_profile): Do not update frequency.
8249 (eliminate_tail_call): Likewise.
8250 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8251 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8252 (optimize_mask_stores): Likewise.
8253 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8254 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8255 (ubsan_expand_ptr_ifn): Update profile.
8256 * value-prof.c (gimple_ic): Simplify.
8257 * value-prof.h (gimple_ic): Update prototype.
8258 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8259 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8261 (want_inline_self_recursive_call_p): Likewise.
8262 (resolve_noninline_speculation): Only cummulate defined counts.
8263 (inline_small_functions): Use nonzero_p.
8264 (ipa_inline): Do not access freed node.
8266 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8268 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8269 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8271 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8273 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8274 non-legitimate address.
8276 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8278 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8279 (*lt0_<mode>di, *lt0_<mode>si): New.
8281 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8283 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8284 TARGET_PAIRED_FLOAT.
8286 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8287 Jim Wilson <jim.wilson@linaro.org>
8289 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8290 * config/aarch64/aarch64-tune.md: Regenerated.
8291 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8292 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8294 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8296 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8297 (arc_expand_prologue): Restore blink for millicode.
8298 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8300 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8303 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8304 gen_vec_duplicate after forcing the scalar into a register.
8306 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8308 * combine (try_combine): Print the insns input to try_combine to the
8311 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8314 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8315 Remove second argument from aarch64_process_target_attr call.
8316 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8318 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8320 (aarch64_handle_attr_arch): Remove second argument.
8321 (aarch64_handle_attr_cpu): Ditto.
8322 (aarch64_handle_attr_tune): Ditto.
8323 (aarch64_handle_attr_isa_flags): Ditto.
8324 (aarch64_process_one_target_attr): Ditto.
8325 (aarch64_process_target_attr): Ditto.
8326 (aarch64_option_valid_attribute_p): Remove second argument.
8327 on aarch64_process_target_attr call.
8329 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8331 * diagnostic.c: Include "selftest-diagnostic.h".
8332 (selftest::assert_location_text): New function.
8333 (selftest::test_diagnostic_get_location_text): New function.
8334 (selftest::diagnostic_c_tests): Call it.
8336 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8338 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8339 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8340 (class selftest::test_diagnostic_context): Move to...
8341 * selftest-diagnostic.c: New file.
8342 * selftest-diagnostic.h: New file.
8344 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8346 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8347 offset range for FT32B.
8348 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8349 * config/ft32/ft32.md: Add TARGET_NOPM.
8350 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8351 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8353 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8355 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8357 2017-11-02 Jeff Law <law@redhat.com>
8359 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8360 virtual keyword on FINAL OVERRIDE members.
8362 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8363 virtuals together. Add virtual destructor.
8364 (substitute_and_fold_engine): Similarly.
8366 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8368 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8370 2017-11-02 Richard Biener <rguenther@suse.de>
8372 PR tree-optimization/82795
8373 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8375 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8377 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8379 (gcc_GAS_CHECK_FEATURE): Remove.
8380 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8381 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8382 * configure: Regenerate.
8384 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8386 * config/arc/arc.c (hwloop_optimize): Account for empty
8389 2017-11-02 Richard Biener <rguenther@suse.de>
8392 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8393 Truncate ARRAY_REF index and element size.
8395 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8397 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8399 2017-11-01 Jeff Law <law@redhat.com>
8401 * tree-ssa-ccp.c (ccp_folder): New class derived from
8402 substitute_and_fold_engine.
8403 (ccp_folder::get_value): New member function.
8404 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8405 (ccp_fold_stmt): Remove prototype.
8406 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8407 * tree-ssa-copy.c (copy_folder): New class derived from
8408 substitute_and_fold_engine.
8409 (copy_folder::get_value): Renamed from get_value.
8410 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8411 * tree-vrp.c (vrp_folder): New class derived from
8412 substitute_and_fold_engine.
8413 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8414 (vrp_folder::get_value): New member function.
8415 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8416 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8417 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8418 provide a class interface to folder/substitute routines.
8419 (ssa_prop_fold_stmt_fn): Remove typedef.
8420 (ssa_prop_get_value_fn): Likewise.
8421 (subsitute_and_fold): Remove prototype.
8422 (replace_uses_in): Likewise.
8423 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8424 Renamed from replace_uses_in. Call the virtual member function
8425 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8426 (substitute_and_fold_dom_walker): Remove initialization of
8427 data member entries for calbacks. Add substitute_and_fold_engine
8428 member and initialize it.
8429 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8430 member functions for get_value, replace_phi_args_in c
8431 replace_uses_in, and fold_stmt calls.
8432 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8433 substitute_and_fold. Remove assert. Update ctor call.
8435 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8436 (ssa_prop_visit_phi_fn): Likewise.
8437 (class ssa_propagation_engine): New class to provide an interface
8439 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8441 (ssa_prop_visit_phi): Likewise.
8442 (ssa_propagation_engine::simulate_stmt): Moved into class.
8443 Call visit_phi/visit_stmt from the class rather than via
8444 file scoped static variables.
8445 (ssa_propagation_engine::simulate_block): Moved into class.
8446 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8447 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8448 set file scoped statics for the visit_stmt/visit_phi callbacks.
8449 * tree-complex.c (complex_propagate): New class derived from
8450 ssa_propagation_engine.
8451 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8452 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8453 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8455 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8456 ssa_propagation_engine.
8457 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8458 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8459 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8460 * tree-ssa-copy.c (copy_prop): New class derived from
8461 ssa_propagation_engine.
8462 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8463 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8464 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8465 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8466 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8467 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8468 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8470 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8472 PR rtl-optimization/82778
8473 PR rtl-optimization/82597
8474 * compare-elim.c (struct comparison): Add in_a_setter field.
8475 (find_comparison_dom_walker::before_dom_children): Remove killed
8476 bitmap and df_simulate_find_defs call, instead walk the defs.
8477 Compute last_setter and initialize in_a_setter. Merge definitions
8478 with first initialization for a few variables.
8479 (try_validate_parallel): Use insn_invalid_p instead of
8480 recog_memoized. Return insn rather than just the pattern.
8481 (try_merge_compare): Fix up comment. Don't uselessly test if
8482 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8484 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8487 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8488 Alan Hayward <alan.hayward@arm.com>
8489 David Sherwood <david.sherwood@arm.com>
8491 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8492 aarch64_hard_regno_nregs to get the number of registers
8495 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8496 Alan Hayward <alan.hayward@arm.com>
8497 David Sherwood <david.sherwood@arm.com>
8499 * config/aarch64/constraints.md (Upl): Rename to...
8501 * config/aarch64/aarch64.md
8502 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8505 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8506 Alan Hayward <alan.hayward@arm.com>
8507 David Sherwood <david.sherwood@arm.com>
8509 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8510 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8514 Alan Hayward <alan.hayward@arm.com>
8515 David Sherwood <david.sherwood@arm.com>
8517 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8518 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8519 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8521 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8522 of a single element.
8523 * config/aarch64/iterators.md: Add a comment above the permute
8524 unspecs to say that they are generated directly by
8525 aarch64_expand_vec_perm_const.
8526 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8528 2017-11-01 Nathan Sidwell <nathan@acm.org>
8530 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8532 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8534 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8535 and medany code models, and describe what they do.
8537 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8539 Revert accidental duplicate:
8541 * combine.c (can_change_dest_mode): Reject changes in
8542 REGMODE_NATURAL_SIZE.
8544 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8546 PR rtl-optimization/64682
8547 PR rtl-optimization/69567
8548 PR rtl-optimization/69737
8549 PR rtl-optimization/82683
8550 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8551 register mentioned in the note, drop the note, unless it came from I3,
8552 in which case it should go to I3 again.
8554 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8556 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8557 and return false if not.
8558 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8560 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8562 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8563 range is known to be empty.
8565 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8566 Alan Hayward <alan.hayward@arm.com>
8567 David Sherwood <david.sherwood@arm.com>
8569 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8570 and CONST_VECTOR_NUNITS instead of computing the number of units from
8571 the byte sizes of the vector and element.
8572 (simplify_binary_operation_1): Likewise.
8573 (simplify_const_binary_operation): Likewise.
8574 (simplify_ternary_operation): Likewise.
8576 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8577 Alan Hayward <alan.hayward@arm.com>
8578 David Sherwood <david.sherwood@arm.com>
8580 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8581 (int_mem_offset): ...this new function.
8582 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8583 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8584 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8587 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8588 Alan Hayward <alan.hayward@arm.com>
8589 David Sherwood <david.sherwood@arm.com>
8591 * lower-subreg.c (interesting_mode_p): New function.
8592 (compute_costs, find_decomposable_subregs, decompose_register)
8593 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8594 (resolve_clobber, dump_choices): Use it.
8596 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8597 Alan Hayward <alan.hayward@arm.com>
8598 David Sherwood <david.sherwood@arm.com>
8600 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8602 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8603 Alan Hayward <alan.hayward@arm.com>
8604 David Sherwood <david.sherwood@arm.com>
8606 * alias.c (find_base_value, find_base_term): Only process integer
8607 truncations. Check the precision rather than the size.
8609 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8610 Alan Hayward <alan.hayward@arm.com>
8611 David Sherwood <david.sherwood@arm.com>
8613 * machmode.h (is_narrower_int_mode): New function
8614 * optabs.c (expand_float, expand_fix): Use it.
8615 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8617 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8618 Alan Hayward <alan.hayward@arm.com>
8619 David Sherwood <david.sherwood@arm.com>
8621 * rtl.h (narrower_subreg_mode): New function.
8622 * ira-color.c (update_costs_from_allocno): Use it.
8624 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8625 Alan Hayward <alan.hayward@arm.com>
8626 David Sherwood <david.sherwood@arm.com>
8628 * optabs-query.h (convert_optab_p): New function, split out from...
8629 (convert_optab_handler): ...here.
8630 (widening_optab_handler): Delete.
8631 (find_widening_optab_handler): Remove permit_non_widening parameter.
8632 (find_widening_optab_handler_and_mode): Likewise. Provide an
8633 override that operates on mode class wrappers.
8634 * optabs-query.c (widening_optab_handler): Delete.
8635 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8636 parameter. Assert that the two modes are the same class and that
8637 the "from" mode is narrower than the "to" mode. Use
8638 convert_optab_handler instead of widening_optab_handler.
8639 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8640 instead of widening_optab_handler.
8641 * expr.c (expand_expr_real_2): Update calls to
8642 find_widening_optab_handler.
8643 * optabs.c (expand_widen_pattern_expr): Likewise.
8644 (expand_binop_directly): Take the insn_code as a parameter.
8645 (expand_binop): Only call find_widening_optab_handler for
8646 conversion optabs; use optab_handler otherwise. Update calls
8647 to find_widening_optab_handler and expand_binop_directly.
8648 Use convert_optab_handler instead of widening_optab_handler.
8649 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8650 find_widening_optab_handler and use scalar_mode rather than
8652 (convert_plusminus_to_widen): Likewise.
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 * machmode.h (fixed_size_mode): New class.
8659 * rtl.h (get_pool_mode): Return fixed_size_mode.
8660 * gengtype.c (main): Add fixed_size_mode.
8661 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8662 (get_raw_arg_mode): Likewise.
8663 * doc/tm.texi: Regenerate.
8664 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8665 * targhooks.c (default_get_reg_raw_mode): Likewise.
8666 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8667 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8668 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8669 (msp430_get_raw_result_mode): Likewise.
8670 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8671 * dbxout.c (dbxout_parms): Require fixed-size modes.
8672 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8673 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8674 * omp-low.c (lower_oacc_reductions): Likewise.
8675 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8676 (simplify_subreg): Update accordingly.
8677 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8678 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8679 that aren't fixed-size.
8680 (get_pool_mode): Return a fixed_size_mode.
8681 (output_constant_pool_2): Take a fixed_size_mode.
8683 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8684 Alan Hayward <alan.hayward@arm.com>
8685 David Sherwood <david.sherwood@arm.com>
8687 * doc/rtl.texi (vec_series): Document.
8688 (const): Say that the operand can be a vec_series.
8689 * rtl.def (VEC_SERIES): New rtx code.
8690 * rtl.h (const_vec_series_p_1): Declare.
8691 (const_vec_series_p): New function.
8692 * emit-rtl.h (gen_const_vec_series): Declare.
8693 (gen_vec_series): Likewise.
8694 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8695 (gen_vec_series): Likewise.
8696 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8697 * simplify-rtx.c (simplify_unary_operation): Handle negations
8699 (simplify_binary_operation_series): New function.
8700 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8701 (test_vector_ops_series): New function.
8702 (test_vector_ops): Call it.
8703 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8704 gen_const_vec_series.
8705 (altivec_lvsr): Likewise.
8706 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8709 Alan Hayward <alan.hayward@arm.com>
8710 David Sherwood <david.sherwood@arm.com>
8712 * doc/rtl.texi (const): Update description of address constants.
8713 Say that vector constants are allowed too.
8714 * common.md (E, F): Use CONSTANT_P instead of checking for
8716 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8717 checking for CONST_VECTOR.
8718 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8720 * expr.c (expand_expr_real_2): Check for vector modes instead
8721 of checking for CONST_VECTOR.
8722 * rtl.h (const_vec_p): New function.
8723 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8724 (unwrap_const_vec_duplicate): Handle them here too.
8726 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8727 David Malcolm <dmalcolm@redhat.com>
8728 Alan Hayward <alan.hayward@arm.com>
8729 David Sherwood <david.sherwood@arm.com>
8731 * rtl.h (vec_duplicate_p): New function.
8732 * selftest-rtl.c (assert_rtx_eq_at): New function.
8733 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8734 (assert_rtx_eq_at): Declare.
8735 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8736 * selftest-run-tests.c (selftest::run_tests): Call it.
8737 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8738 (simplify_unary_operation_1): Recursively handle vector duplicates.
8739 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8741 (simplify_subreg): Handle subregs of vector duplicates.
8742 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8743 (selftest::simplify_rtx_c_tests): New functions.
8745 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8746 Alan Hayward <alan.hayward@arm.com>
8747 David Sherwood <david.sherwood@arm.com>
8749 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8750 (gen_vec_duplicate): Likewise.
8751 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8753 (gen_const_vector): ...here.
8754 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8755 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8756 whose elements are all equal.
8757 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8758 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8759 (simplify_relational_operation): Likewise.
8760 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8762 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8763 (aarch64_expand_vector_init): Likewise.
8764 * config/arm/arm.c (neon_vdup_constant): Likewise.
8765 (neon_expand_vector_init): Likewise.
8766 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8767 (arm_block_set_unaligned_vect): Likewise.
8768 (arm_block_set_aligned_vect): Likewise.
8769 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8770 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8771 (expand_vec_perm_even_odd_pack): Likewise.
8772 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8773 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8774 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8775 gen_const_vec_duplicate.
8776 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8777 * config/mips/mips.c (mips_gen_const_int_vector): Use
8778 gen_const_vec_duplicate.
8779 (mips_expand_vector_init): Use CONST0_RTX.
8780 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8781 (define_split): Use gen_const_vec_duplicate.
8782 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8783 (define_split): Use gen_const_vec_duplicate.
8784 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8785 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8786 * config/spu/spu.c (spu_const): Likewise.
8788 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8789 Alan Hayward <alan.hayward@arm.com>
8790 David Sherwood <david.sherwood@arm.com>
8792 * combine.c (can_change_dest_mode): Reject changes in
8793 REGMODE_NATURAL_SIZE.
8795 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8797 * configure.ac (--enable-libssp): New.
8798 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8799 trying to determine target-specific default. Adjust indentation.
8800 * configure: Regenerated.
8801 * doc/install.texi (Configuration): Expand --disable-libssp
8804 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8806 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8809 2017-10-31 Martin Jambor <mjambor@suse.cz>
8812 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8814 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8816 * auto-profile.c (autofdo_source_profile::read): Use
8817 UNKNOWN_LOCATION rather than 0.
8818 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8819 (warning_at): ...this overload.
8820 (warning_at_rich_loc_n): Rename to...
8821 (warning_n): ...this overload.
8822 (error_at_rich_loc): Rename to...
8823 (error_at): ...this overload.
8824 (pedwarn_at_rich_loc): Rename to...
8825 (pedwarn): ...this overload.
8826 (permerror_at_rich_loc): Rename to...
8827 (permerror): ...this overload.
8828 (inform_at_rich_loc): Rename to...
8829 (inform): ...this overload.
8830 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8831 (diagnostic_n_impl_richloc): Rename to...
8832 (diagnostic_n_impl): ...this rich_location *-based decl.
8833 (inform_at_rich_loc): Rename to...
8834 (inform): ...this, and add an assertion.
8835 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8836 (warning_at_rich_loc): Rename to...
8837 (warning_at): ...this, and add an assertion.
8838 (warning_at_rich_loc_n): Rename to...
8839 (warning_n): ...this, and add an assertion.
8840 (warning_n): Update location_t-based implementation for removal of
8841 location_t-based diagnostic_n_impl.
8842 (pedwarn_at_rich_loc): Rename to...
8843 (pedwarn): ...this, and add an assertion.
8844 (permerror_at_rich_loc): Rename to...
8845 (permerror): ...this, and add an assertion.
8846 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8847 (error_at_rich_loc): Rename to...
8848 (error_at): ...this, and add an assertion.
8849 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8850 (driver::do_spec_on_infiles): Likewise.
8851 * substring-locations.c (format_warning_va): Update for renaming
8852 of inform_at_rich_loc.
8854 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8856 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8857 _Float<N>X built-in functions so that the variant without the
8858 "__builtin_" prefix is only enabled for the GNU C and Objective C
8859 languages when they are in non-strict ANSI/ISO mode.
8860 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8861 * target.def (floatn_builtin_p): Add a target hook to control
8862 whether _Float<N> and _Float<N>X built-in functions without the
8863 "__builtin_" prefix are enabled, and return true for C and
8864 Objective C in the default hook. Include langhooks.h in
8866 * targhooks.h (default_floatn_builtin_p): Likewise.
8867 * targhooks.c (default_floatn_builtin_p): Likewise.
8868 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8869 floatn_builtin_p target hook.
8870 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8872 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8873 Eric Botcazou <ebotcazou@adacore.com>
8875 PR rtl-optimization/81803
8876 * lra-constraints.c (curr_insn_transform): Also reload the whole
8877 register for a strict subreg no wider than a word if this is for
8878 a WORD_REGISTER_OPERATIONS target.
8880 2017-10-31 Jason Merrill <jason@redhat.com>
8882 * gdbinit.in: Skip over inlines from timevar.h.
8884 2017-10-31 Martin Liska <mliska@suse.cz>
8886 * doc/gcov.texi: Document new option.
8887 * gcov.c (print_usage): Likewise print it.
8888 (process_args): Support the argument.
8889 (format_count): New function.
8890 (format_gcov): Use the function.
8892 2017-10-31 Martin Liska <mliska@suse.cz>
8894 * gcov.c (struct name_map): do not use typedef.
8895 Define operator== and operator<.
8896 (name_search): Remove.
8897 (name_sort): Remove.
8898 (main): Do not allocate names.
8899 (process_file): Add vertical space.
8900 (generate_results): Use std::find.
8901 (release_structures): Do not release memory.
8902 (find_source): Use std::find.
8904 2017-10-31 Martin Liska <mliska@suse.cz>
8906 * gcov.c (struct line_info): Remove it's typedef.
8907 (line_info::line_info): Add proper ctor.
8908 (line_info::has_block): Do not use a typedef.
8909 (struct source_info): Do not use typedef.
8910 (circuit): Likewise.
8911 (get_cycles_count): Likewise.
8912 (output_intermediate_file): Iterate via vector iterator.
8913 (add_line_counts): Use std::vector methods.
8914 (accumulate_line_counts): Likewise.
8915 (output_lines): Likewise.
8917 2017-10-31 Martin Liska <mliska@suse.cz>
8919 * gcov.c (struct source_info): Remove typedef.
8920 (source_info::source_info): Add proper ctor.
8921 (accumulate_line_counts): Use struct, not it's typedef.
8922 (output_gcov_file): Likewise.
8923 (output_lines): Likewise.
8924 (main): Do not allocate an array.
8925 (output_intermediate_file): Use size of vector container.
8926 (process_file): Resize the vector.
8927 (generate_results): Do not preallocate, use newly added vector
8929 (release_structures): Do not release sources.
8930 (find_source): Use vector methods.
8931 (add_line_counts): Do not use typedef.
8933 2017-10-31 Martin Liska <mliska@suse.cz>
8935 * doc/gcov.texi: Document that.
8936 * gcov.c (add_line_counts): Mark lines with a non-executed
8938 (output_line_beginning): Handle such lines.
8939 (output_lines): Pass new argument.
8940 (output_intermediate_file): Print it in intermediate format.
8942 2017-10-31 Martin Liska <mliska@suse.cz>
8944 * color-macros.h: New file.
8945 * diagnostic-color.c: Factor out color related to macros to
8947 * doc/gcov.texi: Document -k option.
8948 * gcov.c (INCLUDE_STRING): Include string.h.
8949 (print_usage): Add -k option.
8950 (process_args): Parse it.
8951 (pad_count_string): New function.
8952 (output_line_beginning): Likewise.
8953 (DEFAULT_LINE_START): New macro.
8954 (output_lines): Support color output.
8956 2017-10-31 Martin Liska <mliska@suse.cz>
8958 PR gcov-profile/82633
8959 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8960 their interaction with GCOV infrastructure.
8961 * configure.ac: Add -fkeep-{inline,static}-functions to
8963 * configure: Regenerate.
8965 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8968 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8970 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8973 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8974 into a register if it does not fit into an immediate offset field.
8976 2017-10-31 Olivier Hainque <hainque@adacore.com>
8978 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8980 2017-10-31 Julia Koval <julia.koval@intel.com>
8982 * config.gcc: Add gfniintrin.h.
8983 * config/i386/gfniintrin.h: New.
8984 * config/i386/i386-builtin-types.def
8985 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8986 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8987 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8988 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8989 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8990 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8991 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8992 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8993 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8994 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8995 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8996 * config/i386/immintrin.h: Include gfniintrin.h.
8997 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8999 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9001 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9003 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9005 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9006 (arm_ashldi3_1bit): Remove pattern.
9007 (ashrdi3): Remove shift by 1 expansion.
9008 (arm_ashrdi3_1bit): Remove pattern.
9009 (lshrdi3): Remove shift by 1 expansion.
9010 (arm_lshrdi3_1bit): Remove pattern.
9011 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9012 cost of ashldi3 by 1.
9013 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9014 (<shift>di3_neon): Likewise.
9016 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9018 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9019 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9020 and (*aarch64_simd_mov<VQ:mode>).
9021 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9022 pattern alternative.
9023 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9024 attributes to match pattern alternative.
9026 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9028 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9029 * config/rs6000/emmintrin.h: New file.
9030 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9032 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9034 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9035 * (movdi_vfp_cortexa8): Remove pattern.
9037 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9039 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9041 (Specific, alpha*-dec-osf5.1): Remove.
9042 (Specific, mips-sgi-irix5): Remove.
9043 (Specific, mips-sgi-irix6): Remove.
9045 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9048 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9049 arguments to clear_bit_region_be.
9051 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9053 * gimplify.c: Include memmodel.h.
9055 2017-10-30 Martin Jambor <mjambor@suse.cz>
9057 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9058 condition whether loop should be executed at all.
9060 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9063 gimple folding of vec_madd() intrinsics.
9064 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9065 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9066 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9068 2017-10-30 Richard Biener <rguenther@suse.de>
9070 PR tree-optimization/82762
9072 2017-10-23 Richard Biener <rguenther@suse.de>
9074 PR tree-optimization/82129
9076 2017-08-01 Richard Biener <rguenther@suse.de>
9078 PR tree-optimization/81181
9079 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9080 (compute_antic): ... end of iteration here.
9082 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9084 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9085 -std=iso9899:2017 and -std=gnu17.
9086 * doc/standards.texi (C Language): Document C17 support.
9087 * doc/cpp.texi (Overview): Mention -std=c17.
9088 (Standard Predefined Macros): Document C11 and C17 values of
9089 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9090 * doc/extend.texi (Inline): Do not list individual options for
9091 standards newer than C99.
9092 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9094 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9097 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9099 * asan.c (asan_finish_file): Align asan globals array by shadow
9102 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9105 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9106 (struct store_immediate_info): Add bitregion_start and bitregion_end
9108 (store_immediate_info::store_immediate_info): Add brs and bre
9109 arguments and initialize bitregion_{start,end} from those.
9110 (struct merged_store_group): Add bitregion_start, bitregion_end,
9111 align_base and mask fields. Drop unnecessary struct keyword from
9112 struct store_immediate_info. Add do_merge method.
9113 (clear_bit_region_be): Use memset instead of loop storing zeros.
9114 (merged_store_group::do_merge): New method.
9115 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9116 stores as long as the surrounding bitregions have no gaps.
9117 (merged_store_group::merge_overlapping): Use do_merge.
9118 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9119 is byte aligned, rather than requiring that start and width are
9120 byte aligned. Drop unnecessary struct keyword from
9121 struct store_immediate_info. Allocate and populate also mask array.
9122 Make start of the arrays relative to bitregion_start rather than
9123 start and size them according to bitregion_{end,start} difference.
9124 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9125 struct store_immediate_info.
9126 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9127 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9128 struct keyword from struct store_immediate_info.
9129 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9130 between stores as long as the surrounding bitregions have no gaps.
9132 (struct split_store): Add orig non-static data member.
9133 (split_store::split_store): Initialize orig to false.
9134 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9135 if there is exactly a single original stmt. Change stmts argument
9136 to pointer from reference, if NULL, don't push anything to it. Add
9137 first argument, use it to optimize skipping over orig stmts that
9138 are known to be before bitpos already. Simplify.
9139 (split_group): Return unsigned int count how many stores are or
9140 would be needed rather than a bool. Add allow_unaligned argument.
9141 Change split_stores argument from reference to pointer, if NULL,
9142 only do a dry run computing how many stores would be produced.
9143 Rewritten algorithm to use both alignment and misalign if
9144 !allow_unaligned and handle bitfield stores with gaps.
9145 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9146 from bitregion_start instead of start. Compute allow_unaligned
9147 here, if true, do 2 split_group dry runs to compute which one
9148 produces fewer stores and prefer aligned if equal. Punt if
9149 new count is bigger or equal than original before emitting any
9150 statements, rather than during that. Remove no longer needed
9151 new_ssa_names tracking. Replace num_stmts with
9152 split_stores.length (). Use 32-bit stack allocated entries
9153 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9154 if possible. Handle bitfields with gaps.
9155 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9156 Compute bitregion_{start,end} for the stores and construct
9157 store_immediate_info with that. Formatting fixes.
9159 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9162 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9163 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9165 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9167 * gimplify.c: Include tm_p.h.
9169 * common.opt (gcoff): Re-add as ignored option.
9170 (gcoff1, gcoff2, gcoff3): Likewise.
9172 * Makefile.in (OBJS): Delete sdbout.o.
9173 (GTFILES): Delete $(srcdir)/sdbout.c.
9174 * debug.h: Delete sdb_debug_hooks.
9175 * final.c: Delete sdbout.h include.
9176 (final_scan_insn): Delete SDB_DEBUG check.
9177 (rest_of_clean_state): Likewise.
9178 * output.h: Delete sdb_begin_function_line.
9181 * toplev.c: Delete sdbout.h include.
9182 (process_options): Delete SDB_DEBUG check.
9183 * tree-core.h (tree_type_common): Delete pointer field of
9185 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9186 TYPE_SYMTAB_POINTER.
9187 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9188 (TYPE_SYMTAB_IS_POINTER): Delete.
9189 (TYPE_SYMTAB_IS_DIE): Renumber.
9190 * xcoffout.c: Refer to former sdbout.c file.
9191 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9193 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9194 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9195 (Debugging Options): Delete -gcoff.
9196 (-gstabs): Delete SDB reference.
9198 (-gcoff@var{level}): Delete.
9199 * doc/passes.texi (Debugging information output): Delete SDB and
9200 sdbout.c references.
9201 * doc/tm.texi: Regenerate.
9202 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9203 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9205 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9206 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9207 (SDB_DEBUGGING_INFO): Delete.
9208 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9209 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9210 * target.def (output_source_filename): Delete COFF reference.
9212 * common.opt (gcoff): Delete.
9213 (gxcoff+): Update Negative chain.
9214 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9216 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9217 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9218 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9220 (expand_function_start): Change sdb reference to past tense.
9221 (expand_function_end): Change sdb reference to past tense.
9222 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9223 * opts.c (debug_type_names): Delete coff entry.
9224 (common_handle_option): Delete OPT_gcoff case.
9225 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9227 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9228 * config/cris/cris.h: Delete SDB reference in comment.
9229 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9230 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9231 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9232 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9234 (ix86_expand_prologue): Likewise.
9235 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9236 * config/ia64/ia64.h: Likewise.
9237 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9238 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9240 * config/mmix/mmix.h: Likewise.
9241 * config/nds32/nds32.c: Likewise.
9242 * config/stormy/storym16.h: Likewise.
9243 * config/visium/visium.h: Likewise.
9244 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9246 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9248 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9249 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9250 FRAME_POINTER_REGNUM point at high end of local var area.
9252 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9254 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9255 Move comment around. Do not reset best_edge for a copiable
9256 destination if the copy would cause a partition change.
9257 (better_edge_p): Remove redundant check.
9259 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9261 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9263 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9265 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9266 for math functions that have _Float<N> and _Float<N>X variants.
9267 (mathfn_built_in_2): Add support for math functions that have
9268 _Float<N> and _Float<N>X variants.
9269 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9270 (expand_builtin_mathfn_ternary): Add support for fma with
9271 _Float<N> and _Float<N>X variants.
9272 (expand_builtin): Likewise.
9273 (fold_builtin_3): Likewise.
9274 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9275 create math function _Float<N> and _Float<N>X variants as external
9277 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9278 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9279 the __builtin_ prefix and if not strict ansi, without the prefix.
9280 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9281 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9282 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9283 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9284 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9285 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9286 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9287 function signatures for fma _Float<N> and _Float<N>X variants.
9288 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9289 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9290 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9291 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9292 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9293 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9294 * gencfn-macros.c (print_case_cfn): Add support for math functions
9295 that have _Float<N> and _Float<N>X variants.
9296 (print_define_operator_list): Likewise.
9297 (fltfn_suffixes): Likewise.
9299 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9300 for math functions that have _Float<N> and _Float<N>X variants.
9301 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9302 and _Float<N>X variants.
9303 (COPYSIGN): Likewise.
9306 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9307 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9309 (integer_valued_read_call_p): Likewise.
9310 * fold-const-call.c (fold_const_call_ss): Likewise.
9311 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9312 _Float<N> and _Float<N>X variants.
9313 (fold_const_call_ssss): Add support for fma _Float<N> and
9314 _Float<N>X variants.
9315 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9316 support for copysign and fma _Float<N> and _Float<N>X variants.
9317 (backprop::process_builtin_call_use): Likewise.
9318 * tree-call-cdce.c (can_test_argument_range); Add support for
9319 sqrt _Float<N> and _Float<N>X variants.
9320 (edom_only_function): Likewise.
9321 (get_no_error_domain): Likewise.
9322 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9323 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9324 copysign _Float<N> and _Float<N>X variants.
9325 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9326 handled by machine independent code.
9327 (FMAF128): Likewise.
9328 * doc/cpp.texi (Common Predefined Macros): Document defining
9329 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9330 fma _Float<N> and _Float<N>X variants.
9332 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9335 * config/i386/i386-modes.def (CCFPU): Remove definition.
9336 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9337 (ix86_cc_modes_compatible): Ditto.
9338 (ix86_expand_carry_flag_compare): Ditto.
9339 (ix86_expand_int_movcc): Ditto.
9340 (ix86_expand_int_addcc): Ditto.
9341 (ix86_reverse_condition): Ditto.
9342 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9343 Return true/false for unordered/ordered fp comparisons.
9344 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9345 (ix86_prepare_fp_compare_args): Update for rename.
9346 (ix86_expand_fp_compare): Update for rename. Generate unordered
9347 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9348 (ix86_expand_sse_compare_and_jump): Ditto.
9349 * config/i386/predicates.md (fcmov_comparison_operator):
9350 Remove CCFPU mode handling.
9351 (ix86_comparison_operator): Ditto.
9352 (ix86_carry_flag_operator): Ditto.
9353 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9354 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9355 (*cmpu<mode>_cc_i387): Ditto.
9356 (FPCMP): Remove mode iterator.
9357 (unord): Remove mode attribute.
9358 (unord_subst): New define_subst transformation
9359 (unord): New define_subst attribute.
9361 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9362 (*cmpi<unord>xf_i387): Ditto.
9363 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9364 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9365 using unord_subst transformation.
9366 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9367 (round_saeonly): Also handle CCFP mode.
9368 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9369 Remove UNSPEC_SAHF unspec handling.
9371 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9373 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9375 2017-10-27 Jeff Law <law@redhat.com>
9377 * gimple-ssa-sprintf.c: Include domwalk.h.
9378 (class sprintf_dom_walker): New class, derived from dom_walker.
9379 (sprintf_dom_walker::before_dom_children): New function.
9380 (struct call_info): Moved into sprintf_dom_walker class
9381 (compute_formath_length, handle_gimple_call): Likewise.
9382 (sprintf_length::execute): Call the dominator walker rather
9383 than walking the statements.
9385 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9386 gimple statement locations.
9387 (check_array_bounds): Corresponding changes. Get the statement's
9388 location directly from wi->stmt.
9390 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9393 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9395 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9397 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9398 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9400 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9403 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9404 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9405 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9406 maybe_get_pool_constant.
9407 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9410 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9412 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9413 2.26 caveat. Update gas and gld versions.
9414 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9417 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9419 * cgraph.h (set_malloc_flag): Declare.
9420 * cgraph.c (set_malloc_flag_1): New function.
9421 (set_malloc_flag): Likewise.
9422 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9423 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9425 (read_ipa_call_summary): Add support for reading is_return_callee.
9426 (write_ipa_call_summary): Stream is_return_callee.
9427 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9428 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9429 ipa-prop.h, ipa-fnsummary.h.
9430 (pure_const_names): Change to static.
9431 (malloc_state_e): Define.
9432 (malloc_state_names): Define.
9433 (funct_state_d): Add field malloc_state.
9434 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9435 (check_retval_uses): New function.
9436 (malloc_candidate_p): Likewise.
9437 (analyze_function): Add support for malloc attribute.
9438 (pure_const_write_summary): Stream malloc_state.
9439 (pure_const_read_summary): Add support for reading malloc_state.
9440 (dump_malloc_lattice): New function.
9441 (propagate_malloc): New function.
9442 (warn_function_malloc): New function.
9443 (ipa_pure_const::execute): Call propagate_malloc and
9444 ipa_free_fn_summary.
9445 (pass_local_pure_const::execute): Add support for malloc attribute.
9446 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9447 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9449 2017-10-27 Martin Liska <mliska@suse.cz>
9451 PR gcov-profile/82457
9452 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9453 for fork-like functions to be properly instrumented.
9455 2017-10-27 Richard Biener <rguenther@suse.de>
9458 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9459 info when we redirected EH.
9461 2017-10-26 Michael Collison <michael.collison@arm.com>
9463 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9465 (<optab>_trunchf<GPI:mode>2: New pattern.
9466 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9467 * config/aarch64/iterators.md (wv): New mode attribute.
9468 (vf, VF): New mode attributes.
9469 (vgp, VGP): New mode attributes.
9470 (s): Update attribute with SImode and DImode prefixes.
9472 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9474 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9475 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9476 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9477 (nios2_option_overide): Initialize it. Don't allow R0-relative
9478 addressing with PIC.
9479 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9480 (nios2_symbolic_constant_p): Likewise.
9481 (nios2_legitimate_address_p): Likewise.
9482 (nios2_r0rel_section_name_p): New.
9483 (nios2_symbol_ref_in_r0rel_data_p): New.
9484 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9485 (r0rel_constant_p): New.
9486 (nios2_print_operand_address): Handle r0rel_constant_p.
9487 (nios2_cdx_narrow_form_p): Likewise.
9488 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9489 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9490 (Nios II Options): Document -mr0rel-sec.
9492 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9494 * config/nios2/nios2.c: Include xregex.h.
9495 (nios2_gprel_sec_regex): New.
9496 (nios2_option_overide): Initialize it. Don't allow GP-relative
9497 addressing with PIC.
9498 (nios2_small_section_name_p): Check for regex match.
9499 * config/nios2/nios2.opt (mgprel-sec=): New option.
9500 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9501 (Nios II Options): Document -mgprel-sec.
9503 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9505 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9507 2017-10-26 Tom de Vries <tom@codesourcery.com>
9509 PR tree-optimization/82707
9510 * gimple.c (gimple_copy): Fix unsharing of
9511 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9513 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9515 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9516 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9517 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9518 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9519 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9520 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9521 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9522 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9523 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9524 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9525 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9526 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9527 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9528 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9529 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9530 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9531 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9533 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9535 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9537 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9538 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9539 warning to rs6000.c. Remove the Undocumented flag, since it has
9541 (-mabi=ibmlongdouble): Likewise.
9542 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9543 already set, set the default format for long double.
9544 (rs6000_debug_reg_global): Print whether long double is IBM or
9546 (rs6000_option_override_internal): Rework setting long double
9547 format. Only warn if the user is changing the long double default
9548 and they did not use -Wno-psabi.
9549 * doc/invoke.texi (PowerPC options): Update the documentation for
9550 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9552 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9553 Alan Hayward <alan.hayward@arm.com>
9554 David Sherwood <david.sherwood@arm.com>
9556 * rtl.h (wider_subreg_mode): New function.
9557 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9558 rather than an unsigned int *.
9559 * ira-color.c (regno_max_ref_width): Replace with...
9560 (regno_max_ref_mode): ...this new variable.
9561 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9562 Use wider_subreg_mode.
9563 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9564 rather than an unsigned int *.
9565 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9566 (process_alt_operands): Likewise.
9567 (invariant_p): Likewise.
9568 * lra-spills.c (assign_mem_slot): Likewise.
9569 (add_pseudo_to_slot): Likewise.
9570 * lra.c (collect_non_operand_hard_regs): Likewise.
9571 (add_regs_to_insn_regno_info): Likewise.
9572 * reload1.c (regno_max_ref_width): Replace with...
9573 (regno_max_ref_mode): ...this new variable.
9574 (reload): Update accordingly. Update call to
9575 ira_sort_regnos_for_alter_reg.
9576 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9577 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9578 (scan_paradoxical_subregs): Likewise.
9580 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9582 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9583 (aarch64_frame): Add emit_frame_chain boolean.
9584 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9585 Move eh_return case to aarch64_layout_frame.
9586 (aarch64_layout_frame): Initialize emit_frame_chain.
9587 (aarch64_expand_prologue): Use emit_frame_chain.
9589 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9591 * config/aarch64/aarch64.c (aarch64_layout_frame):
9592 Ensure LR is always stored at the bottom of the callee-saves.
9593 Remove rarely used frame layout which saves callee-saves at top of
9594 frame, so the store of LR can be used as a valid probe in all cases.
9596 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9598 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9599 Improve unaligned TImode/TFmode base/offset split.
9601 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9602 Alan Hayward <alan.hayward@arm.com>
9603 David Sherwood <david.sherwood@arm.com>
9605 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9606 * combine.c (find_single_use_1): Likewise.
9607 (expand_field_assignment): Likewise.
9608 (move_deaths): Likewise.
9609 * lra-constraints.c (simplify_operand_subreg): Likewise.
9610 (curr_insn_transform): Likewise.
9611 * lra.c (collect_non_operand_hard_regs): Likewise.
9612 (add_regs_to_insn_regno_info): Likewise.
9613 * rtlanal.c (reg_referenced_p): Likewise.
9614 (covers_regno_no_parallel_p): Likewise.
9616 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9618 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9619 Don't print any bits outside the precision of the value.
9620 * wide-int.cc (test_printing): Add some new tests.
9622 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9624 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9625 supports -xbrace_comment option.
9626 * configure: Regenerate.
9627 * config.in: Regenerate.
9628 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9629 (ASM_CPU_SPEC): Use it.
9631 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9633 * target.def (static_rtx_alignment): New hook.
9634 * targhooks.h (default_static_rtx_alignment): Declare.
9635 * targhooks.c (default_static_rtx_alignment): New function.
9636 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9637 * doc/tm.texi: Regenerate.
9638 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9639 instead of targetm.constant_alignment. Remove call to
9641 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9642 (cris_preferred_mininum_alignment): New function, split out from...
9643 (cris_constant_alignment): ...here.
9644 (cris_static_rtx_alignment): New function.
9645 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9647 (ix86_constant_alignment): ...here.
9648 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9649 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9650 (mmix_static_rtx_alignment): New function.
9651 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9652 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9654 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9657 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9658 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9660 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9662 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9663 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9664 mark as a sign-extending load.
9665 (local_pic_load_u): Define.
9667 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9670 * fold-const.c (operand_equal_for_comparison_p): Also return true
9671 if ARG0 is a simple variant of ARG1 with narrower precision.
9672 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9674 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9676 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9678 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9679 scatter_static, scatter_per_elt.
9680 * x86-tune-costs.h: Add new cost entries.
9682 2017-10-25 Richard Biener <rguenther@suse.de>
9684 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9685 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9686 class pass_fre): Move to ...
9687 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9688 class pass_fre): ... here and adjust for statistics.
9690 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9693 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9694 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9695 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9696 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9699 2017-10-25 Richard Biener <rguenther@suse.de>
9701 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9702 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9703 eliminate_push_avail, eliminate_insert): Move inside...
9704 (class eliminate_dom_walker): ... this class in preparation
9706 (fini_eliminate): Remove by merging with ...
9707 (eliminate): ... this function. Adjust for class changes.
9708 (pass_pre::execute): Remove fini_eliminate call.
9709 (pass_fre::execute): Likewise.
9711 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9714 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9715 (VPERMI2, VPERMI2I): New mode iterators.
9716 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9717 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9719 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9720 mode iterator. Remove 3 old define_insn patterns.
9721 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9722 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9723 VPERMI2 mode iterator, remove the other two expanders.
9724 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9725 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9726 instructions, remove the other two patterns.
9727 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9728 mode iterator, remove the other two patterns.
9729 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9730 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9731 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9732 and adjust argument order accordingly.
9733 (ix86_expand_vec_perm): Adjust caller.
9734 (expand_vec_perm_1): Likewise.
9735 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9736 (expand_vec_perm_vpermt2_vpshub2): ... this.
9737 (ix86_expand_vec_perm_const_1): Adjust caller.
9738 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9741 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9742 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9743 (vec_shl_<mode>): Remove unused expander.
9744 (avx512bw_<shift_insn><mode>3): New define_insn.
9745 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9746 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9748 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9751 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9754 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9756 PR rtl-optimization/82396
9757 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9758 (autopref_multipass_init): Simplify initialization.
9759 (autopref_rank_data): Simplify sort order.
9760 * gcc/sched-int.h (autopref_multipass_data_): Remove
9761 multi_mem_insn_p, min_offset and max_offset.
9763 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9766 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9767 Check special value of flag_omit_frame_pointer.
9768 (aarch64_can_eliminate): Likewise.
9769 (aarch64_override_options_after_change_1): Simplify handling of
9770 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9772 2017-10-24 Richard Biener <rguenther@suse.de>
9774 PR tree-optimization/82697
9775 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9776 zero for conditional load and unconditional store.
9778 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9780 * doc/install.texi: Document bootstrap-cet.
9782 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9785 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9786 ENDBR instruction at function entrance if function is only
9789 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9792 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9793 patterns to better describe from which operation the CF is computed.
9794 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9795 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9796 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9797 is 0, use _0 suffixed expanders instead of emitting a comparison
9800 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9802 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9803 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9805 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9808 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9809 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9810 * loop-iv.c (iv_get_reaching_def): Likewise.
9811 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9812 variable is promoted and the partition contains undefined values.
9814 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9816 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9818 (nios2_address_cost): Define.
9819 (nios2_legitimize_address): Recognize (exp + constant) directly.
9820 (TARGET_ADDRESS_COST): Define.
9822 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9824 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9825 (nios2_symbolic_memory_operand_p): Declare.
9826 (nios2_split_large_constant): Declare.
9827 (nios2_split_symbolic_memory_operand): Declare.
9828 * config/nios2/nios2.c: Adjust includes.
9829 (nios2_symbolic_constant_allowed): New.
9830 (nios2_symbolic_constant_p): New.
9831 (nios2_plus_symbolic_constant_p): New.
9832 (nios2_valid_addr_expr_p): Recognize addresses involving
9834 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9835 (nios2_symbolic_memory_operand_p): New.
9836 (nios2_large_constant_p): New.
9837 (nios2_split_large_constant): New.
9838 (nios2_split_plus_large_constant): New.
9839 (nios2_split_symbolic_memory_operand): New.
9840 (nios2_legitimize_address): Code refactoring. Handle addresses
9841 involving symbolic constants.
9842 (nios2_emit_move_sequence): Likewise.
9843 (nios2_print_operand): Improve error output.
9844 (nios2_print_operand_address): Handle LO_SUM.
9845 (nios2_cdx_narrow_form_p): Likewise.
9846 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9847 operands involving symbolic constants.
9848 (movhi_internal, movsi_internal): Likewise.
9849 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9850 (extendhisi2, extendqi<mode>2): Likewise.
9852 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9854 * tree-pass.h (PROP_rtl_split_insns): Define.
9855 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9857 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9859 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9861 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9864 * target.def (const_not_ok_for_debug_p): Default to
9865 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9866 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9867 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9868 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9869 which targetm.const_not_ok_for_debug_p returned true.
9870 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9872 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9874 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9875 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9876 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9878 (ix86_const_not_ok_for_debug_p): New function.
9879 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9880 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9882 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9885 * system.h: Conditionally include "unique-ptr.h" if
9886 INCLUDE_UNIQUE_PTR is defined.
9887 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9888 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9890 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9892 * config/rl78/rl78.md: New define_expand "subdi3".
9894 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9897 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9898 DF_REF_INSN if DF_REF_INSN_INFO is false.
9900 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9902 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9903 xmm_move instead of sse_move.
9904 (sse_store_index): New function.
9905 (ix86_register_move_cost): Be more sensible about mismatch stall;
9906 model AVX moves correctly; make difference between sse->integer and
9908 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9909 moves; make difference between SSE and AVX.
9910 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9911 and zmm_move. Increase size of sse load and store tables;
9912 add unaligned load and store tables; add ssemmx_to_integer.
9913 * x86-tune-costs.h: Update all entries according to real
9914 move latencies from Agner Fog's manual and chip documentation.
9916 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9919 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9920 * config/i386/constraints.md (Wf): New constraint.
9921 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9922 (cmp<dwi>_doubleword): Removed.
9923 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9924 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9925 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9926 expand with cmp<dwi>_doubleword. For LTU and GEU use
9927 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9929 * common.opt (gcolumn-info): Enable by default.
9930 * doc/invoke.texi (gcolumn-info): Document new default.
9932 2017-10-23 Richard Biener <rguenther@suse.de>
9934 PR tree-optimization/82672
9935 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9936 Fold the stmt if we propagated into it.
9938 2017-10-23 Richard Biener <rguenther@suse.de>
9940 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9941 (bitmap_remove_expr_from_set): ... this. All callers call this
9942 for non-constant values.
9943 (bitmap_set_subtract): Rename to...
9944 (bitmap_set_subtract_expressions): ... this. Adjust and
9946 (bitmap_set_contains_value): Remove superfluous check.
9947 (bitmap_set_replace_value): Inline into single caller ...
9948 (bitmap_value_replace_in_set): ... here and simplify.
9949 (dependent_clean): Merge into ...
9950 (clean): ... this using an overload. Adjust.
9951 (prune_clobbered_mems): Adjust.
9952 (compute_antic_aux): Likewise.
9953 (compute_partial_antic_aux): Likewise.
9955 2017-10-23 Richard Biener <rguenther@suse.de>
9957 PR tree-optimization/82129
9959 2017-08-01 Richard Biener <rguenther@suse.de>
9961 PR tree-optimization/81181
9962 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9963 (compute_antic): ... end of iteration here.
9965 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9967 * target.def (starting_frame_offset): New hook.
9968 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9969 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9970 * doc/tm.texi.in: Regenerate.
9971 * hooks.h (hook_hwi_void_0): Declare.
9972 * hooks.c (hook_hwi_void_0): New function.
9973 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9974 STARTING_FRAME_OFFSET.
9975 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9976 * reload1.c (reload): Likewise.
9977 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9978 instead of STARTING_FRAME_OFFSET.
9979 * function.c (try_fit_stack_local): Likewise.
9980 (assign_stack_local_1): Likewise
9981 (instantiate_virtual_regs): Likewise.
9982 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9983 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9984 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9985 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9986 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9987 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9988 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9989 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9990 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9991 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9992 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9993 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9994 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9995 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9996 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9997 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9998 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9999 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10000 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10001 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10002 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10003 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10004 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10005 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10006 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10007 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10008 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10009 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10010 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10011 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10012 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10013 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10014 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10015 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10016 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10017 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10018 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10019 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10020 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10021 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10022 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10023 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10024 return a HOST_WIDE_INT.
10025 (avr_builtin_setjmp_frame_value): Use it instead of
10026 STARTING_FRAME_OFFSET.
10027 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10028 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10029 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10031 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10032 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10033 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10034 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10035 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10036 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10037 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10038 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10039 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10040 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10041 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10042 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10044 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10045 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10046 * config/mips/mips.c (mips_compute_frame_info): Refer to
10047 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10048 (mips_starting_frame_offset): New function.
10049 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10050 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10051 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10052 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10053 and return a HOST_WIDE_INT.
10054 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10055 (mmix_initial_elimination_offset): Refer to
10056 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10057 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10058 * config/pa/pa.c (pa_starting_frame_offset): New function.
10059 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10060 (pa_expand_prologue): Likewise.
10061 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10062 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10063 !FRAME_GROWS_DOWNWARD handling to...
10064 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10065 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10066 !FRAME_GROWS_DOWNWARD handling to...
10067 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10068 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10069 !FRAME_GROWS_DOWNWARD handling to...
10070 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10071 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10073 (rs6000_starting_frame_offset): New function.
10074 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10075 !FRAME_GROWS_DOWNWARD handling to...
10076 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10077 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10078 !FRAME_GROWS_DOWNWARD handling to...
10079 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10080 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10081 !FRAME_GROWS_DOWNWARD handling to...
10082 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10083 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10084 (rs6000_starting_frame_offset): New function.
10085 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10086 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10087 * config/vax/vax.c (vax_starting_frame_offset): New function.
10088 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10089 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10090 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10091 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10092 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10093 * system.h (STARTING_FRAME_OFFSET): Poison.
10095 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10097 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10098 SCALAR_TYPE_MODE instead of TYPE_MODE.
10100 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10101 Alan Hayward <alan.hayward@arm.com>
10102 David Sherwood <david.sherwood@arm.com>
10104 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10106 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10107 Alan Hayward <alan.hayward@arm.com>
10108 David Sherwood <david.sherwood@arm.com>
10110 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10112 2017-10-23 Richard Biener <rguenther@suse.de>
10114 PR tree-optimization/82129
10115 * tree-ssa-pre.c (bitmap_set_and): Remove.
10116 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10117 canonicalizing expressions in the set to those with lowest
10118 ID rather than taking that from the first edge.
10120 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10122 * combine.c (rtx_equal_for_field_assignment_p): Use
10123 byte_lowpart_offset.
10125 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10126 Alan Hayward <alan.hayward@arm.com>
10127 David Sherwood <david.sherwood@arm.com>
10129 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10130 to a SUBREG_PROMOTED_VAR.
10132 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10133 Alan Hayward <alan.hayward@arm.com>
10134 David Sherwood <david.sherwood@arm.com>
10136 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10137 (expand_debug_source_expr): Likewise.
10138 * combine.c (combine_simplify_rtx): Likewise.
10139 * cse.c (fold_rtx): Likewise.
10140 * optabs.c (expand_float): Likewise.
10141 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10142 (simplify_binary_operation_1): Likewise.
10144 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10145 Alan Hayward <alan.hayward@arm.com>
10146 David Sherwood <david.sherwood@arm.com>
10148 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10149 (record_promoted_value): Likewise.
10150 * expr.c (expand_expr_real_2): Likewise.
10151 * ree.c (update_reg_equal_equiv_notes): Likewise.
10152 (combine_set_extension): Likewise.
10153 * rtlanal.c (low_bitmask_len): Likewise.
10154 * simplify-rtx.c (neg_const_int): Likewise.
10155 (simplify_binary_operation_1): Likewise.
10157 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10158 Alan Hayward <alan.hayward@arm.com>
10159 David Sherwood <david.sherwood@arm.com>
10161 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10162 * regcprop.c (maybe_mode_change): Likewise.
10163 * reload1.c (alter_reg): Likewise.
10165 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10167 * inchash.h (inchash::hash::add_wide_int): New function.
10168 * lto-streamer-out.c (hash_tree): Use it.
10170 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10172 * inchash.h (inchash::hash::add_wide_int): Rename to...
10173 (inchash::hash::add_hwi): ...this.
10174 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10175 (polymorphic_call_target_hasher::hash): Likewise.
10176 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10177 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10178 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10179 * lto-streamer-out.c (hash_tree): Likewise.
10180 * optc-save-gen.awk: Likewise.
10181 * tree.c (add_expr): Likewise.
10183 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10186 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10187 for ordered inequality comparisons even with TARGET_IEEE_FP.
10189 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10192 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10193 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10194 Expand with cmp<dwi>_doubleword.
10196 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10198 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10199 List CET intrinsics.
10200 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10201 specific to -fcf-protection option.
10203 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10205 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10206 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10207 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10208 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10209 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10210 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10211 (extra_objs): Add cet.o for Linux/x86 targets.
10212 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10213 * config/i386/cet.c: New file.
10214 * config/i386/cetintrin.h: Likewise.
10215 * config/i386/t-cet: Likewise.
10216 * config/i386/cpuid.h (bit_SHSTK): New.
10217 (bit_IBT): Likewise.
10218 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10219 pass IBT and SHSTK bits.
10220 * config/i386/i386-builtin-types.def
10221 (VOID_FTYPE_UNSIGNED_PVOID): New.
10222 (VOID_FTYPE_UINT64_PVOID): Likewise.
10223 * config/i386/i386-builtin.def: Add CET intrinsics.
10224 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10225 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10226 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10227 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10229 * config/i386/i386.c (rest_of_insert_endbranch): New.
10230 (pass_data_insert_endbranch): Likewise.
10231 (pass_insert_endbranch): Likewise.
10232 (make_pass_insert_endbranch): Likewise.
10233 (ix86_notrack_prefixed_insn_p): Likewise.
10234 (ix86_target_string): Add -mibt, -mshstk flags.
10235 (ix86_option_override_internal): Add flag_cf_protection
10237 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10238 (ix86_print_operand): Add 'notrack' prefix output.
10239 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10240 (ix86_expand_builtin): Expand CET intrinsics.
10241 (x86_output_mi_thunk): Add 'endbranch' instruction.
10242 * config/i386/i386.h (TARGET_IBT): New.
10243 (TARGET_IBT_P): Likewise.
10244 (TARGET_SHSTK): Likewise.
10245 (TARGET_SHSTK_P): Likewise.
10246 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10247 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10248 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10249 (builtin_setjmp_setup): New pattern.
10250 (builtin_longjmp): Likewise.
10251 (rdssp<mode>): Likewise.
10252 (incssp<mode>): Likewise.
10253 (saveprevssp): Likewise.
10254 (rstorssp): Likewise.
10255 (wrss<mode>): Likewise.
10256 (wruss<mode>): Likewise.
10257 (setssbsy): Likewise.
10258 (clrssbsy): Likewise.
10259 (nop_endbr): Likewise.
10260 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10262 * config/i386/immintrin.h: Include <cetintrin.h>.
10263 * config/i386/linux-common.h
10264 (file_end_indicate_exec_stack_and_cet): New prototype.
10265 (TARGET_ASM_FILE_END): New.
10267 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10269 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10270 latencies instead of having separate table; make difference between
10271 integer and float costs.
10272 * i386.h (processor_costs): Remove scalar_stmt_cost,
10273 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10274 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10276 * x86-tune-costs.h: Remove entries which has been removed in
10277 procesor_costs from all tables; make cond_taken_branch_cost
10278 and cond_not_taken_branch_cost COST_N_INSNS based.
10280 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10282 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10284 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10286 * config/i386/i386.md (isa): Remove fma_avx512f.
10287 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10288 <avx512>_fmadd_<mode>_mask3<round_name>,
10289 <avx512>_fmsub_<mode>_mask<round_name>,
10290 <avx512>_fmsub_<mode>_mask3<round_name>,
10291 <avx512>_fnmadd_<mode>_mask<round_name>,
10292 <avx512>_fnmadd_<mode>_mask3<round_name>,
10293 <avx512>_fnmsub_<mode>_mask<round_name>,
10294 <avx512>_fnmsub_<mode>_mask3<round_name>,
10295 <avx512>_fmaddsub_<mode>_mask<round_name>,
10296 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10297 <avx512>_fmsubadd_<mode>_mask<round_name>,
10298 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10299 (*vec_widen_umult_even_v16si<mask_name>,
10300 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10301 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10303 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10305 * extend.texi: Add 'nocf_check' documentation.
10306 * gimple.texi: Add second parameter to
10307 gimple_build_call_from_tree.
10308 * invoke.texi: Add -fcf-protection documentation.
10309 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10311 2017-10-20 Richard Biener <rguenther@suse.de>
10313 PR tree-optimization/82473
10314 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10315 the largest input type.
10317 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10319 * c-attribs.c (handle_nocf_check_attribute): New function.
10320 (c_common_attribute_table): Add 'nocf_check' handling.
10321 * gimple-parser.c: Add second argument NULL to
10322 gimple_build_call_from_tree.
10323 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10324 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10326 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10327 * common.opt: Add fcf-protection flag.
10328 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10329 * flag-types.h: Add enum cf_protection_level.
10330 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10331 Add 'nocf_check' attribute propagation to gimple call.
10332 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10333 (gimple_build_call_from_tree): Update prototype.
10334 (gimple_call_nocf_check_p): New function.
10335 (gimple_call_set_nocf_check): Likewise.
10336 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10337 * ipa-icf.c: Add nocf_check attribute in statement hash.
10338 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10339 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10340 * toplev.c (process_options): Add flag_cf_protection handling.
10342 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10344 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10346 2017-10-20 Richard Biener <rguenther@suse.de>
10348 PR tree-optimization/82603
10349 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10350 remove false predicated stores.
10352 2017-10-20 Richard Biener <rguenther@suse.de>
10354 * graphite-isl-ast-to-gimple.c
10355 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10356 Remove return value and simplify, dump copied stmt after lhs
10358 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10359 Reduce dump verbosity.
10360 (gsi_insert_earliest): Likewise.
10361 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10362 * graphite.c (print_global_statistics): Adjust dumping.
10363 (print_graphite_scop_statistics): Likewise.
10364 (print_graphite_statistics): Do not dump loops here.
10365 (graphite_transform_loops): But here.
10367 2017-10-20 Nicolas Roche <roche@adacore.com>
10369 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10370 * configure: Regenerate.
10372 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10375 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10376 functions when optimizing replace GIMPLE_RETURN stmts with
10377 calls to __builtin_unreachable ().
10380 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10381 for -fsanitize=thread link of executables.
10382 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10383 link of executables.
10386 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10387 New mode iterators.
10388 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10389 define_insns for logical vector shifts to use VI248_AVX512BW
10390 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10391 condition, useless isa and prefix attributes. Change the first
10392 2 of these define_insns to ...
10393 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10394 define_insn for avx512vl.
10395 (<shift_insn><mode>3): ... and this, new define_insn without
10396 masking for non-avx512vl.
10399 * config/i386/sse.md (*andnot<mode>3,
10400 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10401 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10402 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10403 not applied use empty suffix even for TARGET_AVX512VL.
10404 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10405 is applied, supply evex,evex or evex,evex,evex instead of just
10408 2017-10-20 Julia Koval <julia.koval@intel.com>
10410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10411 (OPTION_MASK_ISA_GFNI_UNSET): New.
10412 (ix86_handle_option): Handle OPT_mgfni.
10413 * config/i386/cpuid.h (bit_GFNI): New.
10414 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10415 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10416 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10417 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10418 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10419 * config/i386/i386.opt: Add mgfni.
10421 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10423 * config/msp430/msp430.c (msp430_option_override): Disable
10424 -fdelete-null-pointer-checks.
10425 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10427 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10429 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10430 of x87 and SSE instructions.
10432 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10434 * asan.c (create_cond_insert_point): Do not update edge count.
10435 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10436 (afdo_propagate_circuit): Likewise.
10437 (afdo_calculate_branch_prob): Likewise.
10438 (afdo_annotate_cfg): Likewise.
10439 * basic-block.h (struct edge_def): Remove count.
10440 (edge_def::count): New accessor.
10441 * bb-reorder.c (rotate_loop): Update.
10442 (find_traces_1_round): Update.
10443 (connect_traces): Update.
10444 (sanitize_hot_paths): Update.
10445 * cfg.c (unchecked_make_edge): Update.
10446 (make_single_succ_edge): Update.
10447 (check_bb_profile): Update.
10448 (dump_edge_info): Update.
10449 (update_bb_profile_for_threading): Update.
10450 (scale_bbs_frequencies_int): Update.
10451 (scale_bbs_frequencies_gcov_type): Update.
10452 (scale_bbs_frequencies_profile_count): Update.
10453 (scale_bbs_frequencies): Update.
10454 * cfganal.c (connect_infinite_loops_to_exit): Update.
10455 * cfgbuild.c (compute_outgoing_frequencies): Update.
10456 (find_many_sub_basic_blocks): Update.
10457 * cfgcleanup.c (try_forward_edges): Update.
10458 (try_crossjump_to_edge): Update
10459 * cfgexpand.c (expand_gimple_cond): Update
10460 (expand_gimple_tailcall): Update
10461 (construct_exit_block): Update
10462 * cfghooks.c (verify_flow_info): Update
10463 (redirect_edge_succ_nodup): Update
10464 (split_edge): Update
10465 (make_forwarder_block): Update
10466 (duplicate_block): Update
10467 (account_profile_record): Update
10468 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10469 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10470 * cfgloopmanip.c (scale_loop_profile): Update.
10472 (lv_adjust_loop_entry_edge): Update.
10473 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10474 (force_nonfallthru_and_redirect): Update.
10475 (purge_dead_edges): Update.
10476 (rtl_flow_call_edges_add): Update.
10477 * cgraphunit.c (init_lowered_empty_function): Update.
10478 (cgraph_node::expand_thunk): Update.
10479 * gimple-pretty-print.c (dump_probability): Update.
10480 (dump_edge_probability): Update.
10481 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10482 * haifa-sched.c (sched_create_recovery_edges): Update.
10483 * hsa-gen.c (convert_switch_statements): Update.
10484 * ifcvt.c (dead_or_predicable): Update.
10485 * ipa-inline-transform.c (inline_transform): Update.
10486 * ipa-split.c (split_function): Update.
10487 * ipa-utils.c (ipa_merge_profiles): Update.
10488 * loop-doloop.c (add_test): Update.
10489 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10490 * lto-streamer-in.c (input_cfg): Update.
10491 (input_function): Update.
10492 * lto-streamer-out.c (output_cfg): Update.
10493 * modulo-sched.c (sms_schedule): Update.
10494 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10495 * predict.c (maybe_hot_edge_p): Update.
10496 (unlikely_executed_edge_p): Update.
10497 (probably_never_executed_edge_p): Update.
10498 (dump_prediction): Update.
10499 (drop_profile): Update.
10500 (propagate_unlikely_bbs_forward): Update.
10501 (determine_unlikely_bbs): Update.
10502 (force_edge_cold): Update.
10503 * profile.c (compute_branch_probabilities): Update.
10504 * reg-stack.c (better_edge): Update.
10505 * shrink-wrap.c (handle_simple_exit): Update.
10506 * tracer.c (better_p): Update.
10507 * trans-mem.c (expand_transaction): Update.
10508 (split_bb_make_tm_edge): Update.
10509 * tree-call-cdce.c: Update.
10510 * tree-cfg.c (gimple_find_sub_bbs): Update.
10511 (gimple_split_edge): Update.
10512 (gimple_duplicate_sese_region): Update.
10513 (gimple_duplicate_sese_tail): Update.
10514 (gimple_flow_call_edges_add): Update.
10515 (insert_cond_bb): Update.
10516 (execute_fixup_cfg): Update.
10517 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10518 * tree-complex.c (expand_complex_div_wide): Update.
10519 * tree-eh.c (lower_resx): Update.
10520 (unsplit_eh): Update.
10521 (cleanup_empty_eh_move_lp): Update.
10522 * tree-inline.c (copy_edges_for_bb): Update.
10523 (freqs_to_counts): Update.
10524 (copy_cfg_body): Update.
10525 * tree-ssa-dce.c (remove_dead_stmt): Update.
10526 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10527 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10528 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10529 (unloop_loops): Update.
10530 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10531 * tree-ssa-loop-split.c (connect_loops): Update.
10532 (split_loop): Update.
10533 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10534 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10535 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10536 * tree-ssa-reassoc.c (branch_fixup): Update.
10537 * tree-ssa-tail-merge.c (replace_block_by): Update.
10538 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10539 (compute_path_counts): Update.
10540 (update_profile): Update.
10541 (recompute_probabilities): Update.
10542 (update_joiner_offpath_counts): Update.
10543 (estimated_freqs_path): Update.
10544 (freqs_to_counts_path): Update.
10545 (clear_counts_path): Update.
10546 (ssa_fix_duplicate_block_edges): Update.
10547 (duplicate_thread_path): Update.
10548 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10549 (case_bit_test_cmp): Update.
10550 (collect_switch_conv_info): Update.
10551 (gen_inbound_check): Update.
10552 (do_jump_if_equal): Update.
10553 (emit_cmp_and_jump_insns): Update.
10554 * tree-tailcall.c (decrease_profile): Update.
10555 (eliminate_tail_call): Update.
10556 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10557 (vect_do_peeling): Update.
10558 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10559 * ubsan.c (ubsan_expand_null_ifn): Update.
10560 (ubsan_expand_ptr_ifn): Update.
10561 * value-prof.c (gimple_divmod_fixed_value): Update.
10562 (gimple_mod_pow2): Update.
10563 (gimple_mod_subtract): Update.
10564 (gimple_ic): Update.
10565 (gimple_stringop_fixed_value): Update.
10567 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10570 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10572 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10574 PR rtl-optimization/82395
10575 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10576 based on non_spilled_static_chain_regno_p.
10578 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10580 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10581 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10582 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10584 2017-10-19 Martin Sebor <msebor@redhat.com>
10586 PR tree-optimization/82596
10587 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10589 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10591 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10592 (get_mem_refs_of_builtin_call): Likewise.
10593 * builtins.c (expand_builtin_apply): Adjust call to
10594 allocate_dynamic_stack_space.
10595 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10596 the third argument to allocate_dynamic_stack_space, otherwise -1.
10597 (expand_builtin): Deal with all alloca variants.
10598 (is_inexpensive_builtin): Likewise.
10599 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10600 * calls.c (special_function_p): Deal with all alloca variants.
10601 (initialize_argument_information): Adjust call to
10602 allocate_dynamic_stack_space.
10603 (expand_call): Likewise.
10604 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10605 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10606 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10607 use it for the stack usage computation.
10608 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10609 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10610 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10611 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10612 (in_loop_p): Remove first argument and useless check.
10613 (pass_walloca::execute): Remove useless test and adjust call to above.
10614 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10615 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10616 (gimplify_call_expr): Deal with all alloca variants.
10617 * hsa-gen.c (gen_hsa_alloca): Likewise.
10618 (gen_hsa_insns_for_call): Likewise.
10619 * ipa-pure-const.c (special_builtin_state): Likewise.
10620 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10621 * tree-object-size.c (alloc_object_size): Likewise.
10622 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10623 (call_may_clobber_ref_p_1): Likewise.
10624 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10625 (ccp_fold_stmt): Likewise.
10626 (optimize_stack_restore): Likewise.
10627 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10628 (mark_all_reaching_defs_necessary_1): Likewise.
10629 (propagate_necessity): Likewise.
10630 (eliminate_unnecessary_stmts): Likewise.
10631 * tree.c (build_common_builtin_nodes): Build
10632 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10633 (build_alloca_call_expr): New function.
10634 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10635 (CASE_BUILT_IN_ALLOCA): Likewise.
10636 (build_alloca_call_expr): Declare.
10637 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10639 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10642 * dwarf2out.c (new_die_raw): New static inline function.
10643 (new_die): Use it to create the DIE.
10644 (add_AT_external_die_ref): Likewise.
10645 (clone_die): Likewise.
10646 (clone_as_declaration): Likewise.
10647 (dwarf2out_vms_debug_main_pointer): Likewise.
10648 (base_type_die): Likewise. Remove early return for corner cases.
10649 Do not call add_pubtype on the DIE here.
10650 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10651 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10652 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10653 native order exists for base types, attach the DIE manually and call
10654 add_pubtype on it. Do not equate a reverse order DIE to the type.
10656 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10658 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10659 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10661 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10663 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10665 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10666 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10667 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10668 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10669 * opts.c (sanitizer_opts): Add builtin.
10670 * ubsan.c (instrument_builtin): New function.
10671 (pass_ubsan::execute): Call it.
10672 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10673 * doc/invoke.texi: Document -fsanitize=builtin.
10675 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10676 builtins, store max (log2 (align), 0) into uchar field instead of
10677 align into uptr field.
10678 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10679 store uchar 0 field instead of uptr 0 field.
10680 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10681 instead of passing one address of struct with 2 locations pass
10682 two addresses of structs with 1 location each.
10683 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10684 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10685 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10686 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10687 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10688 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10689 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10690 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10692 2017-10-19 Martin Liska <mliska@suse.cz>
10695 * file-find.c (remove_prefix): Remove.
10696 * file-find.h (remove_prefix): Likewise.
10697 * gcc-ar.c: Remove smartness of lookup.
10699 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10701 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10702 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10703 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10705 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10708 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10709 (setcc + and to xor + setcc): New peephole2.
10711 2017-10-19 Tom de Vries <tom@codesourcery.com>
10713 * doc/sourcebuild.texi (Test Directives, Variants of
10714 dg-require-support): Add dg-require-stack-size.
10716 2017-10-19 Martin Liska <mliska@suse.cz>
10719 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10720 that have a large alignment.
10721 (gimplify_target_expr): Likewise.
10723 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10725 PR rtl-optimization/82602
10726 * ira.c (rtx_moveable_p): Return false for volatile asm.
10728 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10731 * config/i386/i386-modes.def (CCGZ): New CC mode.
10732 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10733 * config/i386/predicates.md (ix86_comparison_operator):
10735 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10736 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10737 with double-word subtraction.
10738 (put_condition_code): Handle CCGZmode.
10740 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10742 * wide-int.cc (debug (const wide_int &)): New.
10743 (debug (const wide_int *)): New.
10744 (debug (const widest_int &)): New.
10745 (debug (const widest_int *)): New.
10747 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10749 PR middle-end/82556
10750 * lra-constraints.c (curr_insn_transform): Use non-input operand
10751 instead of output one for matched reload.
10753 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10755 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10756 (tree-ssa-loop-ivopts.h): New header file.
10757 (struct builtin_info): New fields.
10758 (classify_builtin_1): Compute and record base and offset parts for
10759 memset builtin partition by calling strip_offset.
10760 (offset_cmp, fuse_memset_builtins): New functions.
10761 (finalize_partitions): Fuse adjacent memset partitions by calling
10763 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10764 Expose the interface.
10765 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10767 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10769 PR tree-optimization/82574
10770 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10771 that data reference must be executed exactly once per iteration
10772 against the outermost loop in nest.
10773 (classify_partition): Update call to above function.
10775 2017-10-18 Richard Biener <rguenther@suse.de>
10777 PR tree-optimization/82591
10778 * graphite.c (graphite_transform_loops): Move code gen message
10780 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10781 Here. Handle scop_to_isl_ast failing.
10782 (scop_to_isl_ast): Limit the number of ISL operations.
10784 2017-10-18 Richard Biener <rguenther@suse.de>
10786 * graphite-isl-ast-to-gimple.c
10787 (translate_isl_ast_to_gimple::set_rename): Simplify.
10788 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10789 (graphite_copy_stmts_from_block): ... here.
10790 (copy_bb_and_scalar_dependences): Simplify.
10791 (add_parameters_to_ivs_params): Canonicalize.
10792 (generate_entry_out_of_ssa_copies): Simplify.
10793 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10794 by passing in ISL dimension.
10795 (parameter_index_in_region_1): Rename to ...
10796 (parameter_index_in_region): ... this.
10797 (extract_affine): Adjust assert, pass down parameter index.
10798 (add_param_constraints): Use range-info when available.
10799 (build_scop_context): Adjust.
10800 * sese.c (new_sese_info): Adjust.
10801 (free_sese_info): Likewise.
10802 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10803 Remove unused typedefs.
10804 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10806 2017-10-18 Martin Liska <mliska@suse.cz>
10808 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10810 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10812 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10813 (s390_sched_init): Do not reset s390_sched_state if we entered the
10814 current basic block via a fallthru edge and all others are unlikely.
10816 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10818 * config/s390/s390.c (NUM_SIDES): New variable.
10819 (LONGRUNNING_THRESHOLD): New variable.
10820 (LATENCY_FACTOR): New variable.
10821 (s390_sched_score): Decrease score for long-running instructions on
10823 (s390_sched_variable_issue): Perform bookkeeping for long-running
10826 2017-10-18 Richard Biener <rguenther@suse.de>
10828 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10829 Simplify with removal of the parameter rename map.
10830 (set_rename): Likewise.
10831 (should_copy_to_new_region): Likewise.
10832 (graphite_copy_stmts_from_block): Likewise.
10833 (copy_bb_and_scalar_dependences): Remove initialization of
10834 unused copied_bb_map.
10835 (copy_def): Remove.
10836 (copy_internal_parameters): Likewise.
10837 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10838 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10839 Use INTEGRAL_TYPE_P.
10840 (parameter_index_in_region_1): Rename to ...
10841 (assign_parameter_index_in_region): ... this. Assert we have
10842 a parameter we handle.
10843 (scan_tree_for_params): Adjust.
10844 * sese.h (parameter_rename_map_t): Remove.
10845 (struct sese_info_t): Remove unused parameter_rename_map and
10846 copied_bb_map members.
10847 * sese.c (new_sese_info): Adjust.
10848 (free_sese_info): Likewise.
10850 2017-10-18 Martin Liska <mliska@suse.cz>
10853 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10854 on an abnormal edge.
10856 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10858 * doc/invoke.texi (ffunction-sections and fdata-sections):
10861 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10863 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10864 the use statement can throw internally.
10866 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10868 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10869 any RTX present on the RHS of a SET.
10870 * compare-elim.c (try_eliminate_compare): Restore comment.
10872 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10874 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10875 may be also called on tcc_type nodes.
10876 * langhooks.c (lhd_tree_size): Likewise.
10878 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10880 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10881 format_warning_at_substring.
10882 (maybe_warn): Convert source_range * param to a location_t. Pass
10883 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10884 (format_directive): Remove code to extract source_ranges and
10885 source_range * in favor of just a location_t.
10886 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10888 * substring-locations.c (format_warning_va): Convert
10889 source_range * param to a location_t.
10890 (format_warning_at_substring): Likewise.
10891 * substring-locations.h (format_warning_va): Likewise.
10892 (format_warning_at_substring): Likewise.
10894 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10896 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10898 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10899 and scatter/gather ops.
10901 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10902 vec_gather_load and vec_scatter_store.
10903 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10904 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10905 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10906 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10907 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10908 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10910 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10912 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10913 Detect FTST instruction and handle its register pops. Only pop
10914 second operand if can_pop_second_op is true.
10915 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10916 set can_pop_second_op to false in the compare_for_stack_reg call.
10918 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10919 output_fp_compare for stack register operands.
10920 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10921 instructions here. Do not emit stack register pops here. Assert
10922 that FCOMPP pops next to top stack register. Rewrite function.
10924 2017-10-17 Nathan Sidwell <nathan@acm.org>
10926 PR middle-end/82577
10927 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10928 use DECL_ASSEMBLER_NAME_RAW.
10930 PR middle-end/82546
10931 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10934 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10935 Wilco Dijkstra <wilco.dijkstra@arm.com>
10937 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10938 converstion to Pmode from the buf_addr.
10940 2017-10-17 Richard Biener <rguenther@suse.de>
10942 * graphite-dependences.c (scop_get_reads_and_writes): Change
10943 output parameters to references.
10945 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10947 PR 71026/tree-optimization
10948 * fold-const.c (distribute_real_division): Removed.
10949 (fold_binary_loc): Remove calls to distribute_real_divison.
10951 2017-10-17 Richard Biener <rguenther@suse.de>
10953 * graphite-scop-detection.c
10954 (scop_detection::stmt_has_simple_data_refs_p): Always use
10955 the full nest as region.
10956 (try_generate_gimple_bb): Likewise.
10957 * sese.c (scalar_evolution_in_region): Simplify now that
10958 SCEV can handle instantiation in regions.
10959 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10960 in the non-loop part of a function if requested.
10962 2017-10-17 Richard Biener <rguenther@suse.de>
10964 PR tree-optimization/82563
10965 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10967 (graphite_regenerate_ast_isl): Call it.
10968 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10970 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10972 PR tree-optimization/82549
10973 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10974 Formatting fixes. Instead of calling make_bit_field_ref with negative
10977 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10979 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10980 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10981 _mm_maskz_reduce_ss): New.
10982 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10983 __builtin_ia32_reducess_mask): Ditto..
10984 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10985 * config/i386/sse.md (reduces<mode>): Renamed to ...
10986 (reduces<mode><mask_scalar_name>): ... this.
10987 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10989 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10990 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10992 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10994 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10995 * selftest-run-tests.c (selftest::run_tests): Call
10996 selftest::unique_ptr_tests_cc_tests.
10997 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10998 * unique-ptr-tests.cc: New file.
11000 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11003 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11005 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11006 (make_hard_regno_born, make_hard_regno_dead): Update
11007 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11009 2017-10-16 Jeff Law <law@redhat.com>
11011 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11013 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11015 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11017 2017-10-16 Olivier Hainque <hainque@adacore.com>
11019 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11020 with_cpu if we were configured for an e500v2 target cpu name.
11022 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11024 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11025 * doc/invoke.texi: Document +nodsp as a valid extension for
11028 2017-10-16 Martin Liska <mliska@suse.cz>
11030 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11031 (test_set_range): Likewise.
11032 (test_range_functions): Rename to ...
11033 (test_bit_in_range): ... this.
11034 (sbitmap_c_tests): Add new test.
11036 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11038 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11040 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11041 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11043 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11045 * config/aarch64/aarch64-builtins.c
11046 (aarch64_types_quadopu_lane_qualifiers): New.
11047 (TYPES_QUADOPU_LANE): New.
11048 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11049 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11050 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11051 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11052 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11053 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11054 (Vdottype, DOTPROD): New.
11055 (sur): Add SDOT and UDOT.
11057 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11059 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11060 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11061 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11062 Add TARGET_DOTPROD.
11063 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11064 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11065 Enable TARGET_DOTPROD.
11066 (cortex-a75.cortex-a55): Likewise.
11067 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11069 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11071 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11072 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11074 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11076 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11077 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11078 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11079 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11080 * config/arm/types.md (neon_dot, neon_dot_q): New.
11081 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11083 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11085 * config/arm/arm.h (TARGET_DOTPROD): New.
11086 * config/arm/arm.c (arm_arch_dotprod): New.
11087 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11088 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11089 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11090 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11091 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11092 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11093 * doc/invoke.texi (armv8.2-a): Document dotprod
11095 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11097 * i386.c (ix86_vec_cost): New function.
11098 (ix86_rtx_costs): Handle vector operations better.
11099 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11100 * x86-tune-costs.h: Add new costs to all tables.
11102 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11104 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11106 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11107 divsd, sqrtss and sqrtsd
11108 * x86-tune-costs.h: Add new entries to all costs.
11109 (znver1_cost): Fix to match real instruction latencies.
11111 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11112 Michael Collison <michael.collison@arm.com>
11114 * compare-elim.c: Include emit-rtl.h.
11115 (can_merge_compare_into_arith): New function.
11116 (try_validate_parallel): Likewise.
11117 (try_merge_compare): Likewise.
11118 (try_eliminate_compare): Call the above when no previous clobber
11120 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11123 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11125 PR middle-end/62263
11126 PR middle-end/82498
11127 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11128 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11130 PR middle-end/62263
11131 PR middle-end/82498
11132 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11133 to be any operand_equal_p operands. For & (B - 1) require
11134 B to be power of 2. Recognize
11135 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11137 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11140 * optabs.c (expand_memory_blockage): Fix call of
11141 targetm.have_memory_blockage.
11143 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11146 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11147 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11148 objects to extra_objs instead of overwriting it.
11150 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11152 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11153 Use any_fp_register_operand as operand[3] predicate. Simplify
11154 equality test for operands[2] and operands[4] memory location.
11155 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11156 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11157 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11158 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11159 any_fp_register_operand as operand[1] predicate. Simplify
11160 equality test for operands[0] and operands[3] memory location.
11161 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11162 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11163 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11165 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11167 * target-insns.def: Add memory_blockage.
11168 * optabs.c (expand_memory_blockage): New function.
11169 (expand_asm_memory_barrier): Rename ...
11170 (expand_asm_memory_blockage): ... to this.
11171 (expand_mem_thread_fence): Call expand_memory_blockage
11172 instead of expand_asm_memory_barrier.
11173 (expand_mem_singnal_fence): Ditto.
11174 (expand_atomic_load): Ditto.
11175 (expand_atomic_store): Ditto.
11176 * doc/md.texi (Standard Pattern Names For Generation):
11177 Document memory_blockage instruction pattern.
11179 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11181 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11182 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11183 * config/rl78/rl78.md: New define_expand "adddi3".
11185 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11187 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11190 2017-10-13 Jeff Law <law@redhat.com>
11192 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11194 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11197 * internal-fn.c (expand_mul_overflow): If both operands have
11198 the same highpart of -1 or 0 and the topmost bit of lowpart
11199 is different, overflow is if res <= 0 rather than res < 0.
11201 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11203 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11204 TARGET_P9_VECTOR code for unaligned_load case.
11206 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11208 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11210 2017-10-13 Nathan Sidwell <nathan@acm.org>
11212 * tree-core.h (tree_contains_struct): Make bool.
11213 * tree.c (tree_contains_struct): Likewise.
11214 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11215 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11216 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11217 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11218 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11220 2017-10-13 Richard Biener <rguenther@suse.de>
11222 * graphite-isl-ast-to-gimple.c
11223 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11224 parameters and dominance check.
11225 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11226 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11227 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11228 Do not update SSA form here or do intermediate IL verification.
11229 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11230 (graphite_initialize): Remove check on the number of loops in
11231 the function and inline into graphite_transform_loops.
11232 (graphite_finalize): Inline into graphite_transform_loops.
11233 (graphite_transform_loops): Perform SSA update and IL verification
11235 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11237 2017-10-13 Richard Biener <rguenther@suse.de>
11239 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11240 graphite_expression_type_precision): Avoid global constructor
11242 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11243 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11244 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11245 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11246 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11247 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11249 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11252 * config/i386/i386.h (ix86_red_zone_size): New.
11253 * config/i386/i386.md (push peephole2s): Replace
11254 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11256 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * combine.c (can_change_dest_mode): Reject changes in
11261 REGMODE_NATURAL_SIZE.
11263 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11264 Alan Hayward <alan.hayward@arm.com>
11265 David Sherwood <david.sherwood@arm.com>
11267 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11268 (expand_debug_source_expr): Likewise.
11269 * combine.c (combine_simplify_rtx): Likewise.
11270 * cse.c (fold_rtx): Likewise.
11271 * fwprop.c (canonicalize_address): Likewise.
11272 * targhooks.c (default_shift_truncation_mask): Likewise.
11274 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11275 Alan Hayward <alan.hayward@arm.com>
11276 David Sherwood <david.sherwood@arm.com>
11278 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11279 (widened_mode): Likewise.
11280 (expand_unop): Likewise.
11281 * ree.c (transform_ifelse): Likewise.
11282 (merge_def_and_ext): Likewise.
11283 (combine_reaching_defs): Likewise.
11284 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11286 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11287 Alan Hayward <alan.hayward@arm.com>
11288 David Sherwood <david.sherwood@arm.com>
11290 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11291 * combine.c (gen_lowpart_for_combine): Likewise.
11292 * dwarf2out.c (rtl_for_decl_location): Likewise.
11293 * final.c (alter_subreg): Likewise.
11294 * rtlhooks.c (gen_lowpart_general): Likewise.
11295 (gen_lowpart_if_possible): Likewise.
11297 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11298 Alan Hayward <alan.hayward@arm.com>
11299 David Sherwood <david.sherwood@arm.com>
11301 * calls.c (expand_call): Use subreg_lowpart_offset.
11302 * cse.c (cse_insn): Likewise.
11303 * regcprop.c (copy_value): Likewise.
11304 (copyprop_hardreg_forward_1): Likewise.
11306 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11309 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11310 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11311 =Q constraints to +Q and into insn condition add check
11312 that operands[0] and operands[1] are equal.
11313 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11314 =Q constraints to +Q and into insn condition add check
11315 that operands[0] is equal to either operands[1] or operands[2].
11318 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11319 instead of handling MINUS_EXPR twice (once for each argument),
11320 canonicalize operand order and handle just once, use rtype where
11321 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11324 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11325 any values of __C while still being pattern recognizable as a simple
11326 rotate instruction.
11328 2017-10-13 Richard Biener <rguenther@suse.de>
11330 PR tree-optimization/82451
11332 2017-10-02 Richard Biener <rguenther@suse.de>
11334 PR tree-optimization/82355
11335 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11336 a mapping for the enclosing loop but avoid generating one for
11337 the loop tree root.
11338 (copy_bb_and_scalar_dependences): Remove premature codegen
11339 error on PHIs in blocks duplicated into multiple places.
11340 * graphite-scop-detection.c
11341 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11342 in the region use it as loop and nest to analyze the DR in.
11343 (try_generate_gimple_bb): Likewise.
11344 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11345 (add_loop_constraints): For blocks in a loop not in the region
11346 create a dimension with a single iteration.
11347 * sese.h (gbb_loop_at_index): Remove assert.
11349 * cfgloop.c (loop_preheader_edge): For the loop tree root
11350 return the single successor of the entry block.
11351 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11352 Reset the SCEV hashtable and niters.
11353 * graphite-scop-detection.c
11354 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11355 assert that we only have POLYNOMIAL_CHREC that vary in loops
11356 contained in the region.
11357 (scop_detection::graphite_can_represent_expr): Adjust.
11358 (scop_detection::stmt_has_simple_data_refs_p): For loops
11359 not in the region set loop to NULL. The nest is now the
11360 entry edge to the region.
11361 (try_generate_gimple_bb): Likewise.
11362 * sese.c (scalar_evolution_in_region): Adjust for
11363 instantiate_scev change.
11364 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11365 Make nest parameter the edge into the region.
11366 (create_data_ref): Likewise.
11367 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11368 entry edge into a region and adjust instantiate_scev calls.
11369 (create_data_ref): Likewise.
11370 (graphite_find_data_references_in_stmt): Likewise.
11371 (find_data_references_in_stmt): Pass the loop preheader edge
11372 from the nest argument.
11373 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11374 parameter the edge into the region.
11375 (instantiate_parameters): Use the loop preheader edge as entry.
11376 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11378 (get_instantiated_value_entry): Make instantiate_below parameter
11379 the edge into the region.
11380 (instantiate_scev_name): Likewise. Adjust dominance checks,
11381 when we cannot use loop-based instantiation instantiate by
11382 walking use-def chains.
11383 (instantiate_scev_poly): Adjust.
11384 (instantiate_scev_binary): Likewise.
11385 (instantiate_scev_convert): Likewise.
11386 (instantiate_scev_not): Likewise.
11387 (instantiate_array_ref): Remove.
11388 (instantiate_scev_3): Likewise.
11389 (instantiate_scev_2): Likewise.
11390 (instantiate_scev_1): Likewise.
11391 (instantiate_scev_r): Do not blindly handle N-operand trees.
11392 Do not instantiate array-refs. Handle all constants and invariants.
11393 (instantiate_scev): Make instantiate_below parameter
11394 the edge into the region.
11395 (resolve_mixers): Use the loop preheader edge for the region
11396 parameter to instantiate_scev_r.
11397 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11399 2017-10-13 Richard Biener <rguenther@suse.de>
11401 PR tree-optimization/82525
11402 * graphite-isl-ast-to-gimple.c
11403 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11405 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11406 Fail code generation when we cannot represent the isl integer.
11407 (binary_op_to_tree): Elide modulo operations that are no-ops
11408 in the type we code generate. Remove now superfluous code
11411 2017-10-13 Richard Biener <rguenther@suse.de>
11413 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11414 (scop_detection::harmful_loop_in_region): Remove premature
11415 IV type restriction.
11416 (scop_detection::graphite_can_represent_scev): We can handle
11417 pointer IVs just fine.
11419 2017-10-13 Alan Modra <amodra@gmail.com>
11421 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11422 "Clobbers and Scratch Registers". Add paragraph on
11423 alternative to clobbers for scratch registers and OpenBLAS
11426 2017-10-13 Alan Modra <amodra@gmail.com>
11428 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11429 example of a memory input for a string of known length. Move
11430 commentary out of table. Add a number of new examples
11431 covering array memory inputs.
11433 2017-10-12 Martin Liska <mliska@suse.cz>
11435 PR tree-optimization/82493
11436 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11437 (test_range_functions): New function.
11438 (sbitmap_c_tests): Likewise.
11439 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11440 * selftest.h (sbitmap_c_tests): New function.
11442 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11444 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11446 * config/rs6000/amo.h: Fix spacing issue.
11448 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11451 * config/i386/i386.md (*ashl<mode>3_mask_1,
11452 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11453 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11456 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11458 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11459 (profile_probability): Set max_probability
11460 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11462 * profile-count.c (profile_probability::differs_from_p): Do not
11463 rely on max_probaiblity == 10000
11465 2017-10-12 Jeff Law <law@redhat.com>
11467 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11470 2017-10-12 Martin Sebor <msebor@redhat.com>
11474 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11475 (handle_alias_pairs): Call it.
11476 * common.opt (-Wattribute-alias): New option.
11477 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11478 * doc/invoke.texi (-Wattribute-alias): Document.
11480 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11483 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11485 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11487 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11488 (make_hard_regno_born, make_hard_regno_dead): Update
11489 bb_killed_pseudos and bb_gen_pseudos.
11491 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11493 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11495 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11497 * config/alpha/alpha.c (alpha_split_conditional_move):
11498 Use std::swap instead of manually swapping.
11499 (alpha_stdarg_optimize_hook): Ditto.
11500 (alpha_canonicalize_comparison): Ditto.
11502 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11504 * tree-loop-distribution.c (struct builtin_info): New struct.
11505 (struct partition): Refactor fields into struct builtin_info.
11506 (partition_free): Free struct builtin_info.
11507 (build_size_arg_loc, build_addr_arg_loc): Delete.
11508 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11509 information from struct builtin_info.
11510 (find_single_drs): New function refactored from classify_partition.
11511 Also moved builtin validity checks to this function.
11512 (compute_access_range, alloc_builtin): New functions.
11513 (classify_builtin_st, classify_builtin_ldst): New functions.
11514 (classify_partition): Refactor code into functions find_single_drs,
11515 classify_builtin_st and classify_builtin_ldst.
11516 (distribute_loop): Don't do runtime alias check when distributing
11518 (find_seed_stmts_for_distribution): New function.
11519 (pass_loop_distribution::execute): Refactor code finding seed
11520 stmts into above function. Support distribution for the innermost
11521 two-level loop nest. Adjust dump information.
11523 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11525 * tree-loop-distribution.c: Adjust the general comment.
11526 (NUM_PARTITION_THRESHOLD): New macro.
11527 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11528 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11529 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11530 in call to build_partition_graph.
11531 (finalize_partitions): New parameter. Make loop distribution more
11532 conservative by fusing more partitions.
11533 (distribute_loop): Don't do runtime alias check in case of loop nest
11535 (find_seed_stmts_for_distribution): New function.
11536 (prepare_perfect_loop_nest): New function.
11537 (pass_loop_distribution::execute): Refactor code finding seed stmts
11538 and loop nest into above functions. Support loop nest distribution.
11539 Adjust dump information accordingly.
11541 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11543 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11544 and set PTYPE_SEQUENTIAL for merged partition.
11546 2017-10-12 Richard Biener <rguenther@suse.de>
11548 PR tree-optimization/69728
11550 2017-09-19 Richard Biener <rguenther@suse.de>
11552 PR tree-optimization/69728
11553 * graphite-sese-to-poly.c (schedule_error): New global.
11554 (add_loop_schedule): Handle empty domain by failing the
11556 (build_original_schedule): Handle schedule_error.
11558 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11559 domain by returning an unchanged schedule.
11561 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11563 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11564 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11566 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11568 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11571 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11574 * expr.c (store_field): Don't optimize away bitsize == 0 store
11575 from CALL_EXPR with addressable return type.
11577 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11579 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11580 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11581 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11582 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11583 TARGET_ISEL instead of TARGET_ISEL<sel>.
11585 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11587 * config/rs6000/rs6000.c
11588 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11590 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11592 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11594 (reg_or_cint_operand): Fix comment.
11595 (reg_or_zero_operand): New predicate.
11596 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11597 * config/rs6000/rs6000.c (output_isel): Delete.
11598 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11599 instead of reg_or_cint_operand. Output instruction directly (not via
11601 (isel_unsigned_<mode>): Ditto.
11602 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11603 gpc_reg_operand. Add an instruction alternative for this. Output
11604 instruction directly.
11605 (*isel_reversed_unsigned_<mode>): Ditto.
11607 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11609 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11610 (TARGET_CANONICALIZE_COMPARISON): Define.
11612 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11615 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11616 Check whether the dest is REG before adding REG_EQUIV note.
11618 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11621 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11623 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11624 (make_hard_regno_born, make_hard_regno_dead): Update
11625 bb_killed_pseudos and bb_gen_pseudos.
11627 2017-10-11 Nathan Sidwell <nathan@acm.org>
11629 * incpath.h (enum incpath_kind): Name enum, prefix values.
11630 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11631 * incpath.c (heads, tails): Use INC_MAX.
11632 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11633 (merge_include_chains, split_quote_chain,
11634 register_include_chains): Update incpath_kind names.
11635 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11636 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11638 (add_framework_path, darwin_register_objc_includes): Likewise.
11639 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11641 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11643 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11644 Do not use float_operator operator predicate.
11645 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11646 * config/i386/predicates.md (float_operator): Remove predicate.
11648 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11650 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11651 (*jccxf_i387): Ditto.
11652 (*jcc<mode>_i387): Ditto.
11653 (*jccu<mode>_i387): Ditto.
11654 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11655 (*jcc_*_i387 splitters): Remove.
11656 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11657 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11658 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11661 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11663 * profile-count.h (slow_safe_scale_64bit): New function.
11664 (safe_scale_64bit): New inline.
11665 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11666 * profile-count.c: Include wide-int.h
11667 (slow_safe_scale_64bit): New.
11669 2017-10-11 Nathan Sidwell <nathan@acm.org>
11671 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11672 HAS_DECL_ASSEMBLER_NAME_P.
11673 * gimple-expr.c (gimple_decl_printable_name: Check
11674 HAS_DECL_ASSEMBLER_NAME_P too.
11675 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11676 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11677 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11678 * passes.c (rest_of_decl_compilation): Check
11679 HAS_DECL_ASSEMBLER_NAME_P too.
11680 * recog.c (verify_changes): Likewise.
11681 * tree-pretty-print.c (dump_decl_name): Likewise.
11682 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11684 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11685 (SET_DECL_ASSEMBLER_NAME): Use it.
11686 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11687 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11688 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11690 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11692 * config.gcc (i386, x86_64): Add extra objects.
11693 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11694 (ix86_min_insn_size): Declare.
11695 (ix86_issue_rate): Declare.
11696 (ix86_adjust_cost): Declare.
11697 (ia32_multipass_dfa_lookahead): Declare.
11698 (ix86_macro_fusion_p): Declare.
11699 (ix86_macro_fusion_pair_p): Declare.
11700 (ix86_bd_has_dispatch): Declare.
11701 (ix86_bd_do_dispatch): Declare.
11702 (ix86_core2i7_init_hooks): Declare.
11703 (ix86_atom_sched_reorder): Declare.
11704 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11705 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11706 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11707 (rip_relative_addr_p): Rename to ...
11708 (ix86_rip_relative_addr_p): ... this one; export.
11709 (memory_address_length): Update.
11710 (ix86_issue_rate): Move to x86-tune-sched.c.
11711 (ix86_flags_dependent): Move to x86-tune-sched.c.
11712 (ix86_agi_dependent): Move to x86-tune-sched.c.
11713 (exact_dependency_1): Move to x86-tune-sched.c.
11714 (exact_store_load_dependency): Move to x86-tune-sched.c.
11715 (ix86_adjust_cost): Move to x86-tune-sched.c.
11716 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11717 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11718 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11719 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11720 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11721 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11722 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11723 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11724 (min_insn_size): Rename to ...
11725 (ix86_min_insn_size): ... this one; export.
11726 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11727 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11728 (core2i7_first_cycle_multipass_backtrack): Move to
11729 x86-tune-sched-core.c.
11730 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11731 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11732 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11733 (ix86_avoid_jump_mispredicts): Update.
11734 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11735 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11736 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11737 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11738 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11739 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11740 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11741 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11742 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11743 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11744 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11745 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11746 (BIG): Move to ix86-tune-sched-bd.c.
11747 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11748 (enum insn_path): Move to ix86-tune-sched-bd.c.
11749 (get_mem_group): Move to ix86-tune-sched-bd.c.
11750 (is_cmp): Move to ix86-tune-sched-bd.c.
11751 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11752 (is_branch): Move to ix86-tune-sched-bd.c.
11753 (is_prefetch): Move to ix86-tune-sched-bd.c.
11754 (init_window): Move to ix86-tune-sched-bd.c.
11755 (allocate_window): Move to ix86-tune-sched-bd.c.
11756 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11757 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11758 (process_end_window): Move to ix86-tune-sched-bd.c.
11759 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11760 (find_constant): Move to ix86-tune-sched-bd.c.
11761 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11762 (has_immediate): Move to ix86-tune-sched-bd.c.
11763 (get_insn_path): Move to ix86-tune-sched-bd.c.
11764 (get_insn_group): Move to ix86-tune-sched-bd.c.
11765 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11766 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11767 (add_insn_window): Move to ix86-tune-sched-bd.c.
11768 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11769 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11770 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11771 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11772 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11773 (do_dispatch): Move to ix86-tune-sched-bd.c.
11774 (has_dispatch): Move to ix86-tune-sched-bd.c.
11775 * config/i386/t-i386: Add new object files.
11776 * config/i386/x86-tune-costs.h: New file.
11777 * config/i386/x86-tune-sched-atom.c: New file.
11778 * config/i386/x86-tune-sched-bd.c: New file.
11779 * config/i386/x86-tune-sched-core.c: New file.
11780 * config/i386/x86-tune-sched.c: New file.
11782 2017-10-11 Liu Hao <lh_mouse@126.com>
11784 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11785 the generic version below instead.
11786 (should_colorize): Recognize Windows consoles as terminals
11788 * pretty-print.c [__MINGW32__] (write_all): New function.
11789 [__MINGW32__] (find_esc_head): Likewise.
11790 [__MINGW32__] (find_esc_terminator): Likewise.
11791 [__MINGW32__] (eat_esc_sequence): Likewise.
11792 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11794 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11797 2017-10-11 Richard Biener <rguenther@suse.de>
11799 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11800 Properly call analyze_scalar_evolution with the loop of the stmt.
11802 2017-10-11 Richard Biener <rguenther@suse.de>
11804 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11805 * tree-core.h (tree_base): Add chrec_var union member.
11806 * tree.h (CHREC_VAR): Remove.
11807 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11808 * tree-chrec.h (build_polynomial_chrec): Adjust.
11809 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11810 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11812 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11814 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11815 * match.pd: ... here.
11816 ((T) X == (T) Y): Relax condition.
11818 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11820 PR tree-optimization/82472
11821 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11823 (break_alias_scc_partitions): Update postorder number.
11825 2017-10-11 Martin Liska <mliska@suse.cz>
11828 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11830 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11832 2017-10-11 Martin Liska <mliska@suse.cz>
11834 * print-rtl.c (print_insn): Move declaration of idbuf
11835 to same scope as name.
11837 2017-10-11 Martin Liska <mliska@suse.cz>
11842 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11843 volatile arguments.
11845 2017-10-11 Martin Liska <mliska@suse.cz>
11848 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11849 volatile arguments.
11851 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11853 * config.gcc (default_gnu_indirect_function): Default to yes for
11854 arm*-*-linux* with glibc.
11856 2017-10-11 Richard Biener <rguenther@suse.de>
11858 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11859 default-defs and types we do not want to analyze.
11860 (interpret_loop_phi): Replace unreachable code with an assert.
11861 (compute_scalar_evolution_in_loop): Remove and inline ...
11862 (analyze_scalar_evolution_1): ... here, replacing condition with
11863 what makes the intent clearer. Remove handling of cases
11864 get_scalar_evolution now handles.
11866 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11868 PR rtl-optimization/81434
11869 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11870 comment for main loop. In sched_group_found if, also add checks for
11871 pass and min_cost_group.
11873 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11875 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11876 (rs6000_insn_cost): New function.
11877 * config/rs6000/rs6000.md (cost): New attribute.
11879 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11880 H.J. Lu <hongjiu.lu@intel.com>
11884 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11885 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11886 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11887 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11888 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11889 OPTION_MASK_ISA_MMX and some other bit set require both
11890 mmx and the ISAs without the mmx bit.
11891 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11892 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11893 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11894 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11895 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11896 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11897 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11898 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11899 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11900 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11901 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11902 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11903 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11904 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11905 Add OPTION_MASK_ISA_MMX.
11907 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11909 * config.gcc (armv7*-*-freebsd*): New target.
11910 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11912 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11914 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11915 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11918 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11920 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11921 a template parameter.
11922 (WIDE_INT_REF_FOR): Update accordingly.
11923 * tree.h (wi::int_traits <const_tree>): Delete.
11924 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11925 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11926 (wi::tree_to_wide_ref): New typedef.
11927 (wi::to_wide): New function.
11928 * calls.c (get_size_range): Use wi::to_wide when operating on
11929 trees as wide_ints.
11930 * cgraph.c (cgraph_node::create_thunk): Likewise.
11931 * config/i386/i386.c (ix86_data_alignment): Likewise.
11932 (ix86_local_alignment): Likewise.
11933 * dbxout.c (stabstr_O): Likewise.
11934 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11935 * expr.c (const_vector_from_tree): Likewise.
11936 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11937 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11938 (fold_negate_expr_1, int_const_binop_1, const_binop)
11939 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11940 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11941 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11942 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11943 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11944 (fold_not_const, round_up_loc): Likewise.
11945 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11946 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11947 (alloca_call_type): Likewise.
11948 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11949 * godump.c (go_output_typedef): Likewise.
11950 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11951 * internal-fn.c (get_min_precision): Likewise.
11952 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11953 * ipa-polymorphic-call.c
11954 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11955 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11956 (ipa_modify_call_arguments): Likewise.
11957 * match.pd: Likewise.
11958 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11959 * print-tree.c (print_node_brief, print_node): Likewise.
11960 * stmt.c (expand_case): Likewise.
11961 * stor-layout.c (layout_type): Likewise.
11962 * tree-affine.c (tree_to_aff_combination): Likewise.
11963 * tree-cfg.c (group_case_labels_stmt): Likewise.
11964 * tree-data-ref.c (dr_analyze_indices): Likewise.
11965 (prune_runtime_alias_test_list): Likewise.
11966 * tree-dump.c (dequeue_and_dump): Likewise.
11967 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11968 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11969 * tree-pretty-print.c (dump_generic_node): Likewise.
11970 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11971 (simple_iv_with_niters): Likewise.
11972 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11973 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11974 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11975 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11976 (refine_value_range_using_guard, number_of_iterations_ne_max)
11977 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11978 (get_cst_init_from_scev, record_nonwrapping_iv)
11979 (scev_var_range_cant_overflow): Likewise.
11980 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11981 * tree-ssa-pre.c (compute_avail): Likewise.
11982 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11983 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11984 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11985 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11986 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11987 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11988 (dump_case_nodes, try_switch_expansion): Likewise.
11989 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11990 (vect_do_peeling): Likewise.
11991 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11992 * tree-vect-stmts.c (vectorizable_load): Likewise.
11993 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11994 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11995 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11996 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11997 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11998 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11999 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12000 (evrp_dom_walker::before_dom_children): Likewise.
12001 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12002 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12003 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12004 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12005 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12006 (get_range_pos_neg): Likewise.
12007 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12008 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12009 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12010 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12011 * config/avr/avr.c (avr_fold_builtin): Likewise.
12012 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12013 * config/msp430/msp430.c (msp430_attr): Likewise.
12014 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12015 * config/powerpcspe/powerpcspe-c.c
12016 (altivec_resolve_overloaded_builtin): Likewise.
12017 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12018 (rs6000_expand_ternop_builtin): Likewise.
12019 * config/rs6000/rs6000-c.c
12020 (altivec_resolve_overloaded_builtin): Likewise.
12021 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12022 (rs6000_expand_ternop_builtin): Likewise.
12023 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12025 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12027 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12028 when copying loop nest with only one inner loop.
12030 2017-10-10 Richard Biener <rguenther@suse.de>
12032 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12033 blocks if SCEV is active.
12034 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12036 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12037 (scev_initialize): Assert we are not yet initialized.
12039 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12041 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12042 inner loop's exit stmt by making it always exit the loop, otherwise
12043 we would generate an infinite empty loop.
12045 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12047 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12048 renaming variables in new preheader if it's deleted.
12050 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12052 * tree-loop-distribution.c (struct partition): Remove unused field
12053 loops of the structure.
12054 (partition_alloc, partition_free): Ditto.
12055 (build_rdg_partition_for_vertex): Ditto.
12057 2017-10-09 Jeff Law <law@redhat.com>
12059 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12060 return type to match prototype and documentation.
12062 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12064 * config/rs6000/rs6000.c (processor_costs): Move to ...
12065 * config/rs6000/rs6000.h: ... here.
12066 (rs6000_cost): Declare.
12068 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12070 * except.c (setjmp_fn): New global variable.
12071 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12072 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12073 if DONT_USE_BUILTIN_SETJMP is defined.
12075 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12077 * target.def (insn_cost): New hook.
12078 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12079 * doc/tm.texi: Regenerate.
12080 * rtlanal.c (insn_cost): Use the new hook.
12082 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12084 * combine.c (combine_validate_cost): Compute the new insn_cost,
12085 not just pattern_cost.
12086 (try_combine): Adjust comment.
12088 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12090 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12092 * combine.c (uid_insn_cost): Adjust comment.
12093 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12095 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12096 * dse.c (find_shift_sequence): Ditto.
12097 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12098 (bb_valid_for_noce_process_p): Use pattern_cost.
12099 * rtl.h (insn_rtx_cost): Delete.
12100 (pattern_cost): New prototype.
12101 (insn_cost): New prototype.
12102 * rtlanal.c (insn_rtx_cost): Rename to...
12103 (pattern_cost): ... this.
12106 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12108 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12109 (*jcc<mode>_0_r_i387): Ditto.
12110 (*jccxf_r_i387): Ditto.
12111 (*jcc<mode>_r_i387): Ditto.
12112 (*jccu<mode>_r_i387): Ditto.
12113 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12114 (*jcc): Rename from *jcc_1.
12116 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12118 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12119 deferred rescans after the lvx/stvx recombination pre-pass.
12121 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12123 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12124 memory operation instruction support.
12125 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12126 (rs6000-ibm-aix[789]*): Likewise.
12127 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12128 Document new functions.
12130 2017-10-09 Richard Biener <rguenther@suse.de>
12132 PR tree-optimization/82397
12133 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12134 equality only for semantically equal trees.
12136 2017-10-09 Richard Biener <rguenther@suse.de>
12138 PR tree-optimization/82449
12139 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12140 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12141 allow constant addresses.
12142 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12145 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12147 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12150 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12153 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12156 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12159 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12161 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12164 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12165 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12167 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12169 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12170 (WI_BINARY_PREDICATE_RESULT): Likewise.
12171 (wi::binary_traits::operator_result): New type.
12172 (wi::binary_traits::predicate_result): Likewise.
12173 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12174 (generic_wide_int::operator==, generic_wide_int::operator!=)
12175 (generic_wide_int::operator&, generic_wide_int::and_not)
12176 (generic_wide_int::operator|, generic_wide_int::or_not)
12177 (generic_wide_int::operator^, generic_wide_int::operator+
12178 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12180 (operator~, unary operator-, operator==, operator!=, operator&)
12181 (operator|, operator^, operator+, binary operator-, operator*): New
12183 * expr.c (get_inner_reference): Use wi::bit_and_not.
12184 * fold-const.c (fold_binary_loc): Likewise.
12185 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12186 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12187 (bit_value_binop): Likewise.
12188 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12189 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12190 (extract_range_from_binary_expr_1): Likewise.
12191 (masked_increment): Likewise.
12192 (simplify_bit_ops_using_ranges): Likewise.
12194 2017-10-09 Martin Jambor <mjambor@suse.cz>
12197 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12198 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12199 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12201 (hsa_fixup_mov_insn_type): New function.
12202 (hsa_op_with_type::get_in_type): Use it.
12203 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12204 immediates in an assert.
12205 (hsa_op_with_type::extend_int_to_32bit): New method.
12206 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12207 types. Convert to dest type if necessary.
12208 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12209 (reg_for_gimple_ssa): Pass false as min32int to
12210 hsa_type_for_scalar_tree_type.
12211 (gen_hsa_addr): Fixup type when creating addresable temporary.
12212 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12213 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12214 necessary. Call hsa_fixup_mov_insn_type.
12215 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12216 extend operands and convert to dest type if necessary.
12217 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12218 to dest type if necessary.
12219 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12220 if conversion nt necessary and size matches.
12221 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12222 to dest type if necessary.
12223 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12224 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12226 (gen_hsa_clrsb): Likewise.
12227 (gen_hsa_ffs): Likewise.
12228 (gen_hsa_divmod): Extend operands and convert to dest type if
12230 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12232 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12234 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12235 Remove empty default arguments. Use a brace block as output
12237 (conditional return): Ditto.
12239 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12240 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12242 (group_ending_nop): Ditto.
12243 (doloop_end): Ditto.
12244 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12245 (splitters for those): Ditto.
12247 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12249 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12250 a conditional jump (and the compare for it) so that pc_rtx is the
12252 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12253 for the deleted and renamed ctr<mode>_internal[234] patterns.
12254 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12255 Delete second conditional return pattern.
12256 (ctr<mode>_internal2): Delete this second bdnz pattern.
12257 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12258 (ctr<mode>_internal4): Delete this second bdz pattern.
12260 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12262 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12263 (always_initialized_rtx_for_ssa_name_p): New predicate.
12264 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12265 (finish_out_of_ssa): Free new field of SA.
12266 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12267 * tree-ssa-coalesce.c: Include tree-ssa.h.
12268 (get_parm_default_def_partitions): Remove extern keyword.
12269 (get_undefined_value_partitions): New function.
12270 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12271 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12272 uninitialized bits.
12273 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12275 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12277 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12279 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12281 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12282 for targets that preffer 128bit.
12284 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12286 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12288 2017-10-08 Olivier Hainque <hainque@adacore.com>
12290 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12291 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12293 (thumb_set_return_address): Likewise.
12295 2017-10-08 Olivier Hainque <hainque@adacore.com>
12297 * common/config/arm/arm-common.c (arm_except_unwind_info):
12298 Handle DWARF2_UNWIND_INFO.
12300 2017-10-07 Michael Collison <michael.collison@arm.com>
12302 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12305 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12307 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12308 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12309 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12310 defined, force the creation of a new block for a dispatch label.
12312 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12314 * invoke.texi (Wsuggest-attribute=cold): Document.
12315 * common.opt (Wsuggest-attribute=cold): New
12316 * ipa-pure-const.c (warn_function_cold): New function.
12317 * predict.c (compute_function_frequency): Use it.
12318 * predict.h (warn_function_cold): Declare.
12320 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12322 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12325 2017-10-06 Martin Liska <mliska@suse.cz>
12327 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12328 keyword for member functions.
12329 (struct sanopt_tree_couple): New struct.
12330 (struct sanopt_tree_couple_hash): New function.
12331 (struct sanopt_ctx): Add new hash_map.
12332 (has_dominating_ubsan_ptr_check): New function.
12333 (record_ubsan_ptr_check_stmt): Likewise.
12334 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12335 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12336 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12338 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12341 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12342 aarch64_simd_valid_immediate on CONST_VECTORs.
12343 (aarch64_reg_or_bic_imm): Likewise.
12345 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12347 PR rtl-optimization/82396
12348 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12350 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12352 * graphite-dependences.c (scop_get_reads): Move code to...
12353 (scop_get_must_writes): Move code to...
12354 (scop_get_may_writes): Move code to...
12355 (scop_get_reads_and_writes): ... here.
12356 (scop_get_dependences): Call scop_get_reads_and_writes.
12358 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12360 PR tree-optimization/82434
12361 * fold-const.h (can_native_encode_type_p,
12362 can_native_encode_string_p): Remove.
12363 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12364 don't encode anything, just return what would be otherwise returned.
12365 (native_encode_fixed, native_encode_complex, native_encode_vector):
12367 (native_encode_string): Likewise. Inline by hand
12368 can_native_encode_string_p.
12369 (can_native_encode_type_p): Remove.
12370 (can_native_encode_string_p): Remove.
12371 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12372 STRING_CSTs using can_native_encode_string_p, test all
12373 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12374 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12375 argument from native_encode_expr.
12376 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12377 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12380 2017-10-06 Richard Biener <rguenther@suse.de>
12382 PR tree-optimization/82397
12383 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12384 operand_equal_p but rely on data_ref_compare_tree for detecting
12386 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12387 to match up with dr_group_sort_cmp.
12389 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12392 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12394 * config/s390/s390-builtin-types.def: Regenerate.
12396 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12399 * config/s390/s390-builtin-types.def: Regenerate.
12400 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12401 Change flag from B_VXE to B_VX.
12402 (s390_vec_min_dbl): Remove B_VXE flag.
12404 2017-10-06 Richard Biener <rguenther@suse.de>
12406 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12407 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12408 translate_isl_ast_to_gimple::is_valid_rename,
12409 translate_isl_ast_to_gimple::get_rename,
12410 translate_isl_ast_to_gimple::get_def_bb_for_const,
12411 translate_isl_ast_to_gimple::get_new_name,
12412 translate_isl_ast_to_gimple::collect_all_ssa_names,
12413 translate_isl_ast_to_gimple::copy_loop_phi_args,
12414 translate_isl_ast_to_gimple::collect_all_ssa_names,
12415 translate_isl_ast_to_gimple::copy_loop_phi_args,
12416 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12417 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12418 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12419 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12420 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12421 translate_isl_ast_to_gimple::copy_cond_phi_args,
12422 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12423 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12424 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12425 translate_isl_ast_to_gimple::rename_uses,
12426 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12427 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12428 (set_rename_for_each_def): Likewise.
12429 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12430 here. Handle rewriting SCEV analyzable uses here.
12431 (copy_bb_and_scalar_dependences): Generate code for PHI
12433 (graphite_regenerate_ast_isl): Adjust.
12434 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12435 (add_write, add_read): New functions.
12436 (build_cross_bb_scalars_def): Use it and simplify.
12437 (build_cross_bb_scalars_use): Likewise.
12438 (graphite_find_cross_bb_scalar_vars): Inline into...
12439 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12440 simulating out-of-SSA. Compute liveout and add dependencies.
12441 (build_scops): Force an empty entry block.
12442 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12444 (sese_build_liveouts): Declare.
12445 (sese_trivially_empty_bb_p): Likewise.
12446 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12447 compute liveout and debug_liveout.
12448 (sese_bad_liveouts_use): Remove.
12449 (sese_reset_debug_liveouts_bb): Likewise.
12450 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12451 (sese_build_liveouts): Build liveout and debug_liveout and store
12453 (new_sese_info): Adjust.
12454 (free_sese_info): Likewise.
12455 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12456 do not build liveout here.
12457 (move_sese_in_condition): Adjust region entry.
12458 (scev_analyzable_p): Match up with chrec_apply requirements.
12459 (sese_trivially_empty_bb_p): New.
12460 * tree-into-ssa.c (get_reaching_def): Properly support generating
12461 default-defs for incremental rewrite of anonymous names.
12463 2017-10-06 Richard Biener <rguenther@suse.de>
12465 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12466 precision do not perform modulo reduction.
12468 2017-10-06 Richard Biener <rguenther@suse.de>
12470 PR tree-optimization/82436
12471 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12472 conservatively choose the vectorization factor when checking
12473 whether we can perform the required load permutation.
12474 (vect_transform_slp_perm_load): Assert when we may not fail.
12476 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12478 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12479 message for incompatible -msdata=* and -mcall-* options.
12481 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12483 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12484 rate for post-reload scheduling.
12486 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12488 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12490 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12492 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12493 to improve monte carlo in scimark.
12495 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12497 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12498 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12499 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12500 pentium4_cost, nocona_cost): Set reassociation width to 1.
12501 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12502 width to 2 for fp operations and 1 otherwise.
12503 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12505 (atom_cost): Set reassociation width to 2.
12506 (slm_cost, generic_cost): Set fp reassociation width
12507 to 2 and 1 otherwise.
12508 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12509 (core_cost): Set fp reassociation width to 4 and vector to 2.
12510 (ix86_reassociation_width): Rewrite using cost table; special case
12511 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12512 and TARGET_AVX128_OPTIMAL.
12513 * config/i386/i386.h (processor_costs): Add
12514 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12515 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12516 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12517 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12518 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12519 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12521 2017-10-05 Nathan Sidwell <nathan@acm.org>
12523 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12525 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12527 * config/arm/arm.c (arm_test_fpu_data): New.
12528 (arm_run_selftests): Call arm_test_fpu_data.
12530 2017-10-04 Nathan Sidwell <nathan@acm.org>
12532 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12534 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12536 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12538 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12539 check type for aarch64_simd_valid_immediate.
12540 (aarch64_output_simd_mov_immediate): Update prototype.
12541 (aarch64_simd_valid_immediate): Update prototype.
12542 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12543 support for ORR-immediate.
12544 (and<mode>3): modified pattern to add support for BIC-immediate.
12545 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12546 now checks for valid immediate for BIC and ORR based on new enum
12548 (aarch64_output_simd_mov_immediate): Function now used to output
12549 BIC/ORR imm as well based on new enum argument.
12550 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12552 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12553 (aarch64_reg_or_bic_imm): Likewise.
12555 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12557 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12558 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12560 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12564 PR rtl-optimization/82396
12565 * haifa-sched.c (autopref_multipass_init): Simplify
12567 (autopref_rank_data): Simplify sort order.
12568 * sched-int.h (autopref_multipass_data_): Remove
12569 multi_mem_insn_p, min_offset and max_offset.
12571 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12573 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12575 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12577 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12578 vect_doubleint_cvt.
12580 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12582 * doc/sourcebuild.texi: Document vect_long_mult.
12584 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12586 PR tree-optimization/82413
12587 * fold-const.c (build_range_check): Use widest_int when comparing
12588 the maximum ETYPE value with HIGH.
12590 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12592 PR rtl-optimization/82396
12593 * haifa-sched.c (autopref_multipass_init): Simplify
12595 (autopref_rank_data): Simplify sort order.
12596 * sched-int.h (autopref_multipass_data_): Remove
12597 multi_mem_insn_p, min_offset and max_offset.
12599 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12601 PR tree-optimization/82381
12602 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12603 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12606 PR tree-optimization/82374
12607 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12608 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12609 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12610 current_function_decl to the new decl.
12612 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12614 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12615 helper macro for IEEE float128 hardware built-in functions.
12616 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12618 (TRUNCF128_ODD): Likewise.
12619 (ADDF128_ODD): Likewise.
12620 (SUBF128_ODD): Likewise.
12621 (MULF128_ODD): Likewise.
12622 (DIVF128_ODD): Likewise.
12623 (FMAF128_ODD): Likewise.
12624 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12625 UNSPEC_TRUNC_ROUND_TO_ODD.
12626 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12627 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12628 floating point round to odd instructions.
12629 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12630 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12631 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12632 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12633 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12634 (trunc<mode>sf2_hw): Change the truncate with round to odd
12635 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12636 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12637 to odd hardware instructions.
12638 (sub<mode>3_odd): Likewise.
12639 (mul<mode>3_odd): Likewise.
12640 (div<mode>3_odd): Likewise.
12641 (sqrt<mode>2_odd): Likewise.
12642 (fma<mode>4_odd): Likewise.
12643 (fms<mode>4_odd): Likewise.
12644 (nfma<mode>4_odd): Likewise.
12645 (nfms<mode>4_odd): Likewise.
12646 (trunc<mode>df2_odd): Change the truncate with round to odd
12647 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12649 * doc/extend.texi (PowerPC built-in functions): Update documentation
12650 for existing IEEE float128-bit built-in functions. Add built-in
12651 functions that generate the IEEE 128-bit floating point round to
12654 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12656 PR rtl-optimization/77729
12657 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12658 to (X&(C1&~C2))|C2 transformations.
12660 2017-10-03 Martin Jambor <mjambor@suse.cz>
12662 PR tree-optimization/82363
12663 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12664 mismatch, mark lacc written regardless of racc.
12666 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12668 PR tree-optimization/82381
12669 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12670 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12671 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12672 fallthrough into reassoc_stmt_dominates_stmt_p.
12675 * combine.c (combine_instructions): Don't combine in unreachable
12678 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12682 function to not use the have_cpu variable. Do not set cpu_index,
12683 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12684 or the default cpu.
12685 (rs6000_valid_attribute_p): Remove duplicate initializations of
12686 old_optimize and func_optimize.
12687 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12688 (rs6000_activate_target_options): Make global.
12689 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12692 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12694 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12695 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12696 Set *poffset to 0 and *psize and *pmax_size to -1 if
12697 *poffset + *psize overflows in HOST_WIDE_INT.
12699 PR tree-optimization/82387
12700 PR tree-optimization/82388
12701 PR tree-optimization/82389
12702 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12703 instead of live_bytes non-NULL.
12705 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12708 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12711 2017-10-02 Richard Biener <rguenther@suse.de>
12713 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12714 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12715 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12717 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12719 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12720 requested precision matches the type's.
12721 * calls.c (alloc_max_size): Calculate the new candidate size as
12722 a widest_int and use wi::to_widest when comparing it with the
12723 current candidate size.
12724 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12725 zero rather than integer_zero_node.
12726 * match.pd: Check for a no-op conversion before using wi::add
12727 rather than after. Use tree_to_uhwi when summing small shift
12728 counts into an unsigned int.
12730 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12731 Alan Hayward <alan.hayward@arm.com>
12732 David Sherwood <david.sherwood@arm.com>
12735 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12736 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12737 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12738 POINTER_AND_FP_REGS.
12740 2017-10-02 Richard Biener <rguenther@suse.de>
12742 PR tree-optimization/82355
12743 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12744 a mapping for the enclosing loop but avoid generating one for
12745 the loop tree root.
12746 (copy_bb_and_scalar_dependences): Remove premature codegen
12747 error on PHIs in blocks duplicated into multiple places.
12748 * graphite-scop-detection.c
12749 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12750 in the region use it as loop and nest to analyze the DR in.
12751 (try_generate_gimple_bb): Likewise.
12752 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12753 (add_loop_constraints): For blocks in a loop not in the region
12754 create a dimension with a single iteration.
12755 * sese.h (gbb_loop_at_index): Remove assert.
12757 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12759 * omp-expand.c (adjust_context_scope): New function.
12760 (expand_parallel_call): Call adjust_context_scope.
12762 2017-10-01 Jeff Law <law@redhat.com>
12764 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12765 dom_opt_dom_walker class with direct access to private members.
12766 Add comments. Call test_for_singularity.
12767 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12768 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12769 m_dummy_cond anymore.
12770 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12772 (pass_dominator:execute): Build the dummy_cond here and pass it
12773 to the dom_opt_dom_walker ctor.
12774 (test_for_singularity): New function.
12776 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12777 Maya Rashish <coypu@sdf.org>
12779 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12780 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12781 (alpha*-*-netbsd*) Use nbsd_tm_file.
12782 (arm*-*-netbsdelf*) Likewise.
12783 (i[34567]86-*-netbsdelf*) Likewise.
12784 (x86_64-*-netbsd*) Likewise.
12785 (mips*-*-netbsd*) Likewise.
12786 (powerpc-*-netbsd*) Likewise.
12787 (sh*-*-netbsd*) Likewise.
12788 (sparc-*-netbsdelf*) Likewise.
12789 (sparc64-*-netbsd*) Likewise.
12790 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12792 (vax-*-netbsdelf*) Likewise.
12793 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12794 (UINT_FAST8_TYPE) Likewise.
12795 (INT_FAST16_TYPE) Check CHAR_FAST16.
12796 (UINT_FAST16_TYPE) Likewise.
12798 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12801 * config/i386/i386.md
12802 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12803 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12804 *divmodsi4_zext_2): New define_insn_and_split.
12805 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12806 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12807 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12808 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12809 New define_insn_and_split.
12810 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12811 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12812 operands[1] having DImode when mode is SImode.
12814 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12815 always SImode for DIV and MOD in REG_EQUAL notes.
12817 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12819 PR middle-end/82319
12820 * match.pd: Fix handling of NaNs in pattern.
12822 2017-09-29 Jeff Law <law@redhat.com>
12824 * sbitmap.c (bitmap_bit_in_range_p): New function.
12825 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12826 * tree-ssa-dse.c (live_bytes_read): New function.
12827 (dse_classify_store): Ignore reads of dead bytes.
12829 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12830 typos and whitespace errors.
12831 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12832 * config/s390/s390.c (s390_emit_prologue): Likewise.
12834 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12837 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12838 with a symbol for LRA.
12840 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12842 PR rtl-optimization/82338
12843 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12845 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12847 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12848 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12849 (qsort_chk): Declare.
12850 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12851 (qsort_chk): New function.
12853 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12855 PR tree-optimization/82337
12856 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12857 phi definition if the PHI result appears in an abnormal PHI.
12858 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12859 basis appears in an abnormal PHI.
12861 2017-09-29 Richard Biener <rguenther@suse.de>
12863 * graphite-isl-ast-to-gimple.c
12864 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12865 (binary_op_to_tree): Use it.
12866 (get_rename_from_scev): Likewise.
12867 (copy_loop_phi_nodes): Likewise.
12868 (copy_bb_and_scalar_dependences): Likewise.
12869 (translate_pending_phi_nodes): Likewise.
12871 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12874 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12875 for movabsq $(i32 << shift), r64.
12877 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12879 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12880 index when encoding %esp as %rsp to avoid 0x67 prefix.
12882 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12884 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12885 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12887 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12889 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12890 Extensions with more than 16 double VFP registers.
12891 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12892 to_clear_mask and all code related to it. Replace the remaining
12893 entry by a sbitmap and adapt code accordingly.
12895 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12897 * brig-builtins.def: Change pure attributes to const.
12899 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12901 * config.gcc (default_gnu_indirect_function): Default to yes for
12902 sparc*-*-linux* with glibc.
12904 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12906 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12907 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12908 when creating .init_array and .fini_array sections with priority
12911 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12914 * config/aarch64/aarch64.c
12915 (aarch64_builtin_support_vector_misalignment): Always return false
12916 when misalignment is unknown.
12918 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12920 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12921 this function to return false if the definition used by the swap
12922 instruction is artificial, or if the memory address from which the
12923 constant value is loaded is not represented by a base address held
12924 in a register or if the base address register is a frame or stack
12925 pointer. Additionally, return false if the base address of the
12926 loaded constant is a SYMBOL_REF but is not considered to be a
12928 (replace_swapped_load_constant): New function.
12929 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12930 loaded constant vector with a load of a swapped constant vector.
12932 2017-09-27 Carl Love <cel@us.ibm.com>
12934 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12935 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12936 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12939 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12941 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12942 first, always call autopref_rank_data otherwise.
12944 2017-09-27 Richard Biener <rguenther@suse.de>
12946 * graphite-scop-detection.c (find_scop_parameters): Move
12947 loop bound handling ...
12948 (gather_bbs::before_dom_children): ... here, avoiding the need
12949 to build scop_info->loop_nest.
12950 (record_loop_in_sese): Remove.
12951 * sese.h (sese_info_t::loop_nest): Remove.
12952 * sese.c (new_sese_info): Do not allocate loop_nest.
12953 (free_sese_info): Do not free loop_nest.
12955 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12958 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12959 lhs from calls if the lhs has addressable type.
12961 2017-09-27 Richard Biener <rguenther@suse.de>
12963 * graphite.h (scop::max_alias_set): New member.
12964 * graphite-scop-detection.c: Remove references to non-existing
12965 --param in comments.
12966 (build_alias_sets): Record the maximum alias set used for drs.
12967 (build_scops): Support zero as unlimited for
12968 --param graphite-max-arrays-per-scop.
12969 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12970 and inline into ...
12971 (build_poly_sr_1): ... here. Compute alias set based on the
12972 maximum alias set used for drs rather than
12973 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12975 2017-09-27 Richard Biener <rguenther@suse.de>
12977 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12978 --param loop-block-tile-size=0 to disable tiling.
12980 2017-09-27 Richard Biener <rguenther@suse.de>
12982 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12983 (graphite-max-nb-scop-params): Document special value zero.
12984 * domwalk.h (dom_walker::STOP): New symbolical constant.
12985 (dom_walker::dom_walker): Add optional parameter for bb to
12987 (dom_walker::~dom_walker): Declare.
12988 (dom_walker::before_dom_children): Document STOP return value.
12989 (dom_walker::m_user_bb_to_rpo): New member.
12990 (dom_walker::m_bb_to_rpo): Likewise.
12991 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12992 mapping here if not provided by the user.
12993 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12994 provided by the user.
12995 (dom_walker::STOP): Define.
12996 (dom_walker::walk): Do not compute bb to RPO mapping here.
12997 Support STOP return value from before_dom_children to stop
12999 * graphite-optimize-isl.c (optimize_isl): If the schedule
13000 is the same still generate code if -fgraphite-identity
13001 or -floop-parallelize-all are given.
13002 * graphite-scop-detection.c: Include cfganal.h.
13003 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13005 (gather_bbs::before_dom_children): Return STOP for BBs
13007 (build_scops): Compute bb to RPO mapping and pass it to
13008 the domwalk. Treat --param graphite-max-nb-scop-params=0
13009 as not limiting the number of params.
13010 * graphite.c (graphite_initialize): Remove limit on the
13011 number of basic-blocks in a function.
13012 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13013 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13014 default value of 10.
13016 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13018 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13019 Adjust code to eliminate needing to do the shift right 32-bits
13020 operation after XSCVDPSPN.
13022 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13024 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13025 ((X / Y) != 0 -> X >= Y): Likewise.
13027 2017-09-26 Carl Love <cel@us.ibm.com>
13029 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13030 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13031 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13032 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13033 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13034 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13035 definitions and overloading.
13036 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13037 statement for P9V_BUILTIN_XST_LEN_R.
13038 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13039 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13040 define_expand and define_insn for the instructions and builtins.
13041 * doc/extend.texi: Update the built-in documentation file for the new
13042 built-in functions.
13043 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13044 define_insn for the instructions
13046 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13049 * gcc/config/netbsd-protos.h: New file.
13050 * gcc/config/netbsd.c: New file.
13051 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13052 * gcc/config/t-netbsd: New file.
13053 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13054 (tmake_file) Add t-netbsd.
13055 (extra_objs) Add netbsd.o.
13057 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13061 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13063 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13065 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13066 sign extension from a vector register to a GPR by doing a 32-bit
13067 direct move and then an EXTSW.
13068 (extendsi<mode>2 splitter): Likewise.
13069 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13070 right or vector extract after doing XSCVDPSPN. Use
13071 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13073 (movdi_from_sf_zero_ext): Likewise.
13074 (reload_gpr_from_vsxsf): Likewise.
13075 (p8_mfvsrd_4_disf): Delete, no longer used.
13076 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13077 then needing to move the SFmode to a GPR to use the XSCVDPSP
13078 instruction instead of FRSP and XSCVDPSPN.
13079 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13080 it is adjacent to the other XSCVSPDP insns.
13081 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13082 SFmode to be in traditional Altivec registers.
13083 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13084 (vsx_xscvspdpn): Likewise.
13085 (vsx_xscvspdpn_scalar): Likewise.
13087 2017-09-26 Martin Jambor <mjambor@suse.cz>
13089 * tree-sra.c (compare_access_positions): Put integral types first,
13090 stabilize sorting of integral types, remove conditions putting
13091 non-full-precision integers last.
13092 (sort_and_splice_var_accesses): Disable scalarization if a
13093 non-integert would be represented by a non-full-precision integer.
13095 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13097 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13098 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13099 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13100 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13101 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13102 conditionals inside the function instead of around it. Call
13103 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13104 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13106 2017-09-26 Richard Biener <rguenther@suse.de>
13108 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13110 (scop_detection::build_scop_breadth): ... this. Removed.
13111 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13112 (scop_detection::harmful_stmt_in_bb): Likewise.
13113 (scop_detection::graphite_can_represent_stmt): Likewise.
13114 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13115 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13116 (scop_detection::can_represent_loop_1): ... this. Removed.
13117 (scop_detection::harmful_loop_in_region): Simplify after inlining
13118 the above and remove more quadraticness.
13119 (build_scops): Adjust.
13120 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13123 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13126 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13127 REGNO (base) == SP_REG if base is a REG.
13129 PR middle-end/35691
13130 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13131 if it is different SSA_NAME.
13132 (optimize_range_tests_cmp_bitwise): New function.
13133 (optimize_range_tests): Call it.
13135 2017-09-26 Richard Biener <rguenther@suse.de>
13137 PR tree-optimization/82321
13138 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13139 for the def being inside the loop.
13141 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13143 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13145 * config/s390/s390-builtins.def: Fix constraint on op4.
13147 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13149 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13150 independent expanders.
13151 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13152 ("vec_ordered", "vec_unordered"): New expanders.
13154 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13156 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13159 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13161 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13162 vec_unpacks_lo_v16qi.
13163 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13165 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13167 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13168 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13169 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13171 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13173 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13175 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13177 ("*vec_slb<mode>"): New insn pattern.
13178 ("vec_shr_<mode>"): New expander.
13179 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13180 and force the shift count operand to V16QImode.
13181 ("vec_srb<mode>"): Set shift count mode to V16QI.
13183 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13185 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13186 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13187 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13189 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13192 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13194 2017-09-26 Richard Biener <rguenther@suse.de>
13196 PR tree-optimization/82320
13197 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13200 2017-09-25 Jeff Law <law@redhat.com>
13202 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13203 prototype for new argument.
13204 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13205 mostly extracted from rs6000_emit_allocate_stack.
13206 (rs6000_emit_probe_stack_range_stack_clash): New function.
13207 (rs6000_emit_allocate_stack): Call
13208 rs6000_emit_probe_stack_range_stack_clash as needed.
13209 (rs6000_emit_probe_stack_range): Add additional argument
13210 to call to gen_probe_stack_range{si,di}.
13211 (output_probe_stack_range): New.
13212 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13213 (output_probe_stack_range_stack_clash): New.
13214 (rs6000_emit_prologue): Emit notes into dump file as requested.
13215 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13216 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13217 Add additional operand and pass it to output_probe_stack_range.
13219 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13221 PR tree-optimization/82163
13222 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13223 (checking_verify_loop_closed_ssa): New parameter.
13224 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13225 (check_loop_closed_ssa_stmt): Delete.
13226 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13227 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13228 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13231 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13233 * brig-builtins.def: Treat HSAIL barrier builtins as
13234 setjmp/longjump style functions.
13236 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13238 * target.def (constant_alignment): New hook.
13239 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13240 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13241 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13242 * doc/tm.texi: Regenerate.
13243 * targhooks.h (default_constant_alignment): Declare.
13244 (constant_alignment_word_strings): Likewise.
13245 * targhooks.c (default_constant_alignment): New function.
13246 (constant_alignment_word_strings): Likewise.
13247 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13248 instead of CONSTANT_ALIGNMENT.
13249 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13250 (force_const_mem): Likewise.
13251 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13252 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13253 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13254 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13255 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13257 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13258 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13259 constant_alignment_word_strings.
13260 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13261 (CONSTANT_ALIGNMENT): Likewise.
13262 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13263 (arm_constant_alignment): New function.
13264 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13265 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13266 constant_alignment_word_strings.
13267 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13268 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13269 constant_alignment_word_strings.
13270 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13271 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13272 (cris_constant_alignment): New function.
13273 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13274 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13275 (epiphany_constant_alignment): New function.
13276 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13277 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13278 constant_alignment_word_strings.
13279 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13280 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13281 constant_alignment_word_strings.
13282 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13283 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13284 constant_alignment_word_strings.
13285 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13286 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13287 * config/i386/i386.c (ix86_constant_alignment): Make static.
13288 Use the same interface as the target hook.
13289 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13290 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13291 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13292 constant_alignment_word_strings.
13293 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13294 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13295 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13296 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13297 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13298 constant_alignment_word_strings.
13299 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13300 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13301 constant_alignment_word_strings.
13302 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13303 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13304 constant_alignment_word_strings.
13305 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13306 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13308 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13309 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13310 * config/mips/mips.c (mips_constant_alignment): New function.
13311 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13312 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13313 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13314 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13315 (mmix_constant_alignment): Make static. Use the same interface
13316 as the target hook.
13317 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13318 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13319 constant_alignment_word_strings.
13320 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13321 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13322 constant_alignment_word_strings.
13323 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13324 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13325 constant_alignment_word_strings.
13326 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13327 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13328 (rs6000_constant_alignment): New function.
13329 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13330 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13331 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13332 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13333 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13334 (rs6000_constant_alignment): New function.
13335 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13336 * config/s390/s390.c (s390_constant_alignment): New function.
13337 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13338 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13339 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13340 constant_alignment_word_strings.
13341 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13342 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13343 (sparc_constant_alignment): New function.
13344 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13345 * config/spu/spu.c (spu_constant_alignment): New function.
13346 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13347 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13348 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13349 constant_alignment_word_strings.
13350 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13351 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13352 constant_alignment_word_strings.
13353 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13354 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13355 constant_alignment_word_strings.
13356 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13357 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13358 (visium_constant_alignment): New function.
13359 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13360 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13361 (xtensa_constant_alignment): New function.
13362 * system.h (CONSTANT_ALIGNMENT): Poison.
13364 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13366 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13367 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13368 (rs6000_builtin_valid_without_lhs): New helper function.
13369 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13370 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13372 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13374 * target.h (vec_perm_indices): Use unsigned short rather than
13376 (auto_vec_perm_indices): Likewise.
13377 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13378 Use unsigned int rather than unsigned char.
13379 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13381 2017-09-25 Richard Biener <rguenther@suse.de>
13383 * cfgloop.h (sort_sibling_loops): Declare.
13384 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13385 (sort_sibling_loops): New function sorting the sibling loop list
13387 * graphite.c (graphite_transform_loops): Sort sibling loops.
13389 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13391 * target.def (vec_perm_const_ok): Change sel parameter to
13393 * optabs-query.c (can_vec_perm_p): Update accordingly.
13394 * doc/tm.texi: Regenerate.
13395 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13396 auto_vec_perm_indices and remove separate nelt field.
13397 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13398 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13399 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13400 (aarch64_expand_vec_perm_const): Update accordingly.
13401 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13402 to vec_perm_indices.
13403 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13404 auto_vec_perm_indices and remove separate nelt field.
13405 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13406 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13407 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13409 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13410 to vec_perm_indices.
13411 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13412 sel to vec_perm_indices.
13413 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13414 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13415 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13417 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13420 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13423 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13424 on the FUNCTION_DECL function context if it has a DIE that is a
13427 2017-09-25 Richard Biener <rguenther@suse.de>
13429 PR tree-optimization/82285
13430 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13433 2017-09-25 Tom de Vries <tom@codesourcery.com>
13437 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13440 2017-09-25 Richard Biener <rguenther@suse.de>
13442 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13443 ISL errors other than isl_error_quota happen. Dump if the
13444 schedule is the same.
13445 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13446 errors instead of aborting inside ISL.
13448 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13451 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13452 of libgcc_eh for m64.
13453 * config/i386/darwin64.h: Likewise.
13455 2017-09-25 Richard Biener <rguenther@suse.de>
13457 PR middle-end/82144
13458 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13459 attribute for incomplete types nor twice for complete ones.
13461 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13464 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13465 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13468 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13471 * config/i386/i386.opt (mprefer-avx256): Use
13472 ix86_target_flags variable.
13473 * config/i386/i386.c (ix86_target_string): Move
13474 -mprefer-avx256 to flag2_opts.
13476 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13478 PR middle-end/35691
13479 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13480 and x != -1 | y != -1 into (x & y) != -1.
13482 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13484 * config.gcc: Add new case statement to set
13485 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13486 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13487 s390x-*-linux* case statements. Added aarch64 to the list of
13488 supported architectures.
13490 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13492 PR tree-optimization/82289
13493 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13494 STMT_VINFO_RELEVANT_P.
13496 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13497 Alan Hayward <alan.hayward@arm.com>
13498 David Sherwood <david.sherwood@arm.com>
13500 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13501 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13502 (extract_range_from_binary_expr_1): Don't call
13503 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13505 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13506 Alan Hayward <alan.hayward@arm.com>
13507 David Sherwood <david.sherwood@arm.com>
13509 * target.def (preferred_vector_alignment): New hook.
13510 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13512 * doc/tm.texi: Regenerate.
13513 * targhooks.h (default_preferred_vector_alignment): Declare.
13514 * targhooks.c (default_preferred_vector_alignment): New function.
13515 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13517 (DR_TARGET_ALIGNMENT): New macro.
13518 (aligned_access_p): Update commentary.
13519 (vect_known_alignment_in_bytes): New function.
13520 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13522 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13523 Calculate the misalignment based on the target alignment rather than
13525 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13526 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13527 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13528 the target alignment, rather than masking the element misalignment
13529 with the number of elements in a vector. Also use the target
13530 alignment when calculating the maximum number of peels.
13531 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13532 instead of TYPE_ALIGN_UNIT.
13533 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13534 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13535 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13536 (vect_create_data_ref_ptr): Likewise.
13537 (vect_setup_realignment): Realign by ANDing with
13538 -DR_TARGET_MISALIGNMENT.
13539 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13540 the number of peels based on DR_TARGET_ALIGNMENT.
13541 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13542 with the guaranteed alignment boundary when deciding whether
13544 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13545 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13546 (ensure_base_align): Remove stmt_info parameter. Get the
13547 target base alignment from DR_TARGET_ALIGNMENT.
13548 (vectorizable_store): Update call accordingly. Interpret
13549 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13551 (vectorizable_load): Likewise.
13553 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13554 Alan Hayward <alan.hayward@arm.com>
13555 David Sherwood <david.sherwood@arm.com>
13557 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13558 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13559 (vect_enhance_data_refs_alignment): Likewise.
13561 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13563 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13564 error. Only quit immediately if parsing is complete.
13565 (BEGIN): Initialize fatal_err and parse_done.
13566 (begin fpu, end fpu): Check number of arguments.
13567 (begin arch, end arch): Likewise.
13568 (begin cpu, end cpu): Likewise.
13569 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13570 (optalias): Likewise.
13572 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13574 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13575 * config/arm/arm-isa.h: Delete. Move definitions to ...
13576 * arm-cpus.in: ... here. Use new feature and fgroup values.
13577 * config/arm/arm.c (arm_option_override): Use lower case for feature
13579 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13580 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13581 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13583 (print_isa_bits_for): New function.
13584 (gen_isa): New function.
13585 (gen_comm_data): Use print_isa_bits_for.
13586 (define feature): New keyword.
13587 (define fgroup): New keyword.
13588 * config/arm/t-arm (TM_H): Remove.
13589 (GTM_H): Add arm-isa.h.
13590 (arm-isa.h): Add rule to generate file.
13591 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13592 case for feature bit names.
13594 2017-09-22 Richard Biener <rguenther@suse.de>
13596 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13598 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13599 print of no dependency loops ...
13600 * graphite.c (graphite_transform_loops): ... here.
13601 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13603 (same_close_phi_node, remove_duplicate_close_phi,
13604 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13605 (canonicalize_loop_closed_ssa): ... here and simplify.
13606 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13607 (optimize_isl): Use dump_printf_loc to tell when we stopped
13608 optimizing because of an ISL timeout.
13610 2017-09-22 Richard Biener <rguenther@suse.de>
13612 PR tree-optimization/82291
13613 * tree-if-conv.c (predicate_mem_writes): Make sure to
13614 remove writes in blocks predicated with false.
13616 2017-09-22 Richard Biener <rguenther@suse.de>
13618 * sese.c: Include cfganal.h.
13619 (if_region_set_false_region): Remove.
13620 (create_if_region_on_edge): Likewise.
13621 (move_sese_in_condition): Re-implement without destroying
13624 2017-09-22 Richard Biener <rguenther@suse.de>
13626 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13627 Verify both BBs contain loop PHI nodes before dispatching to
13628 copy_loop_phi_args.
13629 (graphite_regenerate_ast_isl): Do not recompute dominators,
13630 do not verify three times. Restructure for clarity.
13631 * graphite-scop-detection.c (same_close_phi_node,
13632 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13633 defined_in_loop_p, canonicalize_loop_closed_ssa,
13634 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13635 checking and SSA rewrite, move to ...
13636 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13637 (graphite_initialize): Do not pass in ctx, do not reset the
13638 SCEV cache, compute only dominators.
13639 (graphite_transform_loops): Allocate ISL ctx after
13640 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13641 Maintain post-dominators only around build_scops.
13642 * sese.c (if_region_set_false_region): Make static. Free
13643 and recompute dominators.
13644 (move_sese_in_condition): Assert we don't get called with
13645 post-dominators computed.
13646 * sese.h (if_region_set_false_region): Remove.
13648 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13650 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13651 mode attribute for TARGET_AVX512VL.
13653 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13655 * config/i386/i386.opt (mprefer-avx256): New option.
13656 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13658 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13659 for TARGET_PREFER_AVX256.
13660 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13662 2017-09-21 Jeff Law <law@redhat.com>
13664 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13665 Fix dump output if the only stack space is for pushed registers.
13667 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13669 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13672 2017-09-21 Martin Sebor <msebor@redhat.com>
13675 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13676 code (in C++) or code that triggers warnings.
13678 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13680 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13682 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13684 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13685 * sched-rgn.c: Ditto.
13686 * sel-sched-ir.c: Ditto.
13688 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13690 * toplev.h (set_random_seed): Adjust return type.
13691 * toplev.c (init_local_tick): Move eager initialization of random_seed
13692 to get_random_seed. Adjust comment.
13693 (init_random_seed): Inline to get_random_seed, delete.
13694 (get_random_seed): Initialize random_seed lazily.
13695 (set_random_seed): Do not return previous value.
13696 (print_switch_value): Do not call get_random_seed.
13698 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13700 * cgraph.c (delete_function_version): New, broken out from...
13701 (cgraph_node::delete_function_version): ...here. Rename to
13702 cgraph_node::delete_function_version_by_decl. Update all uses.
13703 (cgraph_node::remove): Call delete_function_version.
13705 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13708 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13709 VAR_DECLs to which addressable non-volatile parameters are mapped
13710 and for id->retvar after the return value assignment. Clear
13711 id->retval and id->retbnd after inlining.
13713 2017-09-21 Richard Biener <rguenther@suse.de>
13715 PR tree-optimization/82276
13716 PR tree-optimization/82244
13717 * tree-vrp.c (build_assert_expr_for): Set
13718 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13720 (remove_range_assertions): Revert earlier change.
13722 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13725 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13727 2017-09-21 Richard Biener <rguenther@suse.de>
13729 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13730 Restore valid IL after code generation errors.
13731 * graphite.c (graphite_transform_loops): Diagnose code
13732 generation issues as MSG_MISSED_OPTIMIZATION and continue
13733 with processing SCOPs.
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 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13740 rather than choosing between plus_constant and gen_rtx_<CODE>.
13741 * expr.c (emit_push_insn): Likewise.
13742 (expand_expr_real_2): Likewise.
13744 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13745 Alan Hayward <alan.hayward@arm.com>
13746 David Sherwood <david.sherwood@arm.com>
13748 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13750 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13751 Alan Hayward <alan.hayward@arm.com>
13752 David Sherwood <david.sherwood@arm.com>
13754 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13755 calling tree_to_uhwi.
13757 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13758 Alan Hayward <alan.hayward@arm.com>
13759 David Sherwood <david.sherwood@arm.com>
13761 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13762 INTEGER_CST rather than a negative test for ADDR_EXPR.
13764 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13765 Alan Hayward <alan.hayward@arm.com>
13766 David Sherwood <david.sherwood@arm.com>
13768 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13769 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13771 2017-09-21 Richard Biener <rguenther@suse.de>
13773 PR tree-optimization/71351
13774 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13775 graphite_create_new_loop_guard): Remove, fold remaining parts
13777 (translate_isl_ast_node_for): ... here and simplify.
13779 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13782 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13783 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13784 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13785 alternative always use QI mode, for -Os imov (=R,R) alternative
13786 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13789 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13790 Jeff Law <law@redhat.com>
13792 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13793 (allocate_stack_space): New function, partially extracted from
13794 s390_emit_prologue.
13795 (s390_emit_prologue): Track offset to most recent stack probe.
13796 Code to allocate space moved into allocate_stack_space.
13797 Dump actions when no stack is allocated.
13798 (s390_prologue_plus_offset): New function.
13799 (s390_emit_stack_probe): Likewise.
13801 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13803 * common.opt (Wa, Wl, Wp, g, gz=): Add
13805 (gno-column-info): Remove.
13806 (gcolumn-info): Drop RejectNegative.
13807 (gno-): New prefix.
13808 (gno-record-gcc-switches): Remove.
13809 (grecord-gcc-switches): Drop RejectNegative.
13810 (gno-split-dwarf): Remove.
13811 (gsplit-dwarf): Drop RejectNegative.
13812 (gno-strict-dwarf): Remove.
13813 (gstrict-dwarf): Drop RejectNegative.
13814 * config/darwin.opt (gfull, gused): Add RejectNegative.
13815 * dwarf2out.c (gen_producer_string): Drop
13816 gno-record-gcc-switches handler.
13817 * optc-gen.awk: Add g to prefixes with negative forms.
13818 * opts-common.c (remapping_prefix_p): New.
13819 (find_opt): Check it.
13820 (generate_canonical_option): Test g prefix.
13821 (option_map): Add -gno- mapping.
13822 (add_misspelling_candidates): Check remapping_prefix_p.
13824 2017-09-20 Jeff Law <law@redhat.com>
13826 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13827 thinko in stack clash protection support.
13829 * explow.c (compute_stack_clash_protection_loop_data): Use
13830 CONST_INT_P instead of explicit test. Verify object is a
13831 CONST_INT_P before looking at INTVAL.
13832 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13833 instead of explicit test.
13835 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13838 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13839 address instead of to r1 and r11.
13841 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13843 * config.gcc: Support "knm".
13844 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13845 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13847 * config/i386/i386.c (m_KNM): Define.
13848 (processor_target_table): Add "knm".
13850 (ix86_option_override_internal): Add "knm".
13851 (ix86_issue_rate): Add PROCESSOR_KNM.
13852 (ix86_adjust_cost): Ditto.
13853 (ia32_multipass_dfa_lookahead): Ditto.
13854 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13855 (fold_builtin_cpu): Add M_INTEL_KNM.
13856 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13857 (processor_type): Add PROCESSOR_KNM.
13858 * config/i386/x86-tune.def: Add m_KNM.
13859 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13861 2017-09-20 Richard Biener <rguenther@suse.de>
13863 PR tree-optimization/80213
13864 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13865 are allowed in empty BBs as well.
13866 (canonicalize_loop_closed_ssa): Also look for other complex
13868 (scop_detection::get_sese): Include the loop-closed PHI block
13870 (scop_detection::merge_sese): Remove code adding extra blocks.
13871 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13872 (build_scops): Assert the final returned scop is invalid.
13874 2017-09-20 Richard Biener <rguenther@suse.de>
13876 PR tree-optimization/82264
13877 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13879 (vn_phi_lookup): Likewise.
13880 (vn_phi_insert): Likewise.
13882 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13884 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13885 that fits into uhwi or shwi, add DW_AT_const_value regardless
13886 of early_dwarf without going through RTL, using add_AT_unsigned
13889 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13890 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13891 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13892 (DEBUG_MACRO_SECTION): Likewise.
13893 (DEBUG_LTO_MACRO_SECTION): Likewise.
13894 (DEBUG_STR_DWO_SECTION): Likewise.
13895 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13896 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13897 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13898 (DEBUG_LTO_LINE_STR_SECTION): Define.
13899 (init_sections_and_labels): Initialize debug_line_str_section
13900 variable. Initialize debug_loc_section for -gdwarf-5 to
13901 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13903 2017-09-20 Richard Biener <rguenther@suse.de>
13905 * graphite-sese-to-poly.c (extract_affine): Properly handle
13906 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13908 2017-09-20 Richard Biener <rguenther@suse.de>
13910 PR tree-optimization/81373
13911 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13912 Force SESE live-out defs to be handled even if they are
13915 2017-09-19 Jeff Law <law@redhat.com>
13917 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13918 nothing for stack adjustments with REG_STACK_CHECK.
13919 * sched-deps.c (parse_add_or_inc): Reject insns with
13920 REG_STACK_CHECK from dependency breaking.
13921 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13922 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13923 * reg-notes.def (STACK_CHECK): New note.
13925 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13926 (ix86_expand_prologue): Dump stack clash info as needed.
13927 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13929 * function.c (dump_stack_clash_frame_info): New function.
13930 * function.h (dump_stack_clash_frame_info): Prototype.
13931 (enum stack_clash_probes): New enum.
13933 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13934 flag_stack_clash_protection.
13935 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13936 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13937 (arm_frame_pointer_required): Likewise.
13938 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13939 (ia64_expand_prologue): Likewise.
13940 * config/mips/mips.c (mips_expand_prologue): Likewise.
13941 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13942 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13943 (sparc_flat_expand_prologue): Likewise.
13944 * config/spu/spu.c (spu_expand_prologue): Likewise.
13946 * explow.c: Include "params.h".
13947 (anti_adjust_stack_and_probe_stack_clash): New function.
13948 (get_stack_check_protect): Likewise.
13949 (compute_stack_clash_protection_loop_data): Likewise.
13950 (emit_stack_clash_protection_loop_start): Likewise.
13951 (emit_stack_clash_protection_loop_end): Likewise.
13952 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13953 Use anti_adjust_stack_and_probe_stack_clash.
13954 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13955 (emit_stack_clash_protection_loop_start): Likewise.
13956 (emit_stack_clash_protection_loop_end): Likewise.
13957 * rtl.h (get_stack_check_protect): Prototype.
13958 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13959 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13960 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13962 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13964 * doc/tm.texi: Rebuilt.
13965 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13966 get_stack_check_protect.
13967 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13968 * config/arm/arm.c (arm_expand_prologue): Likewise.
13969 (arm_frame_pointer_required): Likewise.
13970 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13971 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13972 * config/mips/mips.c (mips_expand_prologue): Likewise.
13973 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13974 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13975 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13976 (sparc_flat_expand_prologue): Likewise.
13978 * common.opt (-fstack-clash-protection): New option.
13979 * flag-types.h (enum stack_check_type): Note difference between
13980 -fstack-check= and -fstack-clash-protection.
13981 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13982 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13983 * toplev.c (process_options): Issue warnings/errors for cases
13984 not handled with -fstack-clash-protection.
13985 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13986 (-fstack-check): Note additional problem with -fstack-check=generic.
13987 Note that -fstack-check is primarily for Ada and refer users
13988 to -fstack-clash-protection for stack-clash-protection.
13989 Document new params for stack clash protection.
13991 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13993 * config/i386/i386.c (ix86_split_long_move): Do not handle
13994 address used for LEA in a special way.
13996 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13998 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14000 2017-09-19 Martin Sebor <msebor@redhat.com>
14003 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14004 of incompatible types.
14006 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14008 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14009 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14010 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14011 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14013 2017-09-19 Richard Biener <rguenther@suse.de>
14015 PR tree-optimization/82244
14016 * tree-vrp.c (remove_range_assertions): Do not propagate
14017 a constant to abnormals but replace the assert with a copy.
14019 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14021 PR rtl-optimization/57878
14022 PR rtl-optimization/68988
14023 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14024 avoidance test involving non_reload_pseudos. Move frequency test
14025 below the general fragmentation avoidance test.
14027 2017-09-19 Richard Biener <rguenther@suse.de>
14029 PR tree-optimization/69728
14030 * graphite-sese-to-poly.c (schedule_error): New global.
14031 (add_loop_schedule): Handle empty domain by failing the
14033 (build_original_schedule): Handle schedule_error.
14035 2017-09-19 Richard Biener <rguenther@suse.de>
14037 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14038 Do not iterate to sibling loops but only to siblings of inner
14041 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14044 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14045 that the registers are different.
14047 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14049 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14050 to processor_model and "amdfam17h" to arch_names_table.
14051 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14053 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14056 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14058 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14059 Alan Hayward <alan.hayward@arm.com>
14060 David Sherwood <david.sherwood@arm.com>
14062 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14064 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14066 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14067 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14068 vect_schedule_slp_instance.
14069 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14070 Update call to vect_slp_analyze_node_operations. Simplify return
14072 (vect_slp_analyze_bb_1): Update call accordingly.
14073 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14074 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14075 (vect_schedule_slp): Update call accordingly.
14077 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14078 Alan Hayward <alan.hayward@arm.com>
14079 David Sherwood <david.sherwood@arm.com>
14081 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14082 with types that aren't in fact scalar.
14084 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14086 * tree-vect-slp.c (vect_record_max_nunits): New function,
14088 (vect_build_slp_tree_1): ...here.
14089 (vect_build_slp_tree_2): Call it for phis too.
14091 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14093 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14094 to vect_get_vec_def_for_operand when getting the mask operand.
14096 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14097 Alan Hayward <alan.hayward@arm.com>
14098 David Sherwood <david.sherwood@arm.com>
14100 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14103 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14104 Alan Hayward <alan.hayward@arm.com>
14105 David Sherwood <david.sherwood@arm.com>
14107 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14108 calculation for vector booleans.
14110 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14111 Alan Hayward <alan.hayward@arm.com>
14112 David Sherwood <david.sherwood@arm.com>
14114 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14116 (vect_transform_stmt): ...here.
14117 (vect_analyze_stmt): Use it instead of calling
14118 vectorizable_live_operation directly.
14120 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14122 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14123 non-SIMT targets in acc vector loops.
14125 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14127 * configure.ac: Add arc and check if assembler supports gdwarf2.
14128 * configure: Regenerate.
14130 2017-09-18 Richard Biener <rguenther@suse.de>
14132 PR tree-optimization/82220
14133 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14134 epilogue niters from the min_profitable_iters compute.
14136 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14139 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14140 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14141 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14143 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14146 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14147 switching to a new text section.
14149 2017-09-18 Richard Biener <rguenther@suse.de>
14151 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14153 (build_alias_set): Reject aliases with no access function.
14155 2017-09-18 Richard Biener <rguenther@suse.de>
14157 PR tree-optimization/79622
14158 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14160 (build_cross_bb_scalars_use): Likewise.
14162 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14164 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14166 2017-09-18 Alan Modra <amodra@gmail.com>
14169 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14170 stack_pointer_rtx for count 0. Update comments. Break up
14171 large rtl expression.
14173 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14175 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14176 Increase to 20 bytes.
14177 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14178 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14179 or avx version of the stub.
14181 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14184 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14185 compute the minimum stack alignment. Also update preferred stack
14186 boundary for leaf functions.
14188 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14190 PR tree-optimization/82228
14191 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14194 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14196 * common/config/nds32/nds32-common.c
14197 (nds32_option_optimization_table): Refine formatting.
14198 (nds32_option_optimization_table): Use -fsched-pressure and
14199 -fomit-frame-pointer for specific optimization level.
14201 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14203 * config/nds32/nds32.c: Refine formatting and comments.
14204 * config/nds32/nds32.h: Likewise.
14205 * config/nds32/nds32.md: Likewise.
14206 * config/nds32/nds32-cost.c: Likewise.
14207 * config/nds32/nds32-isr.c: Likewise.
14208 * config/nds32/nds32-md-auxiliary.c: Likewise.
14209 * config/nds32/nds32-multiple.md: Likewise.
14210 * config/nds32/nds32-predicates.c: Likewise.
14212 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14213 Jakub Jelinek <jakub@redhat.com>
14215 Add support for -std=c++2a.
14216 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14218 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14220 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14223 * doc/extend.texi (Common Function Attributes): Add
14224 references to ARM, AArch64, and S/390 specific attributes.
14225 (Function Specific Option Pragmas): Add AArch64 and S/390
14226 to list of back ends that support the target pragma.
14228 2017-09-15 Nathan Sidwell <nathan@acm.org>
14230 * doc/standards.texi: Fix C++17 description. Update URLs for
14233 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14235 * common.opt (Wcast-align=strict): New warning option.
14236 * doc/invoke.texi: Document -Wcast-align=strict.
14238 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14240 * cgraph.h (cgraph_thunk_info): Add comments.
14241 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14242 assert for VIRTUAL_* arguments stricter.
14244 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14246 PR tree-optimization/71026
14247 * match.pd: Move RDIV patterns from fold-const.c
14248 * fold-const.c (distribute_real_division): Removed.
14249 (fold_binary_loc): Remove calls to distribute_real_divison.
14251 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14253 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14254 c++1z and gnu++1z as deprecated. Change other references to
14255 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14256 Change -Wc++1z-compat to -Wc++17-compat.
14257 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14258 * dwarf2out.c (highest_c_language): Handle C++17.
14259 (gen_compile_unit_die): Likewise.
14261 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14263 PR rtl-optimization/82192
14264 * combine.c (make_extraction): Don't look through non-paradoxical
14265 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14268 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14269 Alan Hayward <alan.hayward@arm.com>
14270 David Sherwood <david.sherwood@arm.com>
14272 * target.def (function_arg_offset): New hook.
14273 * targhooks.h (default_function_arg_offset): Declare.
14274 * targhooks.c (default_function_arg_offset): New function.
14275 * function.c (locate_and_pad_parm): Use
14276 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14277 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14278 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14279 * doc/tm.texi: Regenerate.
14280 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14281 * config/spu/spu.c (spu_function_arg_offset): New function.
14282 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14283 * system.h (FUNCTION_ARG_OFFSET): Poison.
14285 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14286 Alan Hayard <alan.hayward@arm.com>
14287 David Sherwood <david.sherwood@arm.com>
14289 * target.def (truly_noop_truncation): New hook.
14290 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14291 than TRULY_NOOP_TRUNCATION.
14292 * hooks.h (hook_bool_uint_uint_true): Declare.
14293 * hooks.c (hook_bool_uint_uint_true): New function.
14294 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14295 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14296 * doc/tm.texi: Regenerate.
14297 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14298 rather than TRULY_NOOP_TRUNCATION in comments.
14299 (simplify_comparison): Likewise.
14300 (record_truncated_value): Likewise.
14301 * expmed.c (extract_bit_field_1): Likewise.
14302 (extract_split_bit_field): Likewise.
14303 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14304 instead of TRULY_NOOP_TRUNCATION.
14305 * function.c (assign_parm_setup_block): Likewise.
14306 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14307 * rtlhooks.c: Include target.h.
14308 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14309 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14310 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14311 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14312 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14313 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14314 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14315 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14316 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14317 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14318 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14319 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14320 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14321 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14322 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14323 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14324 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14325 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14326 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14327 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14328 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14329 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14330 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14331 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14332 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14333 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14334 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14335 rather than TRULY_NOOP_TRUNCATION in comments.
14336 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14337 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14338 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14340 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14341 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14342 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14343 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14344 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14345 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14346 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14347 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14348 rather than TRULY_NOOP_TRUNCATION in comments.
14349 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14350 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14351 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14352 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14353 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14354 TRULY_NOOP_TRUNCATION condition.
14355 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14356 (TRULY_NOOP_TRUNCATION): Delete.
14357 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14358 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14359 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14360 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14361 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14362 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14363 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14364 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14365 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14366 rather than TRULY_NOOP_TRUNCATION in comments.
14367 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14368 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14369 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14370 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14371 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14372 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14374 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14377 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14378 (*cmp_ior): Likewise.
14379 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14380 (*ior_scc_scc_cmp): Likewise.
14381 (*and_scc_scc): Likewise.
14382 (*and_scc_scc_cmp): Likewise.
14384 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14385 Alan Hayard <alan.hayward@arm.com>
14386 David Sherwood <david.sherwood@arm.com>
14388 * target.def (can_change_mode_class): New hook.
14389 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14390 (hard_regno_nregs): Likewise.
14391 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14392 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14393 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14394 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14395 (LOAD_EXTEND_OP): Update accordingly.
14396 * doc/tm.texi: Regenerate.
14397 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14398 CANNOT_CHANGE_MODE_CLASS.
14399 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14400 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14401 * combine.c (simplify_set): Update accordingly.
14402 * emit-rtl.c (validate_subreg): Likewise.
14403 * recog.c (general_operand): Likewise.
14404 * regcprop.c (mode_change_ok): Likewise.
14405 * reload1.c (choose_reload_regs): Likewise.
14406 (inherit_piecemeal_p): Likewise.
14407 * rtlanal.c (simplify_subreg_regno): Likewise.
14408 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14409 instead of CANNOT_CHANGE_MODE_CLASS.
14410 (reload_cse_simplify_operands): Likewise.
14411 * reload.c (push_reload): Use targetm.can_change_mode_class
14412 instead of CANNOT_CHANGE_MODE_CLASS.
14413 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14414 REG_CANNOT_CHANGE_MODE_P.
14415 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14416 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14417 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14418 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14419 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14420 (arm_can_change_mode_class): New function.
14421 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14422 than CANNOT_CHANGE_MODE_CLASS in comments.
14423 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14424 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14425 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14426 (ix86_can_change_mode_class): ...this new function, inverting the
14427 sense of the return value.
14428 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14429 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14430 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14431 (ia64_can_change_mode_class): New function.
14432 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14433 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14434 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14435 (m32c_can_change_mode_class): ...this new function, inverting the
14436 sense of the return value.
14437 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14438 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14439 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14440 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14441 (mips_can_change_mode_class): ...this new function, inverting the
14442 sense of the return value.
14443 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14444 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14445 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14446 (msp430_can_change_mode_class): New function.
14447 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14448 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14449 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14450 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14451 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14452 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14453 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14454 (pa_cannot_change_mode_class): Replace with...
14455 (pa_can_change_mode_class): ...this new function, inverting the
14456 sense of the return value.
14457 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14458 than CANNOT_CHANGE_MODE_CLASS in comments.
14459 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14460 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14461 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14462 (pdp11_cannot_change_mode_class): Replace with...
14463 (pdp11_can_change_mode_class): ...this new function, inverting the
14464 sense of the return value.
14465 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14466 * config/powerpcspe/powerpcspe-protos.h
14467 (rs6000_cannot_change_mode_class_ptr): Delete.
14468 * config/powerpcspe/powerpcspe.c
14469 (rs6000_cannot_change_mode_class_ptr): Delete.
14470 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14471 (rs6000_option_override_internal): Assign to
14472 targetm.can_change_mode_class instead of
14473 rs6000_cannot_change_mode_class_ptr.
14474 (rs6000_cannot_change_mode_class): Replace with...
14475 (rs6000_can_change_mode_class): ...this new function, inverting the
14476 sense of the return value.
14477 (rs6000_debug_cannot_change_mode_class): Replace with...
14478 (rs6000_debug_can_change_mode_class): ...this new function.
14479 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14480 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14481 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14482 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14483 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14485 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14486 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14487 (rs6000_option_override_internal): Assign to
14488 targetm.can_change_mode_class instead of
14489 rs6000_cannot_change_mode_class_ptr.
14490 (rs6000_cannot_change_mode_class): Replace with...
14491 (rs6000_can_change_mode_class): ...this new function, inverting the
14492 sense of the return value.
14493 (rs6000_debug_cannot_change_mode_class): Replace with...
14494 (rs6000_debug_can_change_mode_class): ...this new function.
14495 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14496 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14497 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14498 (s390_can_change_mode_class): ...this new function, inverting the
14499 sense of the return value.
14500 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14501 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14502 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14503 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14504 (sh_cannot_change_mode_class): Replace with...
14505 (sh_can_change_mode_class): ...this new function, inverting the
14506 sense of the return value.
14507 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14508 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14509 (sparc_can_change_mode_class): New function.
14510 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14511 * config/spu/spu.c (spu_can_change_mode_class): New function.
14512 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14513 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14514 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14515 (visium_can_change_mode_class): New function.
14516 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14518 2017-09-15 Richard Biener <rguenther@suse.de>
14520 PR tree-optimization/82217
14521 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14522 but not undefined case.
14524 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14527 * postreload.c (reload_cse_simplify_operands): Skip
14528 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14530 2017-09-15 Richard Biener <rguenther@suse.de>
14532 PR tree-optimization/68823
14533 * graphite-scop-detection.c (build_alias_set): If we have a
14534 possible dependence check whether we can handle them by just
14535 looking at the DRs DR_ACCESS_FNs.
14536 (build_scops): If build_alias_set fails, fail the SCOP.
14538 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14540 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14541 to support float128 built-in functions that require the ISA 3.0
14543 (BU_FLOAT128_3_HW): Likewise.
14544 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14545 built-in functions.
14546 (FMAF128): Likewise.
14548 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14549 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14550 floating point instructions.
14551 (rs6000_invalid_builtin): Likewise.
14552 (rs6000_builtin_mask_names): Likewise.
14553 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14554 (RS6000_BTM_FLOAT128_HW): Likewise.
14555 (RS6000_BTM_COMMON): Likewise.
14556 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14558 * doc/extend.texi (RS/6000 built-in functions): Document the
14559 IEEE 128-bit floating point square root and fused multiply-add
14560 built-in functions.
14562 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14564 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14565 reg (r2) isn't in the set of registers defined in the prologue.
14567 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14568 Alan Hayward <alan.hayward@arm.com>
14569 David Sherwood <david.sherwood@arm.com>
14571 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14572 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14573 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14574 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14575 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14577 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14578 max_vectorization_factor.
14579 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14581 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayward <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14586 (vect_worthwhile_without_simd_p): Declare.
14587 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14588 (vectorizable_reduction): Use it.
14589 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14590 (vectorizable_operation): Likewise.
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 * tree-vectorizer.h (vect_get_num_copies): New function.
14597 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14598 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14599 (vectorizable_induction): Likewise.
14600 (vectorizable_live_operation): Likewise.
14601 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14602 (vectorizable_bswap): Likewise.
14603 (vectorizable_call): Likewise.
14604 (vectorizable_conversion): Likewise.
14605 (vectorizable_assignment): Likewise.
14606 (vectorizable_shift): Likewise.
14607 (vectorizable_operation): Likewise.
14608 (vectorizable_store): Likewise.
14609 (vectorizable_load): Likewise.
14610 (vectorizable_condition): Likewise.
14611 (vectorizable_comparison): Likewise.
14612 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14614 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14615 Alan Hayward <alan.hayward@arm.com>
14616 David Sherwood <david.sherwood@arm.com>
14618 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14619 of vect_init_vector.
14621 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14622 Alan Hayward <alan.hayward@arm.com>
14623 David Sherwood <david.sherwood@arm.com>
14625 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14626 an inline wrapper that provides a location.
14627 (gimple_build_vector): Likewise.
14628 * gimple-fold.c (gimple_build_vector_from_val): New function.
14629 (gimple_build_vector): Likewise.
14630 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14631 functions to build the initial value. Always return a gimple value.
14632 (get_initial_defs_for_reduction): Likewise. Only compute
14634 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14635 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14636 (vectorizable_induction): Use gimple_build_vector rather than
14639 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14640 Alan Hayward <alan.hayward@arm.com>
14641 David Sherwood <david.sherwood@arm.com>
14643 * target.h (vec_perm_indices): New typedef.
14644 (auto_vec_perm_indices): Likewise.
14645 * optabs-query.h: Include target.h
14646 (can_vec_perm_p): Take a vec_perm_indices *.
14647 * optabs-query.c (can_vec_perm_p): Likewise.
14648 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14649 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14650 * tree-vect-generic.c (lower_vec_perm): Likewise.
14651 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14652 (vect_grouped_load_supported): Likewise.
14653 (vect_shift_permute_load_chain): Likewise.
14654 (vect_permute_store_chain): Use auto_vec_perm_indices.
14655 (vect_permute_load_chain): Likewise.
14656 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14657 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14658 Update uses of can_vec_perm_p.
14659 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14660 mode with a number of elements. Take a vec_perm_indices *.
14661 (vect_create_epilog_for_reduction): Update accordingly.
14662 Use auto_vec_perm_indices.
14663 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14664 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14665 (vect_transform_slp_perm_load): Likewise.
14666 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14667 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14668 (vect_gen_perm_mask_checked): Likewise.
14669 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14670 (vect_gen_perm_mask_checked): Likewise.
14671 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14672 (vectorizable_store): Likewise.
14673 (vectorizable_load): Likewise.
14674 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14675 (vectorizable_bswap): Likewise.
14677 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14678 Alan Hayward <alan.hayward@arm.com>
14679 David Sherwood <david.sherwood@arm.com>
14681 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14682 * tree.c (build_vector): Likewise.
14683 (build_vector_from_ctor): Update accordingly.
14684 (build_vector_from_val): Likewise.
14685 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14686 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14687 * tree-vect-generic.c (add_rshift): Likewise.
14688 (expand_vector_divmod): Likewise.
14689 (optimize_vector_constructor): Likewise.
14690 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14691 (vect_transform_slp_perm_load): Likewise.
14692 (vect_schedule_slp_instance): Likewise.
14693 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14694 (vectorizable_call): Likewise.
14695 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14696 * expmed.c (make_tree): Likewise.
14697 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14698 a vector passed to build_vector.
14699 (fold_convert_const): Likewise.
14700 (exact_inverse): Likewise.
14701 (fold_ternary_loc): Likewise.
14702 (fold_relational_const): Likewise.
14703 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14704 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14705 (const_unop): Likewise. Store the reduction accumulator in a
14706 variable rather than an array.
14707 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14708 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14709 the new vector, rather than constructing it after the input arrays.
14710 (native_interpret_vector): Use auto_vec<tree> when building
14711 a vector passed to build_vector. Add elements in order.
14712 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14713 auto_vec<tree> when building a vector passed to build_vector.
14714 (vect_create_epilog_for_reduction): Likewise.
14715 (vectorizable_induction): Likewise.
14716 (get_initial_def_for_reduction): Likewise. Fix indentation of
14718 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14720 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14721 passed to build_vector.
14723 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14724 Alan Hayward <alan.hayward@arm.com>
14725 David Sherwood <david.sherwood@arm.com>
14727 * tree-core.h (tree_base::u): Add an "nelts" field.
14728 (tree_vector): Use VECTOR_CST_NELTS as the length.
14729 * tree.c (tree_size): Likewise.
14730 (make_vector): Initialize VECTOR_CST_NELTS.
14731 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14732 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14733 TYPE_VECTOR_SUBPARTS.
14734 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14735 as the number of units, setting it from VECTOR_CST_NELTS.
14736 (const_vector_from_tree): Likewise.
14737 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14738 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14739 (fold_negate_expr_1): Likewise.
14740 (fold_convert_const): Likewise.
14741 (const_binop): Likewise. Differentiate the number of output and
14743 (const_unop): Likewise.
14744 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14745 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14746 in cases that did the opposite.
14748 2017-09-14 Richard Biener <rguenther@suse.de>
14750 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14753 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14755 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14757 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14760 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14761 if and where to split a bb, except for splitting before debug insn
14762 sequences followed by non-label real insn. Delete debug insns
14763 in between basic blocks.
14765 * combine.c (make_compound_operation_int): Formatting fixes.
14767 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14768 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14769 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14770 * config/sol2.h (LINK_EH_SPEC): Likewise.
14771 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14772 * config/s390/linux.h (LINK_SPEC): Likewise.
14773 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14774 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14775 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14776 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14777 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14778 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14779 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14780 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14782 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14784 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14786 (ENDFILE_LINUX_SPEC): Likewise.
14787 (LINK_EH_SPEC): Likewise.
14788 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14789 (LINK_OS_LINUX_SPEC32): Likewise.
14790 (LINK_OS_LINUX_SPEC64): Likewise.
14791 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14792 (LINK_OS_LINUX_SPEC): Likewise.
14794 2017-09-13 Martin Liska <mliska@suse.cz>
14796 PR middle-end/82154
14797 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14798 CASE_HIGH is NULL_TREE.
14800 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14801 Alan Hayward <alan.hayward@arm.com>
14802 David Sherwood <david.sherwood@arm.com>
14804 * target.def (secondary_memory_needed): New hook.
14805 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14806 instead of SECONDARY_MEMORY_NEEDED.
14807 (secondary_memory_needed_mode): Likewise.
14808 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14809 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14810 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14811 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14812 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14813 * doc/tm.texi: Regenerate.
14814 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14815 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14816 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14817 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14818 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14819 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14820 mode argument first and change the reg_class arguments to reg_class_t.
14821 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14822 Make static. Update the call to inline_secondary_memory_needed.
14823 (ix86_register_move_cost): Update the call to
14824 inline_secondary_memory_needed.
14825 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14826 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14828 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14829 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14831 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14832 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14833 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14834 and match hook interface. Add comment from mips.h.
14835 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14836 * config/mmix/mmix.md (truncdfsf2): Refer to
14837 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14839 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14840 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14841 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14842 (pa_secondary_memory_needed): New function.
14843 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14844 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14845 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14846 (pdp11_secondary_memory_needed): Make static and match hook interface.
14847 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14848 * config/powerpcspe/powerpcspe-protos.h
14849 (rs6000_secondary_memory_needed_ptr): Delete.
14850 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14852 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14853 (rs6000_option_override_internal): Assign to
14854 targetm.secondary_memory_needed rather than
14855 rs6000_secondary_memory_needed_ptr.
14856 (rs6000_secondary_memory_needed): Match hook interface.
14857 (rs6000_debug_secondary_memory_needed): Likewise.
14858 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14859 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14860 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14861 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14862 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14863 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14865 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14866 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14867 (rs6000_option_override_internal): Assign to
14868 targetm.secondary_memory_needed rather than
14869 rs6000_secondary_memory_needed_ptr.
14870 (rs6000_secondary_memory_needed): Match hook interface.
14871 (rs6000_debug_secondary_memory_needed): Likewise.
14872 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14873 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14874 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14875 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14876 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14877 (sparc_secondary_memory_needed): New function.
14878 * lra-constraints.c (check_and_process_move): Refer to
14879 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14881 (curr_insn_transform): Likewise.
14882 (process_alt_operands): Use targetm.secondary_memory_needed
14883 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14884 (check_secondary_memory_needed_p): Likewise.
14885 (choose_split_class): Likewise.
14886 * reload.c: Unconditionally include code that was previously
14887 conditional on SECONDARY_MEMORY_NEEDED.
14888 (push_secondary_reload): Use targetm.secondary_memory_needed
14889 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14890 (push_reload): Likewise.
14891 * reload1.c: Unconditionally include code that was previously
14892 conditional on SECONDARY_MEMORY_NEEDED.
14893 (choose_reload_regs): Use targetm.secondary_memory_needed
14894 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14895 (gen_reload): Likewise.
14896 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14898 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14899 Alan Hayward <alan.hayward@arm.com>
14900 David Sherwood <david.sherwood@arm.com>
14902 * target.def (secondary_memory_needed_mode): New hook:
14903 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14904 * targhooks.h (default_secondary_memory_needed_mode): New function.
14905 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14906 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14907 * doc/tm.texi: Regenerate.
14908 * lra-constraints.c (check_and_process_move): Use
14909 targetm.secondary_memory_needed_mode instead of
14910 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14911 (curr_insn_transform): Likewise.
14912 * reload.c (get_secondary_mem): Likewise.
14913 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14914 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14916 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14917 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14918 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14919 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14920 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14922 * config/powerpcspe/powerpcspe-protos.h
14923 (rs6000_secondary_memory_needed_mode): Delete.
14924 * config/powerpcspe/powerpcspe.c
14925 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14926 (rs6000_secondary_memory_needed_mode): Make static.
14927 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14928 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14930 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14932 (rs6000_secondary_memory_needed_mode): Make static.
14933 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14934 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14935 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14936 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14937 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14939 (sparc_secondary_memory_needed_mode): New function.
14940 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14942 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14944 * config/aarch64/constraints.md (Umq): New constraint.
14945 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14947 (mov<mode>): Update condition.
14949 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14951 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14952 when bitposition is the same.
14954 2017-09-13 Richard Biener <rguenther@suse.de>
14956 * dwarf2out.c (output_die_symbol): Remove.
14957 (output_die): Do not output a DIEs symbol.
14959 2017-09-13 Richard Biener <rguenther@suse.de>
14961 PR middle-end/82128
14962 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14963 default-def to avoid breaking iterator update with the weird
14964 interaction with cgraph_update_edges_for_call_stmt_node.
14966 2017-09-13 Richard Biener <rguenther@suse.de>
14968 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14969 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14970 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14971 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14973 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14975 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14976 Disable pc relative literal load irrespective of
14977 TARGET_FIX_ERR_A53_84341 for default.
14979 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14981 * config/sparc/sparc.c (output_return): Output the source location of
14982 the insn in the delay slot, if any.
14983 (output_sibcall): Likewise.
14985 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14988 * common.opt (-static-pie): New alias.
14989 (shared): Negate static-pie.
14990 (-no-pie): Update help text.
14992 (static-pie): New option.
14993 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14994 -static-pie support.
14995 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14996 (LINK_EH_SPEC): Likewise.
14997 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14998 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14999 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15000 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15001 (init_gcc_specs): Likewise.
15002 (init_spec): Likewise.
15003 (display_help): Update help message for -pie.
15004 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15007 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15009 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15010 (movdi_aarch64): Likewise.
15011 (movti_aarch64): Likewise.
15013 2017-09-12 Simon Wright <simon@pushface.org>
15016 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15017 calculation of the minor version, always output as 0.
15019 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15022 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15023 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15024 on it early, rather than manual conversion late. For
15025 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15026 instead of performing manual conversion.
15028 2017-09-12 Carl Love <cel@us.ibm.com>
15030 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15031 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15033 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15034 VMULOSW): Add definitions.
15035 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15036 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15037 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15038 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15039 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15041 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15043 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15045 (movti_aarch64): Likewise.
15046 (movdf_aarch64): Likewise.
15047 (movtf_aarch64): Likewise.
15048 (load_pairdi): Likewise.
15049 (store_pairdi): Likewise.
15050 (load_pairdf): Likewise.
15051 (store_pairdf): Likewise.
15052 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15053 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15054 (ldr_got_small_<mode>): Likewise.
15055 (ldr_got_small_28k_<mode>): Likewise.
15056 (ldr_got_tiny): Likewise.
15057 * config/aarch64/iterators.md (ldst_sz): New.
15058 (ldpstp_sz): Likewise.
15059 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15061 (thunderx_load): Split load_8 to load_16.
15062 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15064 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15065 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15066 (xgene1_store_pair): Split store_8 to store_16.
15067 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15068 (falkor_st_0_st_sd): Split store_8 to store_16.
15070 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15072 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15073 and store1/2/3/4 to store_4/8/12/16.
15074 * config/aarch64/aarch64.md: Update for rename.
15075 * config/arm/arm.md: Likewise.: Likewise.
15076 * config/arm/arm.c: Likewise.
15077 * config/arm/thumb1.md: Likewise.
15078 * config/arm/thumb2.md: Likewise.
15079 * config/arm/vfp.md: Likewise.
15080 * config/arm/arm-generic.md: Likewise.
15081 * config/arm/arm1020e.md: Likewise.
15082 * config/arm/arm1026ejs.md: Likewise.
15083 * config/arm/arm1136jfs.md: Likewise.
15084 * config/arm/arm926ejs.md: Likewise.
15085 * config/arm/cortex-a15.md: Likewise.
15086 * config/arm/cortex-a17.md: Likewise.
15087 * config/arm/cortex-a5.md: Likewise.
15088 * config/arm/cortex-a53.md: Likewise.
15089 * config/arm/cortex-a57.md: Likewise.
15090 * config/arm/cortex-a7.md: Likewise.
15091 * config/arm/cortex-a8.md: Likewise.
15092 * config/arm/cortex-a9.md: Likewise.
15093 * config/arm/cortex-m4.md: Likewise.
15094 * config/arm/cortex-m7.md: Likewise.
15095 * config/arm/cortex-r4.md: Likewise.
15096 * config/arm/exynos-m1.md: Likewise.
15097 * config/arm/fa526.md: Likewise.
15098 * config/arm/fa606te.md: Likewise.
15099 * config/arm/fa626te.md: Likewise.
15100 * config/arm/fa726te.md: Likewise.
15101 * config/arm/fmp626.md: Likewise.
15102 * config/arm/iwmmxt.md: Likewise.
15103 * config/arm/ldmstm.md: Likewise.
15104 * config/arm/marvell-pj4.md: Likewise.
15105 * config/arm/xgene1.md: Likewise.
15106 * config/aarch64/thunderx.md: Likewise.
15107 * config/aarch64/thunderx2t99.md: Likewise.
15108 * config/aarch64/falkor.md: Likewise.
15110 2017-09-12 Martin Liska <mliska@suse.cz>
15112 * attribs.c (private_lookup_attribute): New function.
15113 * attribs.h (private_lookup_attribute): Declared here.
15114 (lookup_attribute): Called from this place.
15116 2017-09-12 Richard Biener <rguenther@suse.de>
15118 PR tree-optimization/82157
15119 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15120 stmts with side-effects.
15122 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15123 Alan Hayward <alan.hayward@arm.com>
15124 David Sherwood <david.sherwood@arm.com>
15126 * target.def (hard_regno_nregs): New hook.
15127 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15128 * targhooks.h (default_hard_regno_nregs): Declare.
15129 * targhooks.c (default_hard_regno_nregs): New function.
15130 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15131 (TARGET_HARD_REGNO_NREGS): ...this hook.
15132 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15133 (CLASS_MAX_NREGS): Likewise.
15134 * doc/tm.texi: Regenerate.
15135 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15136 instead of HARD_REGNO_NREGS.
15137 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15138 HARD_REGNO_NREGS in the comment.
15139 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15140 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15141 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15142 Return an unsigned int.
15143 (TARGET_HARD_REGNO_NREGS): Redefine.
15144 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15145 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15146 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15147 (arc_hard_regno_nregs): New function.
15148 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15149 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15150 (arm_hard_regno_nregs): New function.
15151 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15152 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15153 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15154 (TARGET_HARD_REGNO_NREGS): Redefine.
15155 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15156 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15157 (HARD_REGNO_NREGS): Delete.
15158 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15159 (cr16_hard_regno_nregs): New function.
15160 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15161 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15162 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15163 (cris_hard_regno_nregs): New function.
15164 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15165 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15166 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15167 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15168 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15169 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15170 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15171 (frv_hard_regno_nregs): Make static. Take and return an
15173 (frv_class_max_nregs): Remove outdated copy of documentation.
15174 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15175 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15176 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15177 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15178 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15179 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15180 (TARGET_HARD_REGNO_NREGS): Redefine.
15181 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15182 (CLASS_MAX_NREGS): Update comment.
15183 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15184 (ia64_hard_regno_nregs): New function.
15185 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15186 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15187 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15188 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15189 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15191 (m32c_hard_regno_nregs): Likewise. Make static.
15192 (TARGET_HARD_REGNO_NREGS): Redefine.
15193 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15194 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15195 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15196 (m68k_hard_regno_nregs): New function.
15197 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15198 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15199 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15200 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15201 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15202 Take and return an unsigned int.
15203 (TARGET_HARD_REGNO_NREGS): Redefine.
15204 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15205 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15206 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15207 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15208 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15209 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15210 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15211 (msp430_hard_regno_nregs): Make static. Take and return an
15213 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15214 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15215 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15216 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15217 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15218 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15219 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15220 (TARGET_HARD_REGNO_NREGS): Redefine.
15221 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15222 (PA_HARD_REGNO_NREGS): ...this.
15223 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15224 (PA_HARD_REGNO_NREGS): ...this.
15225 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15226 (pa_hard_regno_nregs): New function.
15227 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15228 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15229 (pdp11_hard_regno_nregs): New function.
15230 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15231 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15232 (rs6000_hard_regno_nregs_hook): New function.
15233 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15234 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15235 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15236 Take and return an unsigned int. Move earlier in file.
15237 (TARGET_HARD_REGNO_NREGS): Redefine.
15238 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15239 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15240 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15241 (rl78_hard_regno_nregs): Make static. Take and return an
15243 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15244 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15245 (rs6000_hard_regno_nregs_hook): New function.
15246 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15247 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15248 (TARGET_HARD_REGNO_NREGS): Redefine.
15249 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15250 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15251 instead of HARD_REGNO_NREGS.
15252 (s390_hard_regno_nregs): New function.
15253 (s390_hard_regno_mode_ok): Add comment from s390.h.
15254 (TARGET_HARD_REGNO_NREGS): Redefine.
15255 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15256 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15257 (sh_hard_regno_nregs): New function.
15258 (sh_pass_in_reg_p): Use it.
15259 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15260 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15261 (sparc_hard_regno_nregs): New function.
15262 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15263 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15264 (spu_function_arg_advance): Use it, supplying a valid register number.
15265 (TARGET_HARD_REGNO_NREGS): Redefine.
15266 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15267 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15268 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15269 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15270 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15271 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15272 (CLASS_MAX_NREGS): Remove copy of old documentation.
15273 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15274 (visium_hard_regno_nregs): New function.
15275 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15276 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15277 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15278 xtensa_hard_regno_nregs): New function.
15279 * system.h (HARD_REGNO_NREGS): Poison.
15281 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15283 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15284 hard_regno_nregs instead of HARD_REGNO_NREGS.
15285 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15286 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15287 (c6x_expand_epilogue): Likewise.
15288 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15289 (frv_read_iacc_argument): Likewise.
15290 * config/sh/sh.c: Include regs.h.
15291 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15292 (regs_used): Likewise.
15293 (output_stack_adjust): Likewise.
15294 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15295 * expmed.c: Include regs.h.
15296 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15297 * ree.c: Include regs.h.
15298 (combine_reaching_defs): Use hard_regno_nregs instead of
15300 (add_removable_extension): Likewise.
15302 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15304 * regs.h (hard_regno_nregs): Turn into a function.
15305 (end_hard_regno): Update accordingly.
15306 * caller-save.c (setup_save_areas): Likewise.
15307 (save_call_clobbered_regs): Likewise.
15308 (replace_reg_with_saved_mem): Likewise.
15309 (insert_restore): Likewise.
15310 (insert_save): Likewise.
15311 * combine.c (can_change_dest_mode): Likewise.
15312 (move_deaths): Likewise.
15313 (distribute_notes): Likewise.
15314 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15315 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15316 (rs6000_split_multireg_move): Likewise.
15317 (rs6000_register_move_cost): Likewise.
15318 (rs6000_memory_move_cost): Likewise.
15319 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15320 (rs6000_split_multireg_move): Likewise.
15321 (rs6000_register_move_cost): Likewise.
15322 (rs6000_memory_move_cost): Likewise.
15323 * cselib.c (cselib_reset_table): Likewise.
15324 (cselib_lookup_1): Likewise.
15325 * emit-rtl.c (set_mode_and_regno): Likewise.
15326 * function.c (aggregate_value_p): Likewise.
15327 * ira-color.c (setup_profitable_hard_regs): Likewise.
15328 (check_hard_reg_p): Likewise.
15329 (calculate_saved_nregs): Likewise.
15330 (assign_hard_reg): Likewise.
15331 (improve_allocation): Likewise.
15332 (calculate_spill_cost): Likewise.
15333 * ira-emit.c (modify_move_list): Likewise.
15334 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15335 (ira_hard_reg_in_set_p): Likewise.
15336 * ira.c (setup_reg_mode_hard_regset): Likewise.
15337 (clarify_prohibited_class_mode_regs): Likewise.
15338 (check_allocation): Likewise.
15339 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15340 (lra_setup_reg_renumber): Likewise.
15341 (setup_try_hard_regno_pseudos): Likewise.
15342 (spill_for): Likewise.
15343 (assign_hard_regno): Likewise.
15344 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15345 * lra-constraints.c (in_class_p): Likewise.
15346 (lra_constraint_offset): Likewise.
15347 (simplify_operand_subreg): Likewise.
15348 (lra_constraints): Likewise.
15349 (split_reg): Likewise.
15350 (split_if_necessary): Likewise.
15351 (invariant_p): Likewise.
15352 (inherit_in_ebb): Likewise.
15353 * lra-lives.c (process_bb_lives): Likewise.
15354 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15355 (get_hard_regs): Likewise.
15356 (do_remat): Likewise.
15357 * lra-spills.c (assign_spill_hard_regs): Likewise.
15358 * mode-switching.c (create_pre_exit): Likewise.
15359 * postreload.c (reload_combine_recognize_pattern): Likewise.
15360 * recog.c (peep2_find_free_register): Likewise.
15361 * regcprop.c (kill_value_regno): Likewise.
15362 (set_value_regno): Likewise.
15363 (copy_value): Likewise.
15364 (maybe_mode_change): Likewise.
15365 (find_oldest_value_reg): Likewise.
15366 (copyprop_hardreg_forward_1): Likewise.
15367 * regrename.c (check_new_reg_p): Likewise.
15368 (regrename_do_replace): Likewise.
15369 * reload.c (push_reload): Likewise.
15370 (combine_reloads): Likewise.
15371 (find_dummy_reload): Likewise.
15372 (operands_match_p): Likewise.
15373 (find_reloads): Likewise.
15374 (find_equiv_reg): Likewise.
15375 (reload_adjust_reg_for_mode): Likewise.
15376 * reload1.c (count_pseudo): Likewise.
15377 (count_spilled_pseudo): Likewise.
15378 (find_reg): Likewise.
15379 (clear_reload_reg_in_use): Likewise.
15380 (free_for_value_p): Likewise.
15381 (allocate_reload_reg): Likewise.
15382 (choose_reload_regs): Likewise.
15383 (reload_adjust_reg_for_temp): Likewise.
15384 (emit_reload_insns): Likewise.
15385 (delete_output_reload): Likewise.
15386 * rtlanal.c (subreg_get_info): Likewise.
15387 * sched-deps.c (sched_analyze_reg): Likewise.
15388 * sel-sched.c (init_regs_for_mode): Likewise.
15389 (mark_unavailable_hard_regs): Likewise.
15390 (choose_best_reg_1): Likewise.
15391 (verify_target_availability): Likewise.
15392 * valtrack.c (dead_debug_insert_temp): Likewise.
15393 * var-tracking.c (track_loc_p): Likewise.
15394 (emit_note_insn_var_location): Likewise.
15395 * varasm.c (make_decl_rtl): Likewise.
15396 * reginfo.c (choose_hard_reg_mode): Likewise.
15397 (init_reg_modes_target): Refer directly to
15398 this_target_regs->x_hard_regno_nregs.
15400 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15402 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15403 instead of hard_regno_nregs.
15405 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15407 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15408 end_hard_regno instead of hard_regno_nregs.
15409 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15410 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15411 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15412 * ira-color.c (improve_allocation): Likewise.
15413 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15414 * lra-lives.c (mark_regno_live): Likewise.
15415 (mark_regno_dead): Likewise.
15416 * lra-remat.c (operand_to_remat): Likewise.
15417 * lra.c (collect_non_operand_hard_regs): Likewise.
15418 * postreload.c (reload_combine_note_store): Likewise.
15419 (move2add_valid_value_p): Likewise.
15420 * reload.c (regno_clobbered_p): Likewise.
15422 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15424 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15426 * config/v850/v850.c (v850_reorg): Likewise.
15427 * reload.c (refers_to_regno_for_reload_p): Likewise.
15428 (find_equiv_reg): Likewise.
15429 * reload1.c (reload_reg_reaches_end_p): Likewise.
15431 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15433 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15435 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15436 * config/arm/arm.c (output_move_neon): Likewise.
15437 (arm_attr_length_move_neon): Likewise.
15438 (neon_split_vcombine): Likewise.
15439 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15440 (c6x_mark_reg_written): Likewise.
15441 (c6x_dwarf_register_span): Likewise.
15442 * config/i386/i386.c (ix86_save_reg): Likewise.
15443 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15444 (rws_access_reg): Likewise.
15445 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15446 * mode-switching.c (create_pre_exit): Likewise.
15447 * ree.c (combine_reaching_defs): Likewise.
15448 (add_removable_extension): Likewise.
15449 * regcprop.c (find_oldest_value_reg): Likewise.
15450 (copyprop_hardreg_forward_1): Likewise.
15451 * reload.c (reload_inner_reg_of_subreg): Likewise.
15452 (push_reload): Likewise.
15453 (combine_reloads): Likewise.
15454 (find_dummy_reload): Likewise.
15455 (reload_adjust_reg_for_mode): Likewise.
15456 * reload1.c (find_reload_regs): Likewise.
15457 (forget_old_reloads_1): Likewise.
15458 (reload_reg_free_for_value_p): Likewise.
15459 (reload_adjust_reg_for_temp): Likewise.
15460 (emit_reload_insns): Likewise.
15461 (delete_output_reload): Likewise.
15462 * sel-sched.c (choose_best_reg_1): Likewise.
15463 (choose_best_pseudo_reg): Likewise.
15465 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15466 Alan Hayward <alan.hayward@arm.com>
15467 David Sherwood <david.sherwood@arm.com>
15469 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15470 * target.def (slow_unaligned_access): New hook.
15471 * targhooks.h (default_slow_unaligned_access): Declare.
15472 * targhooks.c (default_slow_unaligned_access): New function.
15473 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15474 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15475 * doc/tm.texi: Regenerate.
15476 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15477 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15478 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15480 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15481 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15483 (rs6000_slow_unaligned_access): New function.
15484 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15485 (expand_block_compare): Likewise.
15486 (expand_strn_compare): Likewise.
15487 (rs6000_rtx_costs): Likewise.
15488 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15489 (riscv_slow_unaligned_access): Likewise.
15490 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15491 (riscv_slow_unaligned_access_p): ...this and make static.
15492 (riscv_option_override): Update accordingly.
15493 (riscv_slow_unaligned_access): New function.
15494 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15495 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15496 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15497 (rs6000_slow_unaligned_access): New function.
15498 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15499 (rs6000_rtx_costs): Likewise.
15500 * config/rs6000/rs6000-string.c (expand_block_compare)
15501 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15502 of SLOW_UNALIGNED_ACCESS.
15503 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15504 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15505 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15506 of SLOW_UNALIGNED_ACCESS.
15507 * expmed.c (simple_mem_bitfield_p): Likewise.
15508 * expr.c (alignment_for_piecewise_move): Likewise.
15509 (emit_group_load_1): Likewise.
15510 (emit_group_store): Likewise.
15511 (copy_blkmode_from_reg): Likewise.
15512 (emit_push_insn): Likewise.
15513 (expand_assignment): Likewise.
15514 (store_field): Likewise.
15515 (expand_expr_real_1): Likewise.
15516 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15517 * lra-constraints.c (simplify_operand_subreg): Likewise.
15518 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15519 * gimple-ssa-store-merging.c: Likewise in block comment at start
15521 * tree-ssa-strlen.c: Include target.h.
15522 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15523 of SLOW_UNALIGNED_ACCESS.
15524 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15526 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15528 PR rtl-optimization/82185
15529 * expmed.c (emit_store_flag_int): Only test tem if it has been
15532 2017-09-12 Richard Biener <rguenther@suse.de>
15534 PR middle-end/82149
15535 * match.pd ((FTYPE) N CMP CST): Fix typo.
15537 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15539 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15541 (mips_near_type_p): Add 'short_call' attribute as a synonym
15543 * doc/extend.texi (short_call): Document new function attribute.
15545 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15548 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15549 assertion check that in the condition.
15550 (get_atomic_generic_size): Likewise. Before testing if parameter
15551 has pointer type, if it has array type, call for C++
15552 default_conversion to perform array-to-pointer conversion.
15554 2017-09-12 Richard Biener <rguenther@suse.de>
15556 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15557 for operations we cannot scalarize.
15559 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15561 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15562 vectors heap vectors. Clean up comments.
15563 Make visited_bbs a reference.
15564 (profitable_jump_thread_path): Make GC
15565 vectors heap vectors. Clean up comments.
15567 (convert_and_register_jump_thread_path): Make GC vectors heap
15569 (check_subpath_and_update_thread_path): Same. Clean up comments.
15570 Make visited_bbs a reference.
15571 (handle_phi): Abstract common code to to
15572 register_jump_thread_path_if_profitable.
15573 Rename VAR_BB to DEF_BB.
15575 Make GC vectors heap vectors.
15576 Make visited_bbs a reference.
15577 (handle_assignment): Same.
15578 (register_jump_thread_path_if_profitable): New.
15579 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15581 Make GC vectors heap vectors. Clean up comments.
15582 Make visited_bbs a reference.
15583 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15584 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15587 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15590 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15591 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15593 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15595 Revert r251800 and r251799.
15597 2017-09-11 Martin Jambor <mjambor@suse.cz>
15600 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15601 arguments in advance.
15602 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15603 use it to find predecessor edges.
15604 (naive_outof_ssa): Collect vector of predecessors.
15606 2017-09-08 Jason Merrill <jason@redhat.com>
15608 PR c++/70029 - ICE with ref-qualifier and -flto
15609 * langhooks.h (struct lang_hooks_for_types): Add
15610 copy_lang_qualifiers.
15611 * attribs.c (build_type_attribute_qual_variant): Use it.
15612 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15614 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15615 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15617 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15620 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15621 (*mulsi3_sp64): New instruction.
15622 (mulsi3): New expander.
15624 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15626 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15628 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15630 * sancov.c: Include memmodel.h.
15632 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15635 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15638 2017-09-07 Carl Love <cel@us.ibm.com>
15640 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15641 the sldi instruction.
15643 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15645 * sancov.c: Include tm_p.h.
15647 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15650 * output.h (switch_to_other_text_partition): New declaration.
15651 * varasm.c (switch_to_other_text_partition): New function.
15652 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15653 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15654 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15655 to the other text partition before emitting LCL label and switch back
15656 after emitting the word after it.
15658 2017-09-07 Richard Biener <rguenther@suse.de>
15660 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15661 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15662 critical edges here, after loop init.
15663 (pass_data_pre): Remove PROP_no_crit_edges flags.
15664 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15665 for valueization of call args to avoid leaking VN_TOP.
15666 (visit_use): Assert we do not visit default defs.
15667 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15668 Use error_mark_node to more easily detect leaking VN_TOP.
15669 All default-defs are varying, not VN_TOP. Mark them visited.
15670 (run_scc_vn): Make code match comment.
15672 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15674 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15675 OPTION_MASK_FLOAT128_KEYWORD.
15676 (POWERPC_MASKS): Likewise.
15677 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15678 support for the -mfloat128-type option, and make -mfloat128
15679 default on PowerPC Linux systems. Define or undefine
15680 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15681 Define __float128 to be __ieee128 if IEEE 128-bit support is
15682 enabled, or undefine it.
15683 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15684 Delete defining __FLOAT128_TYPE__.
15685 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15686 -mfloat128-type option and make -mfloat128 default on PowerPC
15688 (TARGET_FLOAT128_TYPE): Likewise.
15689 (-mfloat128-type): Likewise.
15690 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15691 Delete the -mfloat128-type option and make -mfloat128 default on
15692 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15693 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15694 128-bit floating point is enabled. Change tests from using
15695 -mfloat128-type to -mfloat128.
15696 (rs6000_mangle_type): Use the correct mangling for the __float128
15697 type even if normal long double is restricted to 64-bits.
15698 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15700 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15701 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15702 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15704 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15705 documentation for -mfloat128.
15707 2017-09-06 Olivier Hainque <hainque@adacore.com>
15709 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15711 2017-09-06 Wish Wu <wishwu007@gmail.com>
15712 Jakub Jelinek <jakub@redhat.com>
15714 * asan.c (initialize_sanitizer_builtins): Add
15715 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15716 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15717 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15718 BT_FN_VOID_UINT64_PTR variables.
15719 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15720 (BT_FN_VOID_UINT16_UINT16): Likewise.
15721 (BT_FN_VOID_UINT32_UINT32): Likewise.
15722 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15723 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15724 (BT_FN_VOID_UINT64_PTR): Likewise.
15725 * common.opt (flag_sanitize_coverage): New variable.
15726 (fsanitize-coverage=trace-pc): Remove.
15727 (fsanitize-coverage=): Add.
15728 * flag-types.h (enum sanitize_coverage_code): New enum.
15729 * fold-const.c (fold_range_test): Disable non-short-circuit
15730 optimization if flag_sanitize_coverage.
15731 (fold_truth_andor): Likewise.
15732 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15733 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15734 (coverage_sanitizer_opts): New array.
15735 (get_closest_sanitizer_option): Add OPTS argument, handle also
15736 OPT_fsanitize_coverage_.
15737 (parse_sanitizer_options): Adjusted to also handle
15738 OPT_fsanitize_coverage_.
15739 (common_handle_option): Add OPT_fsanitize_coverage_.
15740 * sancov.c (instrument_comparison, instrument_switch): New function.
15741 (sancov_pass): Add trace-cmp support.
15742 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15743 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15744 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15745 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15746 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15747 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15748 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15749 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15750 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15751 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15753 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15755 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15756 error. Only quit immediately if parsing is complete.
15757 (BEGIN): Initialize fatal_err and parse_done.
15758 (begin fpu, end fpu): Check number of arguments.
15759 (begin arch, end arch): Likewise.
15760 (begin cpu, end cpu): Likewise.
15761 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15762 (optalias): Likewise.
15764 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15766 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15767 * config/arm/arm-isa.h: Delete. Move definitions to ...
15768 * arm-cpus.in: ... here. Use new feature and fgroup values.
15769 * config/arm/arm.c (arm_option_override): Use lower case for feature
15771 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15772 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15773 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15775 (print_isa_bits_for): New function.
15776 (gen_isa): New function.
15777 (gen_comm_data): Use print_isa_bits_for.
15778 (define feature): New keyword.
15779 (define fgroup): New keyword.
15780 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15781 (arm-isa.h): Add rule to generate file.
15782 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15783 case for feature bit names.
15785 2017-09-06 Richard Biener <rguenther@suse.de>
15787 * tree-ssa-pre.c (NECESSARY): Remove.
15788 (create_expression_by_pieces): Do not touch pass-local flags.
15789 (insert_into_preds_of_block): Likewise.
15790 (do_pre_regular_insertion): Likewise.
15791 (eliminate_insert): Likewise.
15792 (eliminate_dom_walker::before_dom_children): Likewise.
15793 (fini_eliminate): Do not look at inserted_exprs.
15794 (mark_operand_necessary): Remove.
15795 (remove_dead_inserted_code): Replace with simple work-list
15796 algorithm based on inserted_exprs and SSA uses.
15797 (pass_pre::execute): Re-order fini_eliminate and
15798 remove_dead_inserted_code.
15800 2017-09-06 Olivier Hainque <hainque@adacore.com>
15802 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15803 for VxWorks 7. Adjust surrounding comments.
15805 2017-09-06 Richard Biener <rguenther@suse.de>
15807 * gimple-ssa-strength-reduction.c
15808 (find_candidates_dom_walker::before_dom_children): Also allow
15811 2017-09-06 Richard Biener <rguenther@suse.de>
15813 PR tree-optimization/82108
15814 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15815 for gap in the non-permutation SLP case.
15817 2017-09-06 Martin Jambor <mjambor@suse.cz>
15819 PR tree-optimization/82078
15820 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15821 add_access_to_work_queue...
15822 (build_accesses_from_assign): ...here.
15823 (propagate_all_subaccesses): Make sure racc is the group
15824 representative, if there is one.
15826 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15828 PR middle-end/82095
15829 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15832 2017-09-06 Richard Biener <rguenther@suse.de>
15834 * gimple-ssa-strength-reduction.c
15835 (find_candidates_dom_walker::before_doom_children): Use a
15836 type and not a mode check.
15838 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15841 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15843 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15845 2017-09-05 Jeff Law <law@redhat.com>
15847 PR tree-optimization/64910
15848 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15849 cases where we have 3 or more operands.
15851 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15853 PR middle-end/81768
15854 * omp-low.c (lower_omp_for): Recompute tree invariant if
15855 gimple_omp_for_initial/final is ADDR_EXPR.
15857 PR middle-end/81768
15858 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15859 into gimple val before gimplification fo the COND_EXPR.
15861 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15863 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15864 REGION_COPY argument.
15865 (thread_through_all_blocks): Remove unused argument to
15866 duplicate_thread_path.
15868 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15869 Alan Hayward <alan.hayward@arm.com>
15870 David Sherwood <david.sherwood@arm.com>
15872 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15873 Take a scalar_mode rather than a machine_mode.
15874 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15875 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15876 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15877 (aarch64_gen_adjusted_ldpstp): Likewise.
15878 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15881 Alan Hayward <alan.hayward@arm.com>
15882 David Sherwood <david.sherwood@arm.com>
15884 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15885 Take a scalar_int_mode instead of a machine_mode.
15886 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15887 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15888 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15889 (aarch64_simd_attr_length_rglist): Delete.
15890 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15891 a scalar_int_mode instead of a machine_mode.
15892 (aarch64_add_offset): Likewise.
15893 (aarch64_internal_mov_immediate): Likewise
15894 (aarch64_add_constant_internal): Likewise.
15895 (aarch64_add_constant): Likewise.
15896 (aarch64_movw_imm): Likewise.
15897 (aarch64_rtx_arith_op_extract_p): Likewise.
15898 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15899 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15900 Remove assert that the mode isn't a vector.
15901 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15902 (aarch64_expand_mov_immediate): Update calls after above changes.
15903 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15904 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15905 (aarch64_move_imm): Likewise.
15906 (aarch64_can_const_movi_rtx_p): Likewise.
15907 (aarch64_strip_extend): Likewise.
15908 (aarch64_extr_rtx_p): Likewise.
15909 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15910 a CONST_INT when the mode parameter is VOIDmode.
15911 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15913 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15915 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15916 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15917 (bitwise_type_for_mode): Update accordingly.
15919 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15921 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15922 * stor-layout.c (mode_for_size_tree): Likewise.
15923 (mode_for_array): Update accordingly.
15924 (layout_decl): Likewise.
15925 (compute_record_mode): Likewise. Only set the mode once.
15927 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15929 * target.def (get_mask_mode): Change return type to opt_mode.
15931 * doc/tm.texi: Regenerate.
15932 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15933 * targhooks.c (default_get_mask_mode): Likewise.
15934 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15935 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15936 targetm.get_mask_mode.
15937 * tree.c (build_truth_vector_type): Likewise.
15939 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15941 * machmode.h (mode_for_vector): Return an opt_mode.
15942 * stor-layout.c (mode_for_vector): Likewise.
15943 (mode_for_int_vector): Update accordingly.
15944 (layout_type): Likewise.
15945 * config/i386/i386.c (emit_memmov): Likewise.
15946 (ix86_expand_set_or_movmem): Likewise.
15947 (ix86_expand_vector_init): Likewise.
15948 (ix86_get_mask_mode): Likewise.
15949 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15951 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15952 * expmed.c (extract_bit_field_1): Likewise.
15953 * expr.c (expand_expr_real_2): Likewise.
15954 * optabs-query.c (can_vec_perm_p): Likewise.
15955 (can_vec_mask_load_store_p): Likewise.
15956 * optabs.c (expand_vec_perm): Likewise.
15957 * targhooks.c (default_get_mask_mode): Likewise.
15958 * tree-vect-stmts.c (vectorizable_store): Likewise.
15959 (vectorizable_load): Likewise.
15960 (get_vectype_for_scalar_type_and_size): Likewise.
15962 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15964 * machmode.h (mode_for_int_vector): New function.
15965 * stor-layout.c (mode_for_int_vector): Likewise.
15966 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15967 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15968 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15969 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15970 (s390_expand_vcond): Likewise.
15972 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15974 * machmode.h (opt_machine_mode): New type.
15975 (opt_mode<T>): Allow construction from anything that can be
15977 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15978 (mode_for_size): Return an opt_machine_mode.
15979 * stor-layout.c (mode_for_size): Likewise.
15980 (mode_for_size_tree): Update call accordingly.
15981 (bitwise_mode_for_mode): Likewise.
15982 (make_fract_type): Likewise.
15983 (make_accum_type): Likewise.
15984 * caller-save.c (replace_reg_with_saved_mem): Update call
15986 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15987 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15988 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15989 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15990 * expmed.c (extract_bit_field_1): Likewise.
15991 * reload.c (get_secondary_mem): Likewise.
15992 * varasm.c (assemble_integer): Likewise.
15993 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15996 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15998 * machmode.h (decimal_float_mode_for_size): New function.
15999 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16000 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16001 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16002 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16003 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16004 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16006 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16008 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16009 (get_builtin_sync_mode): Likewise.
16010 (expand_ifn_atomic_compare_exchange): Likewise.
16011 (expand_builtin_atomic_clear): Likewise.
16012 (expand_builtin_atomic_test_and_set): Likewise.
16013 (fold_builtin_atomic_always_lock_free): Likewise.
16014 * calls.c (compute_argument_addresses): Likewise.
16015 (emit_library_call_value_1): Likewise.
16016 (store_one_arg): Likewise.
16017 * combine.c (combine_instructions): Likewise.
16018 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16019 * config/arm/arm.c (arm_function_value): Likewise.
16020 (aapcs_allocate_return_reg): Likewise.
16021 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16022 * config/i386/i386.c (construct_container): Likewise.
16023 (ix86_gimplify_va_arg): Likewise.
16024 (ix86_expand_sse_cmp): Likewise.
16025 (emit_memmov): Likewise.
16026 (emit_memset): Likewise.
16027 (expand_small_movmem_or_setmem): Likewise.
16028 (ix86_expand_pextr): Likewise.
16029 (ix86_expand_pinsr): Likewise.
16030 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16031 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16033 * config/mips/mips.c (mips_function_value_1) Likewise.
16034 (mips_block_move_straight): Likewise.
16035 (mips_expand_ins_as_unaligned_store): Likewise.
16036 * config/powerpcspe/powerpcspe.c
16037 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16038 (rs6000_darwin64_record_arg_flush): Likewise.
16039 * config/rs6000/rs6000.c
16040 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16041 (rs6000_darwin64_record_arg_flush): Likewise.
16042 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16043 (sparc_function_value_1): Likewise.
16044 * config/spu/spu.c (adjust_operand): Likewise.
16045 (spu_emit_branch_or_set): Likewise.
16046 (arith_immediate_p): Likewise.
16047 * emit-rtl.c (gen_lowpart_common): Likewise.
16048 * expr.c (expand_expr_real_1): Likewise.
16049 * function.c (assign_parm_setup_block): Likewise.
16050 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16051 * reload1.c (alter_reg): Likewise.
16052 * stor-layout.c (mode_for_vector): Likewise.
16053 (layout_type): Likewise.
16055 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16057 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16058 (spu_convert_move): Likewise.
16059 * lower-subreg.c (resolve_simple_move): Likewise.
16061 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16064 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16065 define_insn to a define_expand.
16066 (altivec_vsum2sws_direct): New define_insn.
16067 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16069 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16071 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16074 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16079 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16080 realign stack if stack alignment needed is less than incoming
16083 2017-09-05 Marek Polacek <polacek@redhat.com>
16086 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16089 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16091 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16093 2017-09-05 Richard Biener <rguenther@suse.de>
16095 PR tree-optimization/82084
16096 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16098 2017-09-05 Richard Biener <rguenther@suse.de>
16100 PR tree-optimization/82102
16101 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16103 2017-09-05 Martin Liska <mliska@suse.cz>
16105 PR tree-optimization/82032
16106 * tree-cfg.c (generate_range_test): New function.
16107 * tree-cfg.h (generate_range_test): Declared here.
16108 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16109 (cleanup_control_expr_graph): Use it.
16110 * tree-switch-conversion.c (try_switch_expansion): Remove
16112 (emit_case_nodes): Use generate_range_test.
16114 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16117 * config/i386/i386.md (*<btsc><mode>_mask): Add
16118 TARGET_USE_BT to insn constraint.
16119 (*btr<mode>_mask): Ditto.
16121 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16123 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16124 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16126 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16129 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16130 TARGET_NEON and TARGET_IWMMXT.
16131 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16132 TARGET_NEON and TARGET_IWMMXT.
16133 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16135 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16137 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16138 (ix86_rewrite_tls_address): Ditto.
16139 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16140 (ix86_rewrite_tls_address_1): Ditto.
16141 (ix86_rewrite_tls_address): Ditto.
16142 * config/i386/predicates.md (tls_address_pattern): New predicate.
16143 * config/i386/i386.md (TLS address splitter): New splitter.
16145 2017-09-04 Richard Biener <rguenther@suse.de>
16147 PR tree-optimization/82084
16148 * fold-const.h (can_native_encode_string_p): Declare.
16149 * fold-const.c (can_native_encode_string_p): Factor out from ...
16150 (native_encode_string): ... here.
16151 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16152 vectorizing stores from constants we later cannot handle.
16154 2017-09-04 Marek Polacek <polacek@redhat.com>
16157 * doc/invoke.texi: Update -Wtautological-compare documentation.
16159 2017-09-04 Jeff Law <law@redhat.com>
16161 PR tree-optimization/64910
16162 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16163 swap the first and last operand if the last is a constant.
16165 2017-09-04 Marek Polacek <polacek@redhat.com>
16168 * convert.c (do_narrow): When sanitizing signed integer overflows,
16169 bail out for signed types.
16170 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16172 2017-09-04 Richard Biener <rguenther@suse.de>
16174 PR tree-optimization/82060
16175 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16176 Move devirtualization after stmt folding and before EH/AB/noreturn
16177 cleanup to get the stmt refs canonicalized. Use a bool instead
16178 of gimple_modified_p since that doesn't work for NOPs. Schedule
16179 NOPs generated by folding for removal.
16181 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16182 Alan Hayward <alan.hayward@arm.com>
16183 David Sherwood <david.sherwood@arm.com>
16185 * coretypes.h (pad_direction): New enum.
16186 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16187 (FUNCTION_ARG_PADDING): Likewise.
16188 * target.def (function_arg_padding): New hook.
16189 * targhooks.h (default_function_arg_padding): Declare.
16190 * targhooks.c (default_function_arg_padding): New function.
16191 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16192 (TARGET_FUNCTION_ARG_PADDING): ...this.
16193 * doc/tm.texi: Regenerate.
16194 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16195 instead of direction.
16196 (compute_argument_addresses): Likewise.
16197 (load_register_parameters): Likewise.
16198 (emit_library_call_value_1): Likewise.
16199 (store_one_arg): Use targetm.calls.function_arg_padding instead
16200 of FUNCTION_ARG_PADDING.
16201 (must_pass_in_stack_var_size_or_pad): Likewise.
16202 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16203 (emit_group_store): Likewise.
16204 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16205 instead of FUNCTION_ARG_PADDING.
16206 (emit_push_insn): Likewise, and propagate enum change throughout
16208 * function.h (direction): Delete.
16209 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16211 * function.c (assign_parm_find_stack_rtl): Likewise.
16212 (assign_parm_setup_block_p): Likewise.
16213 (assign_parm_setup_block): Likewise.
16214 (gimplify_parameters): Likewise.
16215 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16216 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16218 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16219 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16220 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16221 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16222 (aarch64_function_arg_padding): ...this new function.
16223 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16224 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16225 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16226 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16227 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16228 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16229 (arm_pad_arg_upward): Replace with...
16230 (arm_function_arg_padding): ...this new function.
16231 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16233 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16234 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16235 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16236 (ia64_hpux_function_arg_padding): Replace with...
16237 (ia64_function_arg_padding): ...this new function. Use pad_direction
16238 instead of direction. Check for TARGET_HPUX.
16239 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16240 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16241 (iq2000_function_arg_padding): New function.
16242 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16243 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16244 (mips_function_arg_padding): ...this new function.
16245 (mips_pad_reg_upward): Update accordingly.
16246 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16247 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16248 targetm.calls.function_arg_padding.
16249 (FUNCTION_ARG_PADDING): Delete.
16250 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16251 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16252 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16253 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16254 (nios2_block_reg_padding): Return pad_direction instead of direction.
16255 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16256 instead of direction.
16257 (nios2_function_arg_padding): Likewise. Make static.
16258 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16259 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16260 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16261 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16262 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16263 (pa_function_arg_padding): Make static. Return pad_direction instead
16265 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16266 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16267 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16268 instead of direction. Use targetm.calls.function_arg_padding.
16269 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16270 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16271 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16272 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16273 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16275 (function_arg_padding): Rename to...
16276 (rs6000_function_arg_padding): ...this. Make static. Return
16277 pad_direction instead of direction.
16278 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16279 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16280 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16281 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16282 instead of direction. Use targetm.calls.function_arg_padding.
16283 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16284 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16285 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16286 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16287 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16288 (function_arg_padding): Rename to...
16289 (rs6000_function_arg_padding): ...this. Make static. Return
16290 pad_direction instead of direction.
16291 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16292 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16293 * config/s390/s390.c (s390_function_arg_padding): New function.
16294 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16295 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16296 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16297 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16298 (function_arg_padding): Rename to...
16299 (sparc_function_arg_padding): ...this. Make static. Return
16300 pad_direction instead of direction.
16301 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16302 * config/spu/spu.c (spu_function_arg_padding): New function.
16303 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16304 * system.h (FUNCTION_ARG_PADDING): Poison.
16306 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16307 Alan Hayward <alan.hayward@arm.com>
16308 David Sherwood <david.sherwood@arm.com>
16310 * target.def (modes_tieable_p): New hook.
16311 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16312 (TARGET_MODES_TIEABLE_P): ...this.
16313 * doc/tm.texi.in: Regenerate.
16314 * hooks.h (hook_bool_mode_mode_true): Declare.
16315 * hooks.c (hook_bool_mode_mode_true): New function.
16316 * combine.c (subst): Use targetm.modes_tieable_p instead of
16318 * dse.c (find_shift_sequence): Likewise.
16319 * expmed.c (extract_low_bits): Likewise.
16320 * lower-subreg.c: Include target.h.
16321 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16323 * rtlanal.c (rtx_cost): Likewise.
16324 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16325 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16326 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16327 (TARGET_MODES_TIEABLE_P): Redefine.
16328 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16329 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16330 (TARGET_MODES_TIEABLE_P): Redefine.
16331 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16332 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16333 (arc_modes_tieable_p): New function.
16334 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16335 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16336 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16337 (arm_modes_tieable_p): Make static.
16338 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16339 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16340 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16341 (TARGET_MODES_TIEABLE_P): Redefine.
16342 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16343 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16344 (TARGET_MODES_TIEABLE_P): Redefine.
16345 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16346 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16347 (cr16_modes_tieable_p): New function.
16348 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16349 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16350 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16351 (TRULY_NOOP_TRUNCATION): Update comment.
16352 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16353 (TRULY_NOOP_TRUNCATION): Update comment.
16354 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16355 (frv_modes_tieable_p): New function.
16356 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16357 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16358 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16359 (TARGET_MODES_TIEABLE_P): Redefine.
16360 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16361 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16362 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16363 (TARGET_MODES_TIEABLE_P): Redefine.
16364 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16365 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16366 (ia64_modes_tieable_p): New function.
16367 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16368 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16369 (iq2000_modes_tieable_p): New function.
16370 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16371 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16372 (lm32_modes_tieable_p): New function.
16373 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16374 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16375 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16376 (TARGET_MODES_TIEABLE_P): Redefine.
16377 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16378 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16379 (m32r_modes_tieable_p): New function.
16380 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16381 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16382 (m68k_modes_tieable_p): New function.
16383 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16384 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16385 (mcore_modes_tieable_p): New function.
16386 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16387 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16389 (TARGET_MODES_TIEABLE_P): Redefine.
16390 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16391 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16392 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16393 (TARGET_MODES_TIEABLE_P): Redefine.
16394 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16395 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16396 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16397 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16398 (mn10300_modes_tieable_p): ...this and make static.
16399 (TARGET_MODES_TIEABLE_P): Redefine.
16400 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16401 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16402 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16403 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16404 (msp430_modes_tieable_p): Make static.
16405 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16406 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16407 (TARGET_MODES_TIEABLE_P): Redefine.
16408 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16409 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16410 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16411 (TARGET_MODES_TIEABLE_P): Redefine.
16412 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16413 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16414 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16415 (TARGET_MODES_TIEABLE_P): Redefine.
16416 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16417 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16418 (pdp11_modes_tieable_p): New function.
16419 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16420 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16421 (rs6000_modes_tieable_p): New function.
16422 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16423 * config/powerpcspe/powerpcspe.md: Update comment.
16424 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16425 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16426 (TARGET_MODES_TIEABLE_P): Redefine.
16427 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16428 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16429 (rl78_modes_tieable_p): New function.
16430 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16431 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16432 (rs6000_modes_tieable_p): New function.
16433 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16434 * config/rs6000/rs6000.md: Update comment.
16435 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16436 * config/rx/rx.c (rx_modes_tieable_p): New function.
16437 (TARGET_MODES_TIEABLE_P): Redefine.
16438 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16439 * config/s390/s390.c (s390_modes_tieable_p): New function.
16440 (TARGET_MODES_TIEABLE_P): Redefine.
16441 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16442 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16443 (sh_modes_tieable_p): New function.
16444 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16445 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16446 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16447 (sparc_modes_tieable_p): Make static.
16448 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16449 * config/spu/spu.c (spu_modes_tieable_p): New function.
16450 (TARGET_MODES_TIEABLE_P): Redefine.
16451 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16452 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16453 (TARGET_MODES_TIEABLE_P): Redefine.
16454 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16455 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16456 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16457 * config/v850/v850.c (v850_modes_tieable_p): New function.
16458 (TARGET_MODES_TIEABLE_P): Redefine.
16459 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16460 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16461 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16462 (visium_modes_tieable_p): New function.
16463 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16464 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16465 (xtensa_modes_tieable_p): New function.
16466 * system.h (MODES_TIEABLE_P): Poison.
16468 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16469 Alan Hayward <alan.hayward@arm.com>
16470 David Sherwood <david.sherwood@arm.com>
16472 * target.def (hard_regno_mode_ok): New hook.
16473 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16474 (TARGET_HARD_REGNO_MODE_OK): ...this.
16475 * doc/tm.texi.in: Regenerate.
16476 * hooks.h (hook_bool_uint_mode_true): Declare.
16477 * hooks.c (hook_bool_uint_mode_true): New function.
16478 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16479 HARD_REGNO_MODE_OK.
16480 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16481 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16482 instead of HARD_REGNO_MODE_OK.
16483 * caller-save.c: Include target.h.
16484 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16485 HARD_REGNO_MODE_OK.
16486 * combine.c (can_combine_p): Likewise.
16487 (combinable_i3pat): Likewise.
16488 (can_change_dest_mode): Likewise.
16489 * expr.c (init_expr_target): Likewise.
16490 (convert_move): Likewise.
16491 (convert_modes): Likewise.
16492 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16493 (setup_prohibited_mode_move_regs): Likewise.
16494 * ira.h (target_ira): Likewise.
16495 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16496 * lra-constraints.c (process_alt_operands): Likewise.
16497 (split_reg): Likewise.
16498 * recog.c (peep2_find_free_register): Likewise.
16499 * ree.c (combine_reaching_defs): Likewise.
16500 * regcprop.c (maybe_mode_change): Likewise.
16501 * reginfo.c (init_reg_sets_1): Likewise.
16502 (choose_hard_reg_mode): Likewise.
16503 (simplifiable_subregs): Likewise.
16504 * regrename.c (check_new_reg_p): Likewise.
16505 * reload.c (find_valid_class): Likewise.
16506 (find_valid_class_1): Likewise.
16507 (reload_inner_reg_of_subreg): Likewise.
16508 (push_reload): Likewise.
16509 (combine_reloads): Likewise.
16510 (find_dummy_reload): Likewise.
16511 (find_reloads): Likewise.
16512 * reload1.c (find_reg): Likewise.
16513 (set_reload_reg): Likewise.
16514 (allocate_reload_reg): Likewise.
16515 (choose_reload_regs): Likewise.
16516 (reload_adjust_reg_for_temp): Likewise.
16517 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16518 (simplify_subreg_regno): Likewise.
16519 * sel-sched.c (init_regs_for_mode): Likewise.
16520 * varasm.c (make_decl_rtl): Likewise.
16521 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16522 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16523 HARD_REGNO_MODE_OK.
16524 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16525 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16527 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16528 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16530 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16531 (arc_mode_class): Delete.
16532 (HARD_REGNO_MODE_OK): Delete.
16533 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16534 (arc_hard_regno_mode_ok): Rename old array to...
16535 (arc_hard_regno_mode_ok_modes): ...this.
16536 (arc_conditional_register_usage): Update accordingly.
16537 (arc_mode_class): Make static.
16538 (arc_hard_regno_mode_ok): New function.
16539 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16540 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16541 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16542 (arm_hard_regno_mode_ok): Make static.
16543 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16544 HARD_REGNO_MODE_OK.
16545 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16546 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16547 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16550 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16551 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16552 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16553 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16555 * config/bfin/predicates.md (valid_reg_operand): Use
16556 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16557 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16558 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16560 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16561 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16562 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16563 (cr16_hard_regno_mode_ok): Make static and return a bool.
16564 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16565 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16566 (cris_hard_regno_mode_ok): New function.
16567 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16568 (epiphany_mode_class): Delete.
16569 (HARD_REGNO_MODE_OK): Delete.
16570 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16571 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16572 (hard_regno_mode_ok): Rename to...
16573 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16574 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16575 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16576 HARD_REGNO_MODE_OK.
16577 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16578 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16579 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16580 (frv_hard_regno_mode_ok): Make static and return a bool.
16581 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16582 HARD_REGNO_MODE_OK.
16583 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16584 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16585 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16586 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16588 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16589 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16590 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16591 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16594 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16595 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16596 (ia64_hard_regno_mode_ok): New function.
16597 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16598 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16599 (iq2000_hard_regno_mode_ok): New function.
16600 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16601 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16602 (lm32_hard_regno_mode_ok): New function.
16603 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16604 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16605 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16606 instead of HARD_REGNO_MODE_OK.
16607 (m32c_hard_regno_ok): Rename to...
16608 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16609 (m32c_cannot_change_mode_class): Update accordingly.
16610 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16611 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16612 (m32r_mode_class): Delete.
16613 (HARD_REGNO_MODE_OK): Delete.
16614 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16615 (m32r_hard_regno_mode_ok): Rename to...
16616 (m32r_hard_regno_modes): ...this.
16617 (m32r_mode_class): Make static.
16618 (m32r_hard_regno_mode_ok): New function.
16619 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16620 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16621 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16622 (m68k_hard_regno_mode_ok): Make static.
16623 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16624 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16625 (mcore_hard_regno_mode_ok): New function.
16626 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16627 (HARD_REGNO_MODE_OK): Delete.
16628 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16630 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16631 (microblaze_hard_regno_mode_ok): New function.
16632 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16633 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16634 (mips_hard_regno_mode_ok): Delete.
16635 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16636 (mips_hard_regno_mode_ok_p): ...this and make static.
16637 (mips_hard_regno_mode_ok_p): Rename to...
16638 (mips_hard_regno_mode_ok_uncached): ...this.
16639 (mips_hard_regno_mode_ok): New function.
16640 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16641 of HARD_REGNO_MODE_OK.
16642 (mips_option_override): Update after above name changes.
16643 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16644 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16645 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16646 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16647 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16649 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16650 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16651 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16652 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16653 (msp430_hard_regno_mode_ok): Make static and return a bool.
16654 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16655 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16656 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16658 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16659 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16660 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16661 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16662 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16663 (PA_HARD_REGNO_MODE_OK): ...this
16664 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16665 (PA_HARD_REGNO_MODE_OK): ...this.
16666 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16667 (pa_hard_regno_mode_ok): New function.
16668 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16669 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16670 (pdp11_hard_regno_mode_ok): New function.
16671 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16672 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16674 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16677 (rs6000_hard_regno_mode_ok): Rename to...
16678 (rs6000_hard_regno_mode_ok_uncached): ...this.
16679 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16680 (rs6000_hard_regno_mode_ok): New function.
16681 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16682 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16683 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16684 (riscv_hard_regno_mode_ok): ...this and make static.
16685 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16686 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16687 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16688 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16689 (rl78_hard_regno_mode_ok): Make static and return bool.
16690 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16691 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16693 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16695 (rs6000_hard_regno_mode_ok): Rename to...
16696 (rs6000_hard_regno_mode_ok_uncached): ...this.
16697 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16698 (rs6000_hard_regno_mode_ok): New function.
16699 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16700 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16701 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16702 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16703 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16704 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16705 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16706 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16707 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16708 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16709 (sh_hard_regno_mode_ok): Make static.
16710 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16711 instead of HARD_REGNO_MODE_OK.
16712 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16713 (sparc_mode_class): Delete.
16714 (HARD_REGNO_MODE_OK): Delete.
16715 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16716 (hard_regno_mode_classes): Make static.
16717 (sparc_mode_class): Likewise.
16718 (sparc_hard_regno_mode_ok): New function.
16719 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16720 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16721 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16723 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16724 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16725 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16726 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16727 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16729 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16730 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16731 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16732 (visium_hard_regno_mode_ok): New function.
16733 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16734 instead of HARD_REGNO_MODE_OK.
16735 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16736 (HARD_REGNO_MODE_OK): Delete.
16737 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16738 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16739 (xtensa_option_override): Update accordingly.
16740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16741 (xtensa_hard_regno_mode_ok): New function.
16742 * system.h (HARD_REGNO_MODE_OK): Poison.
16744 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16745 Alan Hayward <alan.hayward@arm.com>
16746 David Sherwood <david.sherwood@arm.com>
16748 * target.def (hard_regno_call_part_clobbered): New hook.
16749 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16750 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16751 * doc/tm.texi: Regenerate.
16752 * hooks.h (hook_bool_uint_mode_false): Declare.
16753 * hooks.c (hook_bool_uint_mode_false): New function.
16754 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16755 * cselib.c (cselib_process_insn): Use
16756 targetm.hard_regno_call_part_clobbered instead of
16757 HARD_REGNO_CALL_PART_CLOBBERED.
16758 * ira-conflicts.c (ira_build_conflicts): Likewise.
16759 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16760 * lra-constraints.c (need_for_call_save_p): Likewise.
16761 * lra-lives.c: Include target.h.
16762 (check_pseudos_live_through_calls): Use
16763 targetm.hard_regno_call_part_clobbered instead of
16764 HARD_REGNO_CALL_PART_CLOBBERED.
16765 * regcprop.c: Include target.h.
16766 (copyprop_hardreg_forward_1): Use
16767 targetm.hard_regno_call_part_clobbered instead of
16768 HARD_REGNO_CALL_PART_CLOBBERED.
16769 * reginfo.c (choose_hard_reg_mode): Likewise.
16770 * regrename.c (check_new_reg_p): Likewise.
16771 * reload.c (find_equiv_reg): Likewise.
16772 * reload1.c (emit_reload_insns): Likewise.
16773 * sched-deps.c (deps_analyze_insn): Likewise.
16774 * sel-sched.c (init_regs_for_mode): Likewise.
16775 (mark_unavailable_hard_regs): Likewise.
16776 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16777 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16778 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16780 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16781 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16782 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16784 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16786 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16787 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16788 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16790 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16791 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16792 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16794 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16795 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16797 * config/powerpcspe/powerpcspe.c
16798 (rs6000_hard_regno_call_part_clobbered): New function.
16799 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16800 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16801 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16803 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16804 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16805 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16807 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16808 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16809 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16811 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16812 Alan Hayward <alan.hayward@arm.com>
16813 David Sherwood <david.sherwood@arm.com>
16815 * rtl.h (subreg_memory_offset): Declare.
16816 * emit-rtl.c (subreg_memory_offset): New function.
16817 * expmed.c (store_bit_field_1): Use it.
16818 * expr.c (undefined_operand_subword_p): Likewise.
16819 * simplify-rtx.c (simplify_subreg): Likewise.
16821 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16823 PR rtl-optimization/57448
16826 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16827 using atomic_load pattern.
16828 (expand_atomic_store): Likewise.
16830 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16833 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16834 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16835 handling. Use replace_call_with_value with NULL instead of
16836 gsi_replace, unlink_stmt_vdef and release_defs.
16838 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16841 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16843 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16846 * rtl.h (emit_library_call_value_1): Declare.
16847 (emit_library_call): Replace declaration with a series of overloads.
16848 Remove the parameter count argument.
16849 (emit_library_call_value): Likewise.
16850 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16851 with an "rtx_mode_t *".
16852 (emit_library_call_value): Delete.
16853 (emit_library_call): Likewise.
16854 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16855 (asan_emit_allocas_unpoison): Likewise.
16856 * builtins.c (expand_builtin_powi): Likewise.
16857 (expand_asan_emit_allocas_unpoison): Likewise.
16858 * cfgexpand.c (expand_main_function): Likewise.
16859 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16860 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16861 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16862 * config/arm/arm.c (arm_trampoline_init): Likewise.
16863 (arm_call_tls_get_addr): Likewise.
16864 (arm_expand_divmod_libfunc): Likewise.
16865 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16866 (smulsi3_highpart): Likewise.
16867 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16868 (c6x_expand_compare): Likewise.
16869 (c6x_expand_movmem): Likewise.
16870 * config/frv/frv.c (frv_trampoline_init): Likewise.
16871 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16872 (ix86_expand_divmod_libfunc): Likewise.
16873 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16874 (ia64_expand_compare): Likewise.
16875 (ia64_profile_hook): Likewise.
16876 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16877 (nonlocal_goto): Likewise.
16878 (restore_stack_nonlocal): Likewise.
16879 * config/m32r/m32r.c (block_move_call): Likewise.
16880 (m32r_trampoline_init): Likewise.
16881 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16882 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16883 (m68k_call_m68k_read_tp): Likewise.
16884 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16885 (microblaze_expand_divide): Likewise.
16886 * config/mips/mips.h (mips_args): Likewise.
16887 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16888 (MIPS_ICACHE_SYNC): Likewise.
16889 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16890 (nios2_trampoline_init): Likewise.
16891 * config/pa/pa.c (hppa_tls_call): Likewise.
16892 (pa_trampoline_init): Likewise.
16893 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16894 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16895 (expand_strn_compare): Likewise.
16896 (rs6000_generate_compare): Likewise.
16897 (rs6000_expand_float128_convert): Likewise.
16898 (output_profile_hook): Likewise.
16899 (rs6000_trampoline_init): Likewise.
16900 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16901 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16902 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16903 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16904 (rs6000_generate_compare): Likewise.
16905 (rs6000_expand_float128_convert): Likewise.
16906 (output_profile_hook): Likewise.
16907 (rs6000_trampoline_init): Likewise.
16908 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16909 * config/sh/sh.c (sh_trampoline_init): Likewise.
16910 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16911 (sparc_emit_float_lib_cmp): Likewise.
16912 (sparc32_initialize_trampoline): Likewise.
16913 (sparc64_initialize_trampoline): Likewise.
16914 (sparc_profile_hook): Likewise.
16915 * config/spu/spu.c (ea_load_store): Likewise.
16916 * config/spu/spu.md (floatunssidf2): Likewise.
16917 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16918 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16919 * config/visium/visium.c (expand_block_move_4): Likewise.
16920 (expand_block_move_2): Likewise.
16921 (expand_block_move_1): Likewise.
16922 (expand_block_set_4): Likewise.
16923 (expand_block_set_2): Likewise.
16924 (expand_block_set_1): Likewise.
16925 (visium_trampoline_init): Likewise.
16926 (visium_profile_hook): Likewise.
16927 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16928 (xtensa_setup_frame_addresses): Likewise.
16929 (xtensa_trampoline_init): Likewise.
16930 * except.c (sjlj_emit_function_enter): Likewise.
16931 (sjlj_emit_function_exit): Likewise.
16932 * explow.c (allocate_dynamic_stack_space): Likewise.
16933 (probe_stack_range): Likewise.
16934 * expr.c (convert_mode_scalar): Likewise.
16935 * optabs.c (expand_binop): Likewise.
16936 (expand_twoval_binop_libfunc): Likewise.
16937 (expand_unop): Likewise.
16938 (prepare_cmp_insn): Likewise.
16939 (prepare_float_lib_cmp): Likewise.
16940 (expand_float): Likewise.
16941 (expand_fix): Likewise.
16942 (expand_fixed_convert): Likewise.
16943 (maybe_emit_sync_lock_test_and_set): Likewise.
16944 (expand_atomic_compare_and_swap): Likewise.
16945 (expand_mem_thread_fence): Likewise.
16946 (expand_atomic_fetch_op): Likewise.
16948 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16950 * doc/generic.texi (OpenACC): Adjust URL.
16951 * doc/invoke.texi (C Dialect Options): Ditto.
16953 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16955 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16956 predicate for operand 1. Add (m,<S>) constraint.
16957 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16958 Prevent memory operand 1 with register operand 2.
16960 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16962 PR rtl-optimization/82024
16963 * combine.c (try_combine): If the combination result is a PARALLEL,
16964 and we only need to retain the SET in there that would be placed
16965 at I2, check that we can place that at I3 instead, before doing so.
16967 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16970 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16972 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16973 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16976 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16977 Jeff Law <law@redhat.com>
16979 * varasm.c (bss_initializer_p): Do not put constants into .bss
16980 (categorize_decl_for_section): Handle bss_initializer_p returning
16981 false when DECL_INITIAL is NULL.
16983 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16986 * config/s390/s390.c (s390_can_inline_p): New function.
16988 2017-09-01 Jeff Law <law@redhat.com>
16990 PR tree-optimization/82052
16991 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16992 Always initialize the returned slot after a hash table miss
16993 when INSERT is true.
16995 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16997 * config/s390/s390.md (mem_signal_fence): Remove.
16998 * doc/md.texi (mem_signal_fence): Remove.
16999 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17001 * target-insns.def (mem_signal_fence): Remove.
17003 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17006 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17009 * asan.c (create_odr_indicator): Strip name encoding from assembler
17010 name before appending it after __odr_asan_.
17012 2017-09-01 Martin Liska <mliska@suse.cz>
17014 PR tree-optimization/82059
17015 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17016 frequency only when an edge is redirected.
17018 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17020 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17021 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17022 (arc_conditional_register_usage): Remove ARC600 lp_count
17024 (arc_file_start): Emit Tag_ARC_CPU_variation.
17025 (arc_can_use_doloop_p): New conditions to use ZOLs.
17026 (hwloop_fail): New function.
17027 (hwloop_optimize): Likewise.
17028 (hwloop_pattern_reg): Likewise.
17029 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17030 (arc_reorg_loops): New function, calls reorg_loops.
17031 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17032 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17034 (arc_loop_hazard): Remove function, functionality moved into
17036 (arc_hazard): Remove arc_loop_hazard call.
17037 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17038 into hwloop_optimize.
17039 (arc_label_align): Remove ZOL handling.
17040 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17041 * config/arc/arc.md (doloop_begin): Remove pattern.
17042 (doloop_begin_i): Likewise.
17043 (doloop_end_i): Likewise.
17044 (doloop_fallback): Likewise.
17045 (doloop_fallback_m): Likewise.
17046 (doloop_end): Reimplement expand.
17047 (arc_lp): New pattern for LP instruction.
17048 (loop_end): New pattern.
17049 (loop_fail): Likewise.
17050 (decrement_and_branch_until_zero): Likewise.
17051 * config/arc/arc.opt (mlpc-width): New option.
17052 * doc/invoke.texi (mlpc-width): Document option.
17054 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17056 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17057 (arc_ccfsm_advance): Fix checking for delay slots.
17058 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17060 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17062 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17064 (movhi_insn): Update store instruction constraint which are saving
17065 6-bit short immediates.
17066 (movsi_insn): Consider also short scaled load operations.
17067 (zero_extendhisi2_i): Use Usd constraint instead of T.
17068 (extendhisi2_i): Add q constraint.
17069 (arc_clzsi2): Add type and length attributes.
17070 (arc_ctzsi2): Likewise.
17071 * config/arc/constraints.md (Usc): Update constraint, the
17072 assembler can parse two relocations for a single instruction.
17074 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17076 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17077 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17079 2017-08-31 Olivier Hainque <hainque@adacore.com>
17081 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17082 match as powerpc-wrs-vxworks*.
17084 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17086 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17087 register constraint for by-element operand.
17088 (aarch64_mls_elt_merge<mode>): Likewise.
17090 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17092 * config/arc/arc.c (arc_can_follow_jump): Check for short
17095 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17097 * config.gcc: Use g.opt for arc.
17098 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17099 functionality moved to ...
17100 (legitimate_scaled_address_p): New function, ...here.
17101 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17102 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17103 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17105 (arc_override_options): Handle G option.
17106 (arc_output_pic_addr_const): Correct function definition.
17107 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17108 (arc_decl_anon_ns_mem_p): Delete.
17109 (arc_in_small_data_p): Overhaul this function to take into
17110 consideration the value given via G option.
17111 (arc_rewrite_small_data_1): Renamed and corrected old
17112 arc_rewrite_small_data function.
17113 (arc_rewrite_small_data): New function.
17114 (small_data_pattern): Don't use pic_offset_table_rtx.
17115 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17116 * config/arc/simdext.md (movmisalignv2hi): Use
17117 prepare_move_operands function.
17119 (movmisalign*): Likewise.
17120 * doc/invoke.texi (ARC options): Document -G option.
17122 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17124 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17126 * config/arc/arc.c (arc_print_operand): Output scalled address for
17127 sdata whenever is possible.
17128 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17129 load/stores are available.
17130 (compact_sda_memory_operand): Check for the alignment required by
17131 code density instructions.
17132 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17134 * config/arc/constraints.md (Usd): Update constraint.
17135 (Us0): New constraint.
17136 (Usc): Update constraint.
17138 2017-08-31 Richard Biener <rguenther@suse.de>
17140 PR middle-end/82054
17141 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17142 function only once.
17144 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17146 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17147 Resize type_signature.
17149 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17150 Alan Hayward <alan.hayward@arm.com>
17151 David Sherwood <david.sherwood@arm.com>
17153 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17154 subregs whose inner modes can be stored in GPRs.
17155 (aarch64_classify_index): Likewise.
17157 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17158 Alan Hayward <alan.hayward@arm.com>
17159 David Sherwood <david.sherwood@arm.com>
17161 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17162 (V_INT_EQUIV): ...this.
17163 (v_cmp_result): Rename to...
17164 (v_int_equiv): ...this.
17165 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17166 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17167 (copysign<mode>3): Likewise.
17168 (aarch64_simd_bsl<mode>_internal): Likewise.
17169 (aarch64_simd_bsl<mode>): Likewise.
17170 (vec_cmp<mode><mode>): Likewise.
17171 (vcond<mode><mode>): Likewise.
17172 (vcond<v_cmp_mixed><mode>): Likewise.
17173 (vcondu<mode><v_cmp_mixed>): Likewise.
17174 (aarch64_cm<optab><mode>): Likewise.
17175 (aarch64_cmtst<mode>): Likewise.
17176 (aarch64_fac<optab><mode>): Likewise.
17177 (vec_perm_const<mode>): Likewise.
17178 (vcond_mask_<mode><v_cmp_result>): Rename to...
17179 (vcond_mask_<mode><v_int_equiv>): ...this.
17180 (vec_cmp<mode><v_cmp_result>): Rename to...
17181 (vec_cmp<mode><v_int_equiv>): ...this.
17183 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17184 Alan Hayward <alan.hayward@arm.com>
17185 David Sherwood <david.sherwood@arm.com>
17187 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17189 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17190 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17191 (UNSPEC_LD4_DREG): New unspecs.
17192 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17193 (aarch64_ld2<mode>_dreg_be): Replace with...
17194 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17196 (aarch64_ld3<mode>_dreg_le)
17197 (aarch64_ld3<mode>_dreg_be): Replace with...
17198 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17200 (aarch64_ld4<mode>_dreg_le)
17201 (aarch64_ld4<mode>_dreg_be): Replace with...
17202 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17205 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17207 PR tree-optimization/81987
17208 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17209 insert an initializer in a location not dominated by the stride
17212 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17214 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17215 on the entire header of the finally block in the fallthru case.
17217 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17219 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17221 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17223 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17224 rs6000_emit_move_from_cr and call renamed function.
17225 (rs6000_emit_prologue): Call renamed functions.
17226 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17227 movesi_from_cr, remove volatile CRs.
17229 2017-08-30 Jon Beniston <jon@beniston.com>
17230 Richard Biener <rguenther@suse.de>
17232 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17233 of VECTOR_MODE_P check.
17234 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17235 element vector types.
17237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17239 * df.h (df_read_modify_subreg_p): Remove in favor of...
17240 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17241 const_rtx instead of an rtx.
17242 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17243 * df-problems.c (df_word_lr_mark_ref): Likewise.
17244 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17245 (mark_pseudo_reg_dead): Likewise.
17246 (mark_ref_dead): Likewise.
17247 * reginfo.c (init_subregs_of_mode): Likewise.
17248 * sched-deps.c (sched_analyze_1): Likewise.
17249 * df-scan.c (df_def_record_1): Likewise.
17250 (df_uses_record): Likewise.
17251 (df_read_modify_subreg_p): Remove in favor of...
17252 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17253 const_rtx instead of an rtx.
17255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17256 Alan Hayward <alan.hayward@arm.com>
17257 David Sherwood <david.sherwood@arm.com>
17259 * rtl.h (partial_subreg_p): New function.
17260 * caller-save.c (save_call_clobbered_regs): Use it.
17261 * calls.c (expand_call): Likewise.
17262 * combine.c (combinable_i3pat): Likewise.
17263 (simplify_set): Likewise.
17264 (make_extraction): Likewise.
17265 (make_compound_operation_int): Likewise.
17266 (gen_lowpart_or_truncate): Likewise.
17267 (force_to_mode): Likewise.
17268 (make_field_assignment): Likewise.
17269 (reg_truncated_to_mode): Likewise.
17270 (record_truncated_value): Likewise.
17271 (move_deaths): Likewise.
17272 * cse.c (record_jump_cond): Likewise.
17273 (cse_insn): Likewise.
17274 * cselib.c (cselib_lookup_1): Likewise.
17275 * expmed.c (extract_bit_field_using_extv): Likewise.
17276 * function.c (assign_parm_setup_reg): Likewise.
17277 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17278 * ira-build.c (create_insn_allocnos): Likewise.
17279 * lra-coalesce.c (merge_pseudos): Likewise.
17280 * lra-constraints.c (match_reload): Likewise.
17281 (simplify_operand_subreg): Likewise.
17282 (curr_insn_transform): Likewise.
17283 * lra-lives.c (process_bb_lives): Likewise.
17284 * lra.c (new_insn_reg): Likewise.
17285 (lra_substitute_pseudo): Likewise.
17286 * regcprop.c (mode_change_ok): Likewise.
17287 (maybe_mode_change): Likewise.
17288 (copyprop_hardreg_forward_1): Likewise.
17289 * reload.c (push_reload): Likewise.
17290 (find_reloads): Likewise.
17291 (find_reloads_subreg_address): Likewise.
17292 * reload1.c (alter_reg): Likewise.
17293 (eliminate_regs_1): Likewise.
17294 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17296 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17298 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17299 back to if statements, including unpack.
17301 2017-08-30 Martin Liska <mliska@suse.cz>
17303 PR inline-asm/82001
17304 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17306 (func_checker::compare_asm_inputs_outputs): ... this function.
17307 (func_checker::compare_gimple_asm): Use the function to compare
17308 also ASM constrains.
17309 * ipa-icf-gimple.h: Rename the function.
17311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17312 Alan Hayward <alan.hayward@arm.com>
17313 David Sherwood <david.sherwood@arm.com>
17315 * coretypes.h (complex_mode): New type.
17316 * gdbhooks.py (build_pretty_printer): Handle it.
17317 * machmode.h (complex_mode): New class.
17318 (complex_mode::includes_p): New function.
17319 (is_complex_int_mode): Likewise.
17320 (is_complex_float_mode): Likewise.
17321 * genmodes.c (get_mode_class): Handle complex mode classes.
17322 * function.c (expand_function_end): Use is_complex_int_mode.
17324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17325 Alan Hayward <alan.hayward@arm.com>
17326 David Sherwood <david.sherwood@arm.com>
17328 * coretypes.h (scalar_mode_pod): New typedef.
17329 * gdbhooks.py (build_pretty_printer): Handle it.
17330 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17331 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17332 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17333 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17334 as_a <scalar_mode>.
17336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17337 Alan Hayward <alan.hayward@arm.com>
17338 David Sherwood <david.sherwood@arm.com>
17340 * machmode.h (mode_for_vector): Take a scalar_mode instead
17342 * stor-layout.c (mode_for_vector): Likewise.
17343 * explow.c (promote_mode): Use as_a <scalar_mode>.
17344 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17347 Alan Hayward <alan.hayward@arm.com>
17348 David Sherwood <david.sherwood@arm.com>
17350 * target.def (preferred_simd_mode): Take a scalar_mode
17351 instead of a machine_mode.
17352 * targhooks.h (default_preferred_simd_mode): Likewise.
17353 * targhooks.c (default_preferred_simd_mode): Likewise.
17354 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17355 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17356 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17357 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17358 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17359 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17360 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17361 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17363 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17364 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17365 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17366 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17367 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17368 * doc/tm.texi: Regenerate.
17369 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 * target.def (scalar_mode_supported_p): Take a scalar_mode
17377 instead of a machine_mode.
17378 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17379 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17380 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17381 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17382 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17383 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17384 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17385 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17386 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17387 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17388 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17389 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17390 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17391 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17393 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17394 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17395 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17396 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17397 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17399 * doc/tm.texi: Regenerate.
17401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17402 Alan Hayward <alan.hayward@arm.com>
17403 David Sherwood <david.sherwood@arm.com>
17405 * coretypes.h (opt_scalar_mode): New typedef.
17406 * gdbhooks.py (build_pretty_printers): Handle it.
17407 * machmode.h (mode_iterator::get_2xwider): Add overload for
17409 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17411 * expr.c (convert_mode_scalar): Likewise.
17412 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17413 * optabs.c (expand_float): Likewise.
17414 (expand_fix): Likewise.
17415 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17418 Alan Hayward <alan.hayward@arm.com>
17419 David Sherwood <david.sherwood@arm.com>
17421 * optabs.c (expand_float): Explicitly check for scalars before
17422 using a branching expansion.
17423 (expand_fix): Likewise.
17425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17426 Alan Hayward <alan.hayward@arm.com>
17427 David Sherwood <david.sherwood@arm.com>
17429 * expr.c (convert_mode): Split scalar handling out into...
17430 (convert_mode_scalar): ...this new function. Treat the modes
17433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17434 Alan Hayward <alan.hayward@arm.com>
17435 David Sherwood <david.sherwood@arm.com>
17437 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17439 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17442 Alan Hayward <alan.hayward@arm.com>
17443 David Sherwood <david.sherwood@arm.com>
17445 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17446 rather than a machine_mode.
17447 (fixed_from_string): Likewise.
17448 (fixed_convert): Likewise.
17449 (fixed_convert_from_int): Likewise.
17450 (fixed_convert_from_real): Likewise.
17451 (real_convert_from_fixed): Likewise.
17452 * fixed-value.c (fixed_from_double_int): Likewise.
17453 (fixed_from_string): Likewise.
17454 (fixed_convert): Likewise.
17455 (fixed_convert_from_int): Likewise.
17456 (fixed_convert_from_real): Likewise.
17457 (real_convert_from_fixed): Likewise.
17458 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17461 Alan Hayward <alan.hayward@arm.com>
17462 David Sherwood <david.sherwood@arm.com>
17464 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17465 of separate mode class checks. Do not allow vector modes here.
17466 (immed_wide_int_const): Use as_a <scalar_mode>.
17467 * explow.c (trunc_int_for_mode): Likewise.
17468 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17469 (wi::shwi): Likewise.
17470 (wi::min_value): Likewise.
17471 (wi::max_value): Likewise.
17472 * dwarf2out.c (loc_descriptor): Likewise.
17473 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17474 for CONST_WIDE_INT.
17476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17477 Alan Hayward <alan.hayward@arm.com>
17478 David Sherwood <david.sherwood@arm.com>
17480 * tree.h (SCALAR_TYPE_MODE): New macro.
17481 * expr.c (expand_expr_addr_expr_1): Use it.
17482 (expand_expr_real_2): Likewise.
17483 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17484 (fold_convert_const_fixed_from_int): Likewise.
17485 (fold_convert_const_fixed_from_real): Likewise.
17486 (native_encode_fixed): Likewise
17487 (native_encode_complex): Likewise
17488 (native_encode_vector): Likewise.
17489 (native_interpret_fixed): Likewise.
17490 (native_interpret_real): Likewise.
17491 (native_interpret_complex): Likewise.
17492 (native_interpret_vector): Likewise.
17493 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17494 (simd_clone_adjust_argument_types): Likewise.
17495 (simd_clone_init_simd_arrays): Likewise.
17496 (simd_clone_adjust): Likewise.
17497 * stor-layout.c (layout_type): Likewise.
17498 * tree.c (build_minus_one_cst): Likewise.
17499 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17500 * tree-inline.c (estimate_move_cost): Likewise.
17501 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17502 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17503 (vectorizable_reduction): Likewise.
17504 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17505 (vect_recog_mixed_size_cond_pattern): Likewise.
17506 (check_bool_pattern): Likewise.
17507 (adjust_bool_pattern): Likewise.
17508 (search_type_for_mask_1): Likewise.
17509 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17510 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17511 (vectorizable_load): Likewise.
17512 (vectorizable_store): Likewise.
17513 * ubsan.c (ubsan_encode_value): Likewise.
17514 * varasm.c (output_constant): Likewise.
17516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17517 Alan Hayward <alan.hayward@arm.com>
17518 David Sherwood <david.sherwood@arm.com>
17520 * coretypes.h (scalar_mode): New class.
17521 * machmode.h (scalar_mode): Likewise.
17522 (scalar_mode::includes_p): New function.
17523 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17524 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17525 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17526 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17527 * expmed.c (store_bit_field_1): Likewise.
17528 (extract_bit_field_1): Likewise.
17529 * expr.c (write_complex_part): Likewise.
17530 (read_complex_part): Likewise.
17531 (emit_move_complex_push): Likewise.
17532 (expand_expr_real_2): Likewise.
17533 * function.c (assign_parm_setup_reg): Likewise.
17534 (assign_parms_unsplit_complex): Likewise.
17535 * optabs.c (expand_binop): Likewise.
17536 * rtlanal.c (subreg_get_info): Likewise.
17537 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17538 * varasm.c (output_constant_pool_2): Likewise.
17540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17541 Alan Hayward <alan.hayward@arm.com>
17542 David Sherwood <david.sherwood@arm.com>
17544 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17546 (expmed_mult_highpart_optab): Likewise.
17547 (expmed_mult_highpart): Likewise.
17549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17550 Alan Hayward <alan.hayward@arm.com>
17551 David Sherwood <david.sherwood@arm.com>
17553 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17554 instead of a machine_mode.
17555 (builtin_memset_read_str): Likewise.
17556 * builtins.c (c_readstr): Likewise.
17557 (builtin_memcpy_read_str): Likewise.
17558 (builtin_strncpy_read_str): Likewise.
17559 (builtin_memset_read_str): Likewise.
17560 (builtin_memset_gen_str): Likewise.
17561 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17562 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17563 instead of a machine_mode.
17564 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17566 (make_extraction): Likewise.
17567 (try_widen_shift_mode): Take and return scalar_int_modes instead
17569 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17570 a scalar_int_mode instead of a machine_mode.
17571 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17572 (avr_addr_space_pointer_mode): Likewise.
17573 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17574 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17575 (msp430_unwind_word_mode): Likewise.
17576 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17577 (spu_addr_space_pointer_mode): Likewise.
17578 (spu_addr_space_address_mode): Likewise.
17579 (spu_libgcc_cmp_return_mode): Likewise.
17580 (spu_libgcc_shift_count_mode): Likewise.
17581 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17582 (rl78_addr_space_pointer_mode): Likewise.
17583 (fl78_unwind_word_mode): Likewise.
17584 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17586 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17587 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17588 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17589 (mips_valid_pointer_mode): Likewise.
17590 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17591 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17592 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17594 (ft32_addr_space_address_mode): Likewise.
17595 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17596 scalar_int_mode instead of a machine_mode.
17597 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17599 (m32c_addr_space_address_mode): Likewise.
17600 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17601 (rs6000_eh_return_filter_mode): Likewise.
17602 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17603 (rs6000_eh_return_filter_mode): Likewise.
17604 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17605 (s390_libgcc_shift_count_mode): Likewise.
17606 (s390_unwind_word_mode): Likewise.
17607 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17609 * target.def (mode_rep_extended): Likewise.
17610 (valid_pointer_mode): Likewise.
17611 (addr_space.valid_pointer_mode): Likewise.
17612 (eh_return_filter_mode): Return a scalar_int_mode rather than
17614 (libgcc_cmp_return_mode): Likewise.
17615 (libgcc_shift_count_mode): Likewise.
17616 (unwind_word_mode): Likewise.
17617 (addr_space.pointer_mode): Likewise.
17618 (addr_space.address_mode): Likewise.
17619 * doc/tm.texi: Regenerate.
17620 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17622 (do_jump): Use scalar_int_mode for local variables.
17623 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17624 rather than a machine_mode.
17625 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17626 (scompare_loc_descriptor_wide): Likewise.
17627 (scompare_loc_descriptor_narrow): Likewise.
17628 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17630 * except.c (sjlj_emit_dispatch_table): Likewise.
17631 (expand_builtin_eh_copy_values): Likewise.
17632 * explow.c (convert_memory_address_addr_space_1): Likewise.
17633 Take a scalar_int_mode rather than a machine_mode.
17634 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17635 than a machine_mode.
17636 (memory_address_addr_space): Use scalar_int_mode for local variables.
17637 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17638 rather than a machine_mode.
17639 * expmed.c (mask_rtx): Likewise.
17640 (init_expmed_one_conv): Likewise.
17641 (expand_mult_highpart_adjust): Likewise.
17642 (extract_high_half): Likewise.
17643 (expmed_mult_highpart_optab): Likewise.
17644 (expmed_mult_highpart): Likewise.
17645 (expand_smod_pow2): Likewise.
17646 (expand_sdiv_pow2): Likewise.
17647 (emit_store_flag_int): Likewise.
17648 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17650 (extract_low_bits): Likewise.
17651 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17653 * expr.c (pieces_addr::adjust): Likewise.
17654 (can_store_by_pieces): Likewise.
17655 (store_by_pieces): Likewise.
17656 (clear_by_pieces_1): Likewise.
17657 (expand_expr_addr_expr_1): Likewise.
17658 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17659 (expand_expr_real_1): Likewise.
17660 (try_casesi): Likewise.
17661 * final.c (shorten_branches): Likewise.
17662 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17663 type of "mode" to machine_mode.
17664 * internal-fn.c (expand_arith_overflow_result_store): Take a
17665 scalar_int_mode rather than a machine_mode.
17666 (expand_mul_overflow): Use scalar_int_mode for local variables.
17667 * loop-doloop.c (doloop_modify): Likewise.
17668 (doloop_optimize): Likewise.
17669 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17670 than a machine_mode.
17671 (expand_doubleword_shift_condmove): Likewise.
17672 (expand_doubleword_shift): Likewise.
17673 (expand_doubleword_clz): Likewise.
17674 (expand_doubleword_popcount): Likewise.
17675 (expand_doubleword_parity): Likewise.
17676 (expand_absneg_bit): Use scalar_int_mode for local variables.
17677 (prepare_float_lib_cmp): Likewise.
17678 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17679 rather than a machine_mode.
17680 (convert_memory_address_addr_space): Likewise.
17681 (get_mode_bounds): Likewise.
17682 (get_address_mode): Return a scalar_int_mode rather than a
17684 * rtlanal.c (get_address_mode): Likewise.
17685 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17686 than a machine_mode.
17687 * targhooks.c (default_mode_rep_extended): Likewise.
17688 (default_valid_pointer_mode): Likewise.
17689 (default_addr_space_valid_pointer_mode): Likewise.
17690 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17691 than a machine_mode.
17692 (default_libgcc_cmp_return_mode): Likewise.
17693 (default_libgcc_shift_count_mode): Likewise.
17694 (default_unwind_word_mode): Likewise.
17695 (default_addr_space_pointer_mode): Likewise.
17696 (default_addr_space_address_mode): Likewise.
17697 * targhooks.h (default_eh_return_filter_mode): Likewise.
17698 (default_libgcc_cmp_return_mode): Likewise.
17699 (default_libgcc_shift_count_mode): Likewise.
17700 (default_unwind_word_mode): Likewise.
17701 (default_addr_space_pointer_mode): Likewise.
17702 (default_addr_space_address_mode): Likewise.
17703 (default_mode_rep_extended): Take a scalar_int_mode rather than
17705 (default_valid_pointer_mode): Likewise.
17706 (default_addr_space_valid_pointer_mode): Likewise.
17707 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17709 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17710 rather than a machine_mode.
17711 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17712 for local variables.
17713 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17714 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17715 than a machine_mode.
17717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17718 Alan Hayward <alan.hayward@arm.com>
17719 David Sherwood <david.sherwood@arm.com>
17721 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17722 the mode argument to scalar_int_mode.
17723 (do_jump_by_parts_zero_rtx): Likewise.
17724 (do_jump_by_parts_equality_rtx): Likewise.
17725 (do_jump_by_parts_greater): Take a mode argument.
17726 (do_jump_by_parts_equality): Likewise.
17727 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
17734 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17735 (jump_table_for_label): Likewise.
17736 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17737 instead of an rtx_insn *.
17738 (shorten_branches): Use dyn_cast instead of LABEL_P and
17739 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17740 rtx_jump_table_data::get_data_mode.
17741 (final_scan_insn): Likewise.
17743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17744 Alan Hayward <alan.hayward@arm.com>
17745 David Sherwood <david.sherwood@arm.com>
17747 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17748 trying to combine a full-register integer set with a subreg
17751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17752 Alan Hayward <alan.hayward@arm.com>
17753 David Sherwood <david.sherwood@arm.com>
17755 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17756 that is always either address_mode or pointer_mode.
17758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17759 Alan Hayward <alan.hayward@arm.com>
17760 David Sherwood <david.sherwood@arm.com>
17762 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17763 when the two are known to be equal.
17765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17766 Alan Hayward <alan.hayward@arm.com>
17767 David Sherwood <david.sherwood@arm.com>
17769 * simplify-rtx.c (simplify_const_unary_operation): Use
17770 is_a <scalar_int_mode> instead of checking for a nonzero
17771 precision. Forcibly convert op_mode to a scalar_int_mode
17772 in that case. More clearly differentiate the operand and
17773 result modes and use the former when deciding what the value
17774 of a count-bits operation should be. Use is_int_mode instead
17775 of checking for a MODE_INT. Remove redundant check for whether
17776 this mode has a zero precision.
17778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17779 Alan Hayward <alan.hayward@arm.com>
17780 David Sherwood <david.sherwood@arm.com>
17782 * optabs.c (widen_leading): Change the type of the mode argument
17783 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17784 (widen_bswap): Likewise.
17785 (expand_parity): Likewise.
17786 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17787 (expand_ffs): Likewise.
17788 (epand_unop): Check for scalar integer modes before calling the
17791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17792 Alan Hayward <alan.hayward@arm.com>
17793 David Sherwood <david.sherwood@arm.com>
17795 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17797 (expand_expr_real_1): Update call accordingly.
17799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17800 Alan Hayward <alan.hayward@arm.com>
17801 David Sherwood <david.sherwood@arm.com>
17803 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17804 value_mode arguments. Use scalar_int_mode internally.
17805 (store_bit_field_1): Rename the new integer mode from imode
17806 to op0_mode and use it instead of GET_MODE (op0). Update calls
17807 to store_split_bit_field, store_bit_field_using_insv and
17808 store_fixed_bit_field.
17809 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17810 Use scalar_int_mode internally. Use a bit count rather than a mode
17811 when calculating the largest bit size for get_best_mode.
17812 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17813 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17814 Remove assertion that OP0 has a scalar integer mode.
17815 (store_split_bit_field): Add op0_mode and value_mode arguments.
17816 Update calls to extract_fixed_bit_field.
17817 (extract_bit_field_using_extv): Add an op0_mode argument.
17818 Use scalar_int_mode internally.
17819 (extract_bit_field_1): Rename the new integer mode from imode to
17820 op0_mode and use it instead of GET_MODE (op0). Update calls to
17821 extract_split_bit_field, extract_bit_field_using_extv and
17822 extract_fixed_bit_field.
17823 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17824 to extract_split_bit_field and extract_fixed_bit_field_1.
17825 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17826 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17827 on the target mode.
17828 (extract_split_bit_field): Add an op0_mode argument. Update call
17829 to extract_fixed_bit_field.
17831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17832 Alan Hayward <alan.hayward@arm.com>
17833 David Sherwood <david.sherwood@arm.com>
17835 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17836 * explow.c (hard_function_value): Likewise.
17837 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17838 convert_to_mode call outside the loop.
17839 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17840 for the mode iterator. Require the mode specified by max_pieces
17842 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17844 (copy_blkmode_to_reg): Likewise.
17845 (set_storage_via_setmem): Likewise.
17846 * optabs.c (prepare_cmp_insn): Likewise.
17847 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17848 * stor-layout.c (finish_bitfield_representative): Likewise.
17850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17851 Alan Hayward <alan.hayward@arm.com>
17852 David Sherwood <david.sherwood@arm.com>
17854 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17855 * expr.c (convert_move): Use them.
17856 (convert_modes): Likewise.
17857 (store_expr_with_bounds): Likewise.
17859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17860 Alan Hayward <alan.hayward@arm.com>
17861 David Sherwood <david.sherwood@arm.com>
17863 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17864 parameter for the mode of "x". Remove the "known_x", "known_mode"
17865 and "known_ret" arguments. Change the type of the mode argument
17866 to scalar_int_mode.
17867 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17868 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17869 (reg_num_sign_bit_copies_for_combine): Likewise.
17870 * rtlanal.c (nonzero_bits1): Likewise.
17871 (num_sign_bit_copies1): Likewise.
17872 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17873 (reg_num_sign_bit_copies_general): Likewise.
17874 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17875 (reg_nonzero_bits_general): Likewise.
17877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17878 Alan Hayward <alan.hayward@arm.com>
17879 David Sherwood <david.sherwood@arm.com>
17881 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17882 than in subroutines. Return 1 for non-integer modes.
17883 (cached_num_sign_bit_copies): Change the type of the mode parameter
17884 to scalar_int_mode.
17885 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17886 classes. Handle CONST_INT_P first and then check whether X also
17887 has a scalar integer mode. Check the same thing for inner registers
17888 of a SUBREG and for values that are being extended or truncated.
17890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17891 Alan Hayward <alan.hayward@arm.com>
17892 David Sherwood <david.sherwood@arm.com>
17894 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17895 in subroutines. Return the mode mask for non-integer modes.
17896 (cached_nonzero_bits): Change the type of the mode parameter
17897 to scalar_int_mode.
17898 (nonzero_bits1): Likewise. Remove early exit for other mode
17899 classes. Handle CONST_INT_P first and then check whether X
17900 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17907 Return a scalar_int_mode and assert that the size is greater than
17909 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17910 (op_by_pieces_d::op_by_pieces_d): Likewise.
17911 (op_by_pieces_d::run): Likewise.
17912 (can_store_by_pieces): Likewise.
17914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17915 Alan Hayward <alan.hayward@arm.com>
17916 David Sherwood <david.sherwood@arm.com>
17918 * combine.c (extract_left_shift): Add a mode argument and update
17920 (make_compound_operation_int): Change the type of the mode parameter
17921 to scalar_int_mode and update the call to extract_left_shift.
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 (simplify_and_const_int): Change the type of the mode
17928 parameter to scalar_int_mode.
17929 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
17936 scalar_int_mode (rather than VOIDmode) before testing its
17938 (simplify_comparison): Move COMPARISON_P handling out of the
17939 loop and restrict the latter part of the loop to scalar_int_modes.
17940 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17941 and when considering SUBREG_REGs. Use is_int_mode instead of
17942 checking GET_MODE_CLASS against MODE_INT.
17944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17945 Alan Hayward <alan.hayward@arm.com>
17946 David Sherwood <david.sherwood@arm.com>
17948 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17949 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17950 shift_unit_mode and for modes involved in scalar shifts.
17952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17953 Alan Hayward <alan.hayward@arm.com>
17954 David Sherwood <david.sherwood@arm.com>
17956 * combine.c (force_int_to_mode): New function, split out from...
17957 (force_to_mode): ...here. Keep xmode up-to-date and use it
17958 instead of GET_MODE (x).
17960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17961 Alan Hayward <alan.hayward@arm.com>
17962 David Sherwood <david.sherwood@arm.com>
17964 * optabs-query.h (extraction_insn::struct_mode): Change type to
17965 opt_scalar_int_mode and update comment.
17966 (extraction_insn::field_mode): Change type to scalar_int_mode.
17967 (extraction_insn::pos_mode): Likewise.
17968 * combine.c (make_extraction): Update accordingly.
17969 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17970 (get_optab_extraction_insn): Likewise.
17971 * recog.c (simplify_while_replacing): Likewise.
17972 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17973 parameter to opt_scalar_int_mode.
17975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17976 Alan Hayward <alan.hayward@arm.com>
17977 David Sherwood <david.sherwood@arm.com>
17979 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17980 to a scalar_int_mode instead of a machine_mode.
17981 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17982 (get_best_mode): Return a boolean and use a pointer argument to store
17983 the selected mode. Replace the limit mode parameter with a bit limit.
17984 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17985 for the values returned by bit_field_mode_iterator::next_mode.
17986 (store_bit_field): Update call to get_best_mode.
17987 (store_fixed_bit_field): Likewise.
17988 (extract_fixed_bit_field): Likewise.
17989 * expr.c (optimize_bitfield_assignment_op): Likewise.
17990 * fold-const.c (optimize_bit_field_compare): Likewise.
17991 (fold_truth_andor_1): Likewise.
17992 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17993 Update for new type of m_mode.
17994 (get_best_mode): As above.
17996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17997 Alan Hayward <alan.hayward@arm.com>
17998 David Sherwood <david.sherwood@arm.com>
18000 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18001 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18002 (store_bit_field): Check is_a <scalar_int_mode> before calling
18003 strict_volatile_bitfield_p.
18004 (extract_bit_field): Likewise.
18006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18007 Alan Hayward <alan.hayward@arm.com>
18008 David Sherwood <david.sherwood@arm.com>
18010 * target.def (cstore_mode): Return a scalar_int_mode.
18011 * doc/tm.texi: Regenerate.
18012 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18013 * targhooks.h (default_cstore_mode): Likewise.
18014 * targhooks.c (default_cstore_mode): Likewise, using a forced
18016 * expmed.c (emit_cstore): Expect the target of the cstore to be
18019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18020 Alan Hayward <alan.hayward@arm.com>
18021 David Sherwood <david.sherwood@arm.com>
18023 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18025 (niter_desc): Likewise mode.
18026 (iv_analyze): Add a mode parameter.
18028 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18029 and change its type to scalar_int_mode.
18030 * loop-iv.c: Update commentary at head of file.
18031 (iv_constant): Pass the mode paraeter before the rtx it describes
18032 and change its type to scalar_int_mode. Remove VOIDmode handling.
18033 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18034 (iv_extend): Likewise.
18035 (shorten_into_mode): Likewise.
18036 (iv_add): Use scalar_int_mode.
18037 (iv_mult): Likewise.
18038 (iv_shift): Likewise.
18039 (canonicalize_iv_subregs): Likewise.
18040 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18041 it describes and change its mode to scalar_int_mode. Also change
18042 the type of the returned inner_mode to scalar_int_mode.
18043 (get_biv_step): Likewise, turning outer_mode from a pointer
18044 into a direct parameter. Update call to get_biv_step_1.
18045 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18046 iv_constant and get_biv_step.
18047 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18048 and change its type to scalar_int_mode. Don't initialise iv->mode
18049 to VOIDmode and remove later checks for its still being VOIDmode.
18050 Update calls to iv_analyze_op and iv_analyze_expr. Check
18051 is_a <scalar_int_mode> when changing the mode under consideration.
18052 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18053 Update call to iv_analyze_expr.
18054 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18055 inner register is not also a scalar_int_mode. Update call to
18057 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18058 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18059 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18060 separate mode class checks. Update calls to iv_analyze. Remove
18061 fix-up of VOIDmodes after iv_analyze_biv.
18062 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18063 don't have a scalar_int_mode. Update call to biv_p.
18065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18066 Alan Hayward <alan.hayward@arm.com>
18067 David Sherwood <david.sherwood@arm.com>
18069 * cfgexpand.c (convert_debug_memory_address): Use
18070 as_a <scalar_int_mode>.
18071 * combine.c (expand_compound_operation): Likewise.
18072 (make_extraction): Likewise.
18073 (change_zero_ext): Likewise.
18074 (simplify_comparison): Likewise.
18075 * cse.c (cse_insn): Likewise.
18076 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18077 (mem_loc_descriptor): Likewise.
18078 (loc_descriptor): Likewise.
18079 * expmed.c (init_expmed_one_mode): Likewise.
18080 (synth_mult): Likewise.
18081 (emit_store_flag_1): Likewise.
18082 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18083 of a comparison with size.
18084 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18085 (reduce_to_bit_field_precision): Likewise.
18086 * function.c (expand_function_end): Likewise.
18087 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18088 * loop-doloop.c (doloop_modify): Likewise.
18089 * optabs.c (expand_binop): Likewise.
18090 (expand_unop): Likewise.
18091 (expand_copysign_absneg): Likewise.
18092 (prepare_cmp_insn): Likewise.
18093 (maybe_legitimize_operand): Likewise.
18094 * recog.c (const_scalar_int_operand): Likewise.
18095 * rtlanal.c (get_address_mode): Likewise.
18096 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18097 (simplify_cond_clz_ctz): Likewise.
18098 * tree-nested.c (get_nl_goto_field): Likewise.
18099 * tree.c (build_vector_type_for_mode): Likewise.
18100 * var-tracking.c (use_narrower_mode): Likewise.
18102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18103 Alan Hayward <alan.hayward@arm.com>
18104 David Sherwood <david.sherwood@arm.com>
18106 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18107 * builtins.c (expand_builtin_signbit): Use it.
18108 * cfgexpand.c (expand_debug_expr): Likewise.
18109 * dojump.c (do_jump): Likewise.
18110 (do_compare_and_jump): Likewise.
18111 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18112 * expmed.c (make_tree): Likewise.
18113 * expr.c (expand_expr_real_2): Likewise.
18114 (expand_expr_real_1): Likewise.
18115 (try_casesi): Likewise.
18116 * fold-const-call.c (fold_const_call_ss): Likewise.
18117 * fold-const.c (unextend): Likewise.
18118 (extract_muldiv_1): Likewise.
18119 (fold_single_bit_test): Likewise.
18120 (native_encode_int): Likewise.
18121 (native_encode_string): Likewise.
18122 (native_interpret_int): Likewise.
18123 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18124 * internal-fn.c (expand_addsub_overflow): Likewise.
18125 (expand_neg_overflow): Likewise.
18126 (expand_mul_overflow): Likewise.
18127 (expand_arith_overflow): Likewise.
18128 * match.pd: Likewise.
18129 * stor-layout.c (layout_type): Likewise.
18130 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18131 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18132 * tree-ssanames.c (get_range_info): Likewise.
18133 * tree-switch-conversion.c (array_value_type) Likewise.
18134 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18135 (vect_recog_divmod_pattern): Likewise.
18136 (vect_recog_mixed_size_cond_pattern): Likewise.
18137 * tree-vrp.c (extract_range_basic): Likewise.
18138 (simplify_float_conversion_using_ranges): Likewise.
18139 * tree.c (int_fits_type_p): Likewise.
18140 * ubsan.c (instrument_bool_enum_load): Likewise.
18141 * varasm.c (mergeable_string_section): Likewise.
18142 (narrowing_initializer_constant_valid_p): Likewise.
18143 (output_constant): Likewise.
18145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18146 Alan Hayward <alan.hayward@arm.com>
18147 David Sherwood <david.sherwood@arm.com>
18149 * machmode.h (NARROWEST_INT_MODE): New macro.
18150 * expr.c (alignment_for_piecewise_move): Use it instead of
18151 GET_CLASS_NARROWEST_MODE (MODE_INT).
18152 (push_block): Likewise.
18153 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18155 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18158 Alan Hayward <alan.hayward@arm.com>
18159 David Sherwood <david.sherwood@arm.com>
18161 * postreload.c (move2add_valid_value_p): Change the type of the
18162 mode parameter to scalar_int_mode.
18163 (move2add_use_add2_insn): Add a mode parameter and use it instead
18165 (move2add_use_add3_insn): Likewise.
18166 (reload_cse_move2add): Update accordingly.
18168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18169 Alan Hayward <alan.hayward@arm.com>
18170 David Sherwood <david.sherwood@arm.com>
18172 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18174 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18175 * optabs.c (expand_unop): 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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18182 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18183 (popcount_loc_descriptor): Likewise.
18184 (bswap_loc_descriptor): Likewise.
18185 (rotate_loc_descriptor): Likewise.
18186 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18187 calling the functions above.
18189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18190 Alan Hayward <alan.hayward@arm.com>
18191 David Sherwood <david.sherwood@arm.com>
18193 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18195 (try_combine): Likewise.
18196 (simplify_if_then_else): Likewise.
18197 * cse.c (cse_insn): Likewise.
18198 * dwarf2out.c (mem_loc_descriptor): Likewise.
18199 * emit-rtl.c (gen_lowpart_common): Likewise.
18200 * simplify-rtx.c (simplify_truncation): Likewise.
18201 (simplify_binary_operation_1): Likewise.
18202 (simplify_const_relational_operation): Likewise.
18203 (simplify_ternary_operation): Likewise.
18204 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18207 Alan Hayward <alan.hayward@arm.com>
18208 David Sherwood <david.sherwood@arm.com>
18210 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18211 * reload.c (push_reload): Likewise.
18212 (find_reloads): Likewise.
18214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18215 Alan Hayward <alan.hayward@arm.com>
18216 David Sherwood <david.sherwood@arm.com>
18218 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18219 (make_compound_operation_int): Likewise.
18220 (change_zero_ext): Likewise.
18221 * expr.c (convert_move): Likewise.
18222 (convert_modes): Likewise.
18223 * fwprop.c (forward_propagate_subreg): Likewise.
18224 * loop-iv.c (get_biv_step_1): Likewise.
18225 * optabs.c (widen_operand): Likewise.
18226 * postreload.c (move2add_valid_value_p): Likewise.
18227 * recog.c (simplify_while_replacing): Likewise.
18228 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18229 (simplify_binary_operation_1): Likewise. Remove redundant
18230 mode equality check.
18232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18233 Alan Hayward <alan.hayward@arm.com>
18234 David Sherwood <david.sherwood@arm.com>
18236 * combine.c (combine_simplify_rtx): Add checks for
18237 is_a <scalar_int_mode>.
18238 (simplify_if_then_else): Likewise.
18239 (make_field_assignment): Likewise.
18240 (simplify_comparison): Likewise.
18241 * ifcvt.c (noce_try_bitop): Likewise.
18242 * loop-invariant.c (canonicalize_address_mult): Likewise.
18243 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18246 Alan Hayward <alan.hayward@arm.com>
18247 David Sherwood <david.sherwood@arm.com>
18249 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18250 is_a <scalar_int_mode> instead of != BLKmode.
18252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18253 Alan Hayward <alan.hayward@arm.com>
18254 David Sherwood <david.sherwood@arm.com>
18256 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18257 instead of != VOIDmode.
18258 * combine.c (if_then_else_cond): Likewise.
18259 (change_zero_ext): Likewise.
18260 * dwarf2out.c (mem_loc_descriptor): Likewise.
18261 (loc_descriptor): Likewise.
18262 * rtlanal.c (canonicalize_condition): Likewise.
18263 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18266 Alan Hayward <alan.hayward@arm.com>
18267 David Sherwood <david.sherwood@arm.com>
18269 * simplify-rtx.c (simplify_binary_operation_1): Use
18270 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18273 Alan Hayward <alan.hayward@arm.com>
18274 David Sherwood <david.sherwood@arm.com>
18276 * wide-int.h (int_traits<unsigned char>) New class.
18277 (int_traits<unsigned short>) Likewise.
18278 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18279 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18281 * combine.c (set_nonzero_bits_and_sign_copies): Use
18282 is_a <scalar_int_mode>.
18283 (find_split_point): Likewise.
18284 (combine_simplify_rtx): Likewise.
18285 (simplify_logical): Likewise.
18286 (expand_compound_operation): Likewise.
18287 (expand_field_assignment): Likewise.
18288 (make_compound_operation): Likewise.
18289 (extended_count): Likewise.
18290 (change_zero_ext): Likewise.
18291 (simplify_comparison): Likewise.
18292 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18293 (ucompare_loc_descriptor): Likewise.
18294 (minmax_loc_descriptor): Likewise.
18295 (mem_loc_descriptor): Likewise.
18296 (loc_descriptor): Likewise.
18297 * expmed.c (init_expmed_one_mode): Likewise.
18298 * lra-constraints.c (lra_constraint_offset): Likewise.
18299 * optabs.c (prepare_libcall_arg): Likewise.
18300 * postreload.c (move2add_note_store): Likewise.
18301 * reload.c (operands_match_p): Likewise.
18302 * rtl.h (load_extend_op): Likewise.
18303 * rtlhooks.c (gen_lowpart_general): Likewise.
18304 * simplify-rtx.c (simplify_truncation): Likewise.
18305 (simplify_unary_operation_1): Likewise.
18306 (simplify_binary_operation_1): Likewise.
18307 (simplify_const_binary_operation): Likewise.
18308 (simplify_const_relational_operation): Likewise.
18309 (simplify_subreg): Likewise.
18310 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18311 * var-tracking.c (adjust_mems): Likewise.
18312 (prepare_call_arguments): Likewise.
18314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18315 Alan Hayward <alan.hayward@arm.com>
18316 David Sherwood <david.sherwood@arm.com>
18318 * machmode.h (is_int_mode): New fuction.
18319 * combine.c (find_split_point): Use it.
18320 (combine_simplify_rtx): Likewise.
18321 (simplify_if_then_else): Likewise.
18322 (simplify_set): Likewise.
18323 (simplify_shift_const_1): Likewise.
18324 (simplify_comparison): Likewise.
18325 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18326 * cse.c (notreg_cost): Likewise.
18327 (cse_insn): Likewise.
18328 * cselib.c (cselib_lookup_1): Likewise.
18329 * dojump.c (do_jump_1): Likewise.
18330 (do_compare_rtx_and_jump): Likewise.
18331 * dse.c (get_call_args): Likewise.
18332 * dwarf2out.c (rtl_for_decl_init): Likewise.
18333 (native_encode_initializer): Likewise.
18334 * expmed.c (emit_store_flag_1): Likewise.
18335 (emit_store_flag): Likewise.
18336 * expr.c (convert_modes): Likewise.
18337 (store_field): Likewise.
18338 (expand_expr_real_1): Likewise.
18339 * fold-const.c (fold_read_from_constant_string): Likewise.
18340 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18341 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18342 * optabs.c (expand_binop): Likewise.
18343 (expand_unop): Likewise.
18344 (expand_abs_nojump): Likewise.
18345 (expand_one_cmpl_abs_nojump): Likewise.
18346 * simplify-rtx.c (mode_signbit_p): Likewise.
18347 (val_signbit_p): Likewise.
18348 (val_signbit_known_set_p): Likewise.
18349 (val_signbit_known_clear_p): Likewise.
18350 (simplify_relational_operation_1): Likewise.
18351 * tree.c (vector_type_mode): Likewise.
18353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18354 Alan Hayward <alan.hayward@arm.com>
18355 David Sherwood <david.sherwood@arm.com>
18357 * machmode.h (smallest_mode_for_size): Fix formatting.
18358 (smallest_int_mode_for_size): New function.
18359 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18360 instead of smallest_mode_for_size.
18361 * combine.c (make_extraction): Likewise.
18362 * config/arc/arc.c (arc_expand_movmem): Likewise.
18363 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18364 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18365 * config/s390/s390.c (s390_expand_insv): Likewise.
18366 * config/sparc/sparc.c (assign_int_registers): Likewise.
18367 * config/spu/spu.c (spu_function_value): Likewise.
18368 (spu_function_arg): Likewise.
18369 * coverage.c (get_gcov_type): Likewise.
18370 (get_gcov_unsigned_t): Likewise.
18371 * dse.c (find_shift_sequence): Likewise.
18372 * expmed.c (store_bit_field_1): Likewise.
18373 * expr.c (convert_move): Likewise.
18374 (store_field): Likewise.
18375 * internal-fn.c (expand_arith_overflow): Likewise.
18376 * optabs-query.c (get_best_extraction_insn): Likewise.
18377 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18378 * stor-layout.c (layout_type): Likewise.
18379 (initialize_sizetypes): Likewise.
18380 * targhooks.c (default_get_mask_mode): Likewise.
18381 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18384 Alan Hayward <alan.hayward@arm.com>
18385 David Sherwood <david.sherwood@arm.com>
18387 * machmode.h (opt_mode::else_blk): New function.
18388 (int_mode_for_mode): Declare.
18389 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18390 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18392 * cfgexpand.c (expand_debug_expr): Likewise.
18393 * combine.c (gen_lowpart_or_truncate): Likewise.
18394 (gen_lowpart_for_combine): Likewise.
18395 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18396 * config/avr/avr.c (avr_to_int_mode): Likewise.
18397 (avr_out_plus_1): Likewise.
18398 (avr_out_plus): Likewise.
18399 (avr_out_round): Likewise.
18400 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18401 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18402 (s390_expand_vcond): Likewise.
18403 * config/spu/spu.c (spu_split_immediate): Likewise.
18404 (spu_expand_mov): Likewise.
18405 * dse.c (get_stored_val): Likewise.
18406 * expmed.c (store_bit_field_1): Likewise.
18407 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18409 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18410 (extract_low_bits): Likewise.
18411 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18412 handling rather than repeating the check.
18413 (emit_group_store): Likewise.
18414 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18415 * optabs.c (expand_absneg_bit): Likewise.
18416 (expand_copysign_absneg): Likewise.
18417 (expand_copysign_bit): Likewise.
18418 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18419 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18420 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18421 * var-tracking.c (prepare_call_arguments): Likewise.
18422 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18423 int_mode_for_mode instead of mode_for_size.
18424 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18427 Alan Hayward <alan.hayward@arm.com>
18428 David Sherwood <david.sherwood@arm.com>
18430 * machmode.h (int_mode_for_size): New function.
18431 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18432 instead of mode_for_size.
18433 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18435 * combine.c (expand_field_assignment): Use int_mode_for_size
18436 instead of mode_for_size.
18437 (make_extraction): Likewise.
18438 (simplify_shift_const_1): Likewise.
18439 (simplify_comparison): Likewise.
18440 * dojump.c (do_jump): Likewise.
18441 * dwarf2out.c (mem_loc_descriptor): Likewise.
18442 * emit-rtl.c (init_derived_machine_modes): Likewise.
18443 * expmed.c (flip_storage_order): Likewise.
18444 (convert_extracted_bit_field): Likewise.
18445 * expr.c (copy_blkmode_from_reg): Likewise.
18446 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18447 * internal-fn.c (expand_mul_overflow): Likewise.
18448 * lower-subreg.c (simple_move): Likewise.
18449 * optabs-libfuncs.c (init_optabs): Likewise.
18450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18451 * tree.c (vector_type_mode): Likewise.
18452 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18453 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18454 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18455 * tree-vect-stmts.c (vectorizable_load): Likewise.
18456 (vectorizable_store): Likewise.
18458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18459 Alan Hayward <alan.hayward@arm.com>
18460 David Sherwood <david.sherwood@arm.com>
18462 * coretypes.h (pod_mode): New type.
18463 (scalar_int_mode_pod): New typedef.
18464 * machmode.h (pod_mode): New class.
18465 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18466 * genmodes.c (emit_mode_int_n): Update accordingly.
18467 * lower-subreg.h (target_lower_subreg): Change type to
18468 scalar_int_mode_pod.
18469 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18470 scalar_int_mode_pod.
18472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18473 Alan Hayward <alan.hayward@arm.com>
18474 David Sherwood <david.sherwood@arm.com>
18476 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18477 machine_mode to scalar_int_mode.
18478 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18479 (rs6000_option_override_internal): Remove cast to int.
18480 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18481 machine_mode to scalar_int_mode.
18482 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18483 (rs6000_option_override_internal): Remove cast to int.
18484 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18485 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18487 * config/s390/s390.c (s390_expand_builtin): Likewise.
18488 * coretypes.h (scalar_int_mode): New type.
18489 (opt_scalar_int_mode): New typedef.
18490 * machmode.h (scalar_int_mode): New class.
18491 (scalar_int_mode::includes_p): New function.
18492 (byte_mode): Change type to scalar_int_mode.
18493 (word_mode): Likewise.
18494 (ptr_mode): Likewise.
18495 * emit-rtl.c (byte_mode): Likewise.
18496 (word_mode): Likewise.
18497 (ptr_mode): Likewise.
18498 (init_derived_machine_modes): Update accordingly.
18499 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18500 and MODE_PARTIAL_INT.
18501 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18502 opt_scalar_int_mode.
18504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18505 Alan Hayward <alan.hayward@arm.com>
18506 David Sherwood <david.sherwood@arm.com>
18508 * target.def (libgcc_floating_mode_supported_p): Take a
18510 * doc/tm.texi: Regenerate.
18511 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18513 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18514 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18518 Alan Hayward <alan.hayward@arm.com>
18519 David Sherwood <david.sherwood@arm.com>
18521 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18522 * doc/tm.texi: Regenerate.
18523 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18524 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18525 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18526 * targhooks.h (default_floatn_mode): Likewise.
18527 * targhooks.c (default_floatn_mode): Likewise.
18528 * tree.c (build_common_tree_nodes): Update accordingly.
18530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18531 Alan Hayward <alan.hayward@arm.com>
18532 David Sherwood <david.sherwood@arm.com>
18534 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18535 (mode_iterator::iterate_p): Likewise.
18536 (mode_iterator::get_wider): Likewise.
18537 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18540 Alan Hayward <alan.hayward@arm.com>
18541 David Sherwood <david.sherwood@arm.com>
18543 * coretypes.h (opt_scalar_float_mode): New typedef.
18544 * machmode.h (float_mode_for_size): New function.
18545 * emit-rtl.c (double_mode): Delete.
18546 (init_emit_once): Use float_mode_for_size.
18547 * stor-layout.c (layout_type): Likewise.
18548 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18551 Alan Hayward <alan.hayward@arm.com>
18552 David Sherwood <david.sherwood@arm.com>
18554 * output.h (assemble_real): Take a scalar_float_mode.
18555 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18556 * config/arm/arm.md (consttable_4): Likewise.
18557 (consttable_8): Likewise.
18558 (consttable_16): Likewise.
18559 * config/mips/mips.md (consttable_float): Likewise.
18560 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18561 * varasm.c (assemble_real): Take a scalar_float_mode.
18562 (output_constant_pool_2): Update accordingly.
18563 (output_constant): Likewise.
18565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18566 Alan Hayward <alan.hayward@arm.com>
18567 David Sherwood <david.sherwood@arm.com>
18569 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18570 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18571 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18572 (native_encode_real): Likewise.
18573 (native_interpret_real): Likewise.
18574 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18575 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18578 Alan Hayward <alan.hayward@arm.com>
18579 David Sherwood <david.sherwood@arm.com>
18581 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18582 <scalar_float_mode>. Simplify.
18583 (gen_extend_conv_libfunc): Likewise.
18585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18586 Alan Hayward <alan.hayward@arm.com>
18587 David Sherwood <david.sherwood@arm.com>
18589 * coretypes.h (scalar_float_mode): New type.
18590 * machmode.h (mode_traits::from_int): Use machine_mode if
18591 USE_ENUM_MODES is defined.
18592 (is_a): New function.
18594 (dyn_cast): Likewise.
18595 (scalar_float_mode): New class.
18596 (scalar_float_mode::includes_p): New function.
18597 (is_float_mode): Likewise.
18598 * gdbhooks.py (MachineModePrinter): New class.
18599 (build_pretty_printer): Use it for scalar_float_mode.
18600 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18601 (format_helper::format_helper): Turn into a template.
18602 * genmodes.c (get_mode_class): New function.
18603 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18604 or machine_mode if none.
18605 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18606 as_a <scalar_float_mode>.
18607 * dwarf2out.c (mem_loc_descriptor): Likewise.
18608 (insert_float): Likewise.
18609 (add_const_value_attribute): Likewise.
18610 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18611 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18612 (expand_unop): Update accordingly.
18613 (expand_abs_nojump): Likewise.
18614 (expand_copysign_absneg): Take a scalar_float_mode.
18615 (expand_copysign_bit): Likewise.
18616 (expand_copysign): Update accordingly.
18618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18619 Alan Hayward <alan.hayward@arm.com>
18620 David Sherwood <david.sherwood@arm.com>
18622 * coretypes.h (opt_mode): New class.
18623 * machmode.h (opt_mode): Likewise.
18624 (opt_mode::else_void): New function.
18625 (opt_mode::require): Likewise.
18626 (opt_mode::exists): Likewise.
18627 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18628 (GET_MODE_2XWIDER_MODE): Likewise.
18629 (mode_iterator::get_wider): Update accordingly.
18630 (mode_iterator::get_2xwider): Likewise.
18631 (mode_iterator::get_known_wider): Likewise, turning into a template.
18632 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18633 forcing a wider mode to exist.
18634 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18635 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18636 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18637 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18638 * lower-subreg.c (init_lower_subreg): Likewise.
18639 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18640 on the final iteration.
18641 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18642 a wider mode exists before asking for a move pattern.
18643 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18644 forcing a wider mode to exist.
18645 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18646 returning false if no such mode exists.
18647 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18648 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18649 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18650 Avoid checking for a MODE_INT if we already know the mode is not a
18652 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18653 forcing a wider mode to exist.
18654 (expmed_mult_highpart_optab): Likewise.
18655 (expmed_mult_highpart): Likewise.
18656 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18658 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18659 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18660 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18661 * internal-fn.c (expand_mul_overflow): Update use of
18662 GET_MODE_2XWIDER_MODE.
18663 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18664 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18665 GET_MODE_WIDER_MODE.
18666 (convert_plusminus_to_widen): Likewise.
18667 * tree-switch-conversion.c (array_value_type): Likewise.
18668 * var-tracking.c (emit_note_insn_var_location): Likewise.
18669 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18670 Return false inside rather than outside the loop if no wider mode
18672 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18673 and GET_MODE_2XWIDER_MODE
18674 (can_compare_p): Use else_void.
18675 * gdbhooks.py (OptMachineModePrinter): New class.
18676 (build_pretty_printer): Use it for opt_mode.
18678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18679 Alan Hayward <alan.hayward@arm.com>
18680 David Sherwood <david.sherwood@arm.com>
18682 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18683 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18686 Alan Hayward <alan.hayward@arm.com>
18687 David Sherwood <david.sherwood@arm.com>
18689 * machmode.h (mode_traits): New structure.
18690 (get_narrowest_mode): New function.
18691 (mode_iterator::start): Likewise.
18692 (mode_iterator::iterate_p): Likewise.
18693 (mode_iterator::get_wider): Likewise.
18694 (mode_iterator::get_known_wider): Likewise.
18695 (mode_iterator::get_2xwider): Likewise.
18696 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18697 (FOR_EACH_MODE): Likewise.
18698 (FOR_EACH_MODE_FROM): Likewise.
18699 (FOR_EACH_MODE_UNTIL): Likewise.
18700 (FOR_EACH_WIDER_MODE): Likewise.
18701 (FOR_EACH_2XWIDER_MODE): Likewise.
18702 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18703 * combine.c (simplify_comparison): Likewise
18704 * config/i386/i386.c (type_natural_mode): Likewise.
18705 * cse.c (cse_insn): Likewise.
18706 * dse.c (find_shift_sequence): Likewise.
18707 * emit-rtl.c (init_derived_machine_modes): Likewise.
18708 (init_emit_once): Likewise.
18709 * explow.c (hard_function_value): Likewise.
18710 * expmed.c (extract_fixed_bit_field_1): Likewise.
18711 (extract_bit_field_1): Likewise.
18712 (expand_divmod): Likewise.
18713 (emit_store_flag_1): Likewise.
18714 * expr.c (init_expr_target): Likewise.
18715 (convert_move): Likewise.
18716 (alignment_for_piecewise_move): Likewise.
18717 (widest_int_mode_for_size): Likewise.
18718 (emit_block_move_via_movmem): Likewise.
18719 (copy_blkmode_to_reg): Likewise.
18720 (set_storage_via_setmem): Likewise.
18721 (compress_float_constant): Likewise.
18722 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18723 * optabs-query.c (get_best_extraction_insn): Likewise.
18724 * optabs.c (expand_binop): Likewise.
18725 (expand_twoval_unop): Likewise.
18726 (expand_twoval_binop): Likewise.
18727 (widen_leading): Likewise.
18728 (widen_bswap): Likewise.
18729 (expand_parity): Likewise.
18730 (expand_unop): Likewise.
18731 (prepare_cmp_insn): Likewise.
18732 (prepare_float_lib_cmp): Likewise.
18733 (expand_float): Likewise.
18734 (expand_fix): Likewise.
18735 (expand_sfix_optab): Likewise.
18736 * postreload.c (move2add_use_add2_insn): Likewise.
18737 * reg-stack.c (reg_to_stack): Likewise.
18738 * reginfo.c (choose_hard_reg_mode): Likewise.
18739 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18740 * stor-layout.c (mode_for_size): Likewise.
18741 (smallest_mode_for_size): Likewise.
18742 (mode_for_vector): Likewise.
18743 (finish_bitfield_representative): Likewise.
18744 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18745 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18746 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18747 * var-tracking.c (prepare_call_arguments): Likewise.
18749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18750 Alan Hayward <alan.hayward@arm.com>
18751 David Sherwood <david.sherwood@arm.com>
18753 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18754 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18755 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18756 * machmode.h (mode_size): Move earlier in file.
18757 (mode_precision): Likewise.
18758 (mode_inner): Likewise.
18759 (mode_nunits): Likewise.
18760 (mode_unit_size): Likewise.
18761 (unit_unit_precision): Likewise.
18762 (mode_wider): Likewise.
18763 (mode_2xwider): Likewise.
18764 (machine_mode): New class.
18765 (mode_to_bytes): New function.
18766 (mode_to_bits): Likewise.
18767 (mode_to_precision): Likewise.
18768 (mode_to_inner): Likewise.
18769 (mode_to_unit_size): Likewise.
18770 (mode_to_unit_precision): Likewise.
18771 (mode_to_nunits): Likewise.
18772 (GET_MODE_SIZE): Use mode_to_bytes.
18773 (GET_MODE_BITSIZE): Use mode_to_bits.
18774 (GET_MODE_PRECISION): Use mode_to_precision.
18775 (GET_MODE_INNER): Use mode_to_inner.
18776 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18777 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18778 (GET_MODE_NUNITS): Use mode_to_nunits.
18779 * system.h (ALWAYS_INLINE): New macro.
18780 * config/powerpcspe/powerpcspe-c.c
18781 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18782 int for arg1_mode and arg2_mode.
18784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18785 Alan Hayward <alan.hayward@arm.com>
18786 David Sherwood <david.sherwood@arm.com>
18788 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18789 Prefix mode names with E_ in case statements.
18790 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18791 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18792 (aarch64_split_simd_move): Likewise.
18793 (aarch64_gen_storewb_pair): Likewise.
18794 (aarch64_gen_loadwb_pair): Likewise.
18795 (aarch64_gen_store_pair): Likewise.
18796 (aarch64_gen_load_pair): Likewise.
18797 (aarch64_get_condition_code_1): Likewise.
18798 (aarch64_constant_pool_reload_icode): Likewise.
18799 (get_rsqrte_type): Likewise.
18800 (get_rsqrts_type): Likewise.
18801 (get_recpe_type): Likewise.
18802 (get_recps_type): Likewise.
18803 (aarch64_gimplify_va_arg_expr): Likewise.
18804 (aarch64_simd_container_mode): Likewise.
18805 (aarch64_emit_load_exclusive): Likewise.
18806 (aarch64_emit_store_exclusive): Likewise.
18807 (aarch64_expand_compare_and_swap): Likewise.
18808 (aarch64_gen_atomic_cas): Likewise.
18809 (aarch64_emit_bic): Likewise.
18810 (aarch64_emit_atomic_swap): Likewise.
18811 (aarch64_emit_atomic_load_op): Likewise.
18812 (aarch64_evpc_trn): Likewise.
18813 (aarch64_evpc_uzp): Likewise.
18814 (aarch64_evpc_zip): Likewise.
18815 (aarch64_evpc_ext): Likewise.
18816 (aarch64_evpc_rev): Likewise.
18817 (aarch64_evpc_dup): Likewise.
18818 (aarch64_gen_ccmp_first): Likewise.
18819 (aarch64_gen_ccmp_next): Likewise.
18820 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18821 (alpha_emit_xfloating_libcall): Likewise.
18822 (emit_insxl): Likewise.
18823 (alpha_arg_type): Likewise.
18824 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18825 (arc_preferred_simd_mode): Likewise.
18826 (arc_secondary_reload): Likewise.
18827 (get_arc_condition_code): Likewise.
18828 (arc_print_operand): Likewise.
18829 (arc_legitimate_constant_p): Likewise.
18830 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18831 * config/arc/arc.md (casesi_load): Likewise.
18832 (casesi_compact_jump): Likewise.
18833 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18834 (cc_use_register): Likewise.
18835 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18836 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18837 (arm_init_iwmmxt_builtins): Likewise.
18838 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18839 (neon_expand_vector_init): Likewise.
18840 (arm_attr_length_move_neon): Likewise.
18841 (maybe_get_arm_condition_code): Likewise.
18842 (arm_emit_vector_const): Likewise.
18843 (arm_preferred_simd_mode): Likewise.
18844 (arm_output_iwmmxt_tinsr): Likewise.
18845 (thumb1_output_casesi): Likewise.
18846 (thumb2_output_casesi): Likewise.
18847 (arm_emit_load_exclusive): Likewise.
18848 (arm_emit_store_exclusive): Likewise.
18849 (arm_expand_compare_and_swap): Likewise.
18850 (arm_evpc_neon_vuzp): Likewise.
18851 (arm_evpc_neon_vzip): Likewise.
18852 (arm_evpc_neon_vrev): Likewise.
18853 (arm_evpc_neon_vtrn): Likewise.
18854 (arm_evpc_neon_vext): Likewise.
18855 (arm_validize_comparison): Likewise.
18856 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18857 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18858 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18859 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18860 (c6x_preferred_simd_mode): Likewise.
18861 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18862 (epiphany_rtx_costs): Likewise.
18863 * config/epiphany/predicates.md (proper_comparison_operator):
18865 * config/frv/frv.c (condexec_memory_operand): Likewise.
18866 (frv_emit_move): Likewise.
18867 (output_move_single): Likewise.
18868 (output_condmove_single): Likewise.
18869 (frv_hard_regno_mode_ok): Likewise.
18870 (frv_matching_accg_mode): Likewise.
18871 * config/h8300/h8300.c (split_adds_subs): Likewise.
18872 (h8300_rtx_costs): Likewise.
18873 (h8300_print_operand): Likewise.
18874 (compute_mov_length): Likewise.
18875 (output_logical_op): Likewise.
18876 (compute_logical_op_length): Likewise.
18877 (compute_logical_op_cc): Likewise.
18878 (h8300_shift_needs_scratch_p): Likewise.
18879 (output_a_shift): Likewise.
18880 (compute_a_shift_length): Likewise.
18881 (compute_a_shift_cc): Likewise.
18882 (expand_a_rotate): Likewise.
18883 (output_a_rotate): Likewise.
18884 * config/i386/i386.c (classify_argument): Likewise.
18885 (function_arg_advance_32): Likewise.
18886 (function_arg_32): Likewise.
18887 (function_arg_64): Likewise.
18888 (function_value_64): Likewise.
18889 (ix86_gimplify_va_arg): Likewise.
18890 (ix86_legitimate_constant_p): Likewise.
18891 (put_condition_code): Likewise.
18892 (split_double_mode): Likewise.
18893 (ix86_avx256_split_vector_move_misalign): Likewise.
18894 (ix86_expand_vector_logical_operator): Likewise.
18895 (ix86_split_idivmod): Likewise.
18896 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18897 (ix86_build_const_vector): Likewise.
18898 (ix86_build_signbit_mask): Likewise.
18899 (ix86_match_ccmode): Likewise.
18900 (ix86_cc_modes_compatible): Likewise.
18901 (ix86_expand_branch): Likewise.
18902 (ix86_expand_sse_cmp): Likewise.
18903 (ix86_expand_sse_movcc): Likewise.
18904 (ix86_expand_int_sse_cmp): Likewise.
18905 (ix86_expand_vec_perm_vpermi2): Likewise.
18906 (ix86_expand_vec_perm): Likewise.
18907 (ix86_expand_sse_unpack): Likewise.
18908 (ix86_expand_int_addcc): Likewise.
18909 (ix86_split_to_parts): Likewise.
18910 (ix86_vectorize_builtin_gather): Likewise.
18911 (ix86_vectorize_builtin_scatter): Likewise.
18912 (avx_vpermilp_parallel): Likewise.
18913 (inline_memory_move_cost): Likewise.
18914 (ix86_tieable_integer_mode_p): Likewise.
18915 (x86_maybe_negate_const_int): Likewise.
18916 (ix86_expand_vector_init_duplicate): Likewise.
18917 (ix86_expand_vector_init_one_nonzero): Likewise.
18918 (ix86_expand_vector_init_one_var): Likewise.
18919 (ix86_expand_vector_init_concat): Likewise.
18920 (ix86_expand_vector_init_interleave): Likewise.
18921 (ix86_expand_vector_init_general): Likewise.
18922 (ix86_expand_vector_set): Likewise.
18923 (ix86_expand_vector_extract): Likewise.
18924 (emit_reduc_half): Likewise.
18925 (ix86_emit_i387_round): Likewise.
18926 (ix86_mangle_type): Likewise.
18927 (ix86_expand_round_sse4): Likewise.
18928 (expand_vec_perm_blend): Likewise.
18929 (canonicalize_vector_int_perm): Likewise.
18930 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18931 (expand_vec_perm_1): Likewise.
18932 (expand_vec_perm_interleave3): Likewise.
18933 (expand_vec_perm_even_odd_pack): Likewise.
18934 (expand_vec_perm_even_odd_1): Likewise.
18935 (expand_vec_perm_broadcast_1): Likewise.
18936 (ix86_vectorize_vec_perm_const_ok): Likewise.
18937 (ix86_expand_vecop_qihi): Likewise.
18938 (ix86_expand_mul_widen_hilo): Likewise.
18939 (ix86_expand_sse2_abs): Likewise.
18940 (ix86_expand_pextr): Likewise.
18941 (ix86_expand_pinsr): Likewise.
18942 (ix86_preferred_simd_mode): Likewise.
18943 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18944 * config/i386/sse.md (*andnot<mode>3): Likewise.
18945 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18946 (*<code><mode>3): Likewise.
18947 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18948 (ia64_expand_atomic_op): Likewise.
18949 (ia64_arg_type): Likewise.
18950 (ia64_mode_to_int): Likewise.
18951 (ia64_scalar_mode_supported_p): Likewise.
18952 (ia64_vector_mode_supported_p): Likewise.
18953 (expand_vec_perm_broadcast): Likewise.
18954 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18955 (iq2000_function_arg_advance): Likewise.
18956 (iq2000_function_arg): Likewise.
18957 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18958 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18959 (m68k_libcall_value): Likewise.
18960 (m68k_function_value): Likewise.
18961 (sched_attr_op_type): Likewise.
18962 * config/mcore/mcore.c (mcore_output_move): Likewise.
18963 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18965 (microblaze_function_arg): Likewise.
18966 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18967 (mips_print_operand): Likewise.
18968 (mips_mode_ok_for_mov_fmt_p): Likewise.
18969 (mips_vector_mode_supported_p): Likewise.
18970 (mips_preferred_simd_mode): Likewise.
18971 (mips_expand_vpc_loongson_even_odd): Likewise.
18972 (mips_expand_vec_unpack): Likewise.
18973 (mips_expand_vi_broadcast): Likewise.
18974 (mips_expand_vector_init): Likewise.
18975 (mips_expand_vec_reduc): Likewise.
18976 (mips_expand_msa_cmp): Likewise.
18977 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18978 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18979 (cc_flags_for_mode): Likewise.
18980 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18981 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18982 (nds32_output_casesi_pc_relative): Likewise.
18983 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18984 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18985 (nvptx_gen_unpack): Likewise.
18986 (nvptx_gen_pack): Likewise.
18987 (nvptx_gen_shuffle): Likewise.
18988 (nvptx_gen_wcast): Likewise.
18989 (nvptx_preferred_simd_mode): Likewise.
18990 * config/pa/pa.c (pa_secondary_reload): Likewise.
18991 * config/pa/predicates.md (base14_operand): Likewise.
18992 * config/powerpcspe/powerpcspe-c.c
18993 (altivec_resolve_overloaded_builtin): Likewise.
18994 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18996 (rs6000_preferred_simd_mode): Likewise.
18997 (output_vec_const_move): Likewise.
18998 (rs6000_expand_vector_extract): Likewise.
18999 (rs6000_split_vec_extract_var): Likewise.
19000 (reg_offset_addressing_ok_p): Likewise.
19001 (rs6000_legitimate_offset_address_p): Likewise.
19002 (rs6000_legitimize_address): Likewise.
19003 (rs6000_emit_set_const): Likewise.
19004 (rs6000_const_vec): Likewise.
19005 (rs6000_emit_move): Likewise.
19006 (spe_build_register_parallel): Likewise.
19007 (rs6000_darwin64_record_arg_recurse): Likewise.
19008 (swap_selector_for_mode): Likewise.
19009 (spe_init_builtins): Likewise.
19010 (paired_init_builtins): Likewise.
19011 (altivec_init_builtins): Likewise.
19012 (do_load_for_compare): Likewise.
19013 (rs6000_generate_compare): Likewise.
19014 (rs6000_expand_float128_convert): Likewise.
19015 (emit_load_locked): Likewise.
19016 (emit_store_conditional): Likewise.
19017 (rs6000_output_function_epilogue): Likewise.
19018 (rs6000_handle_altivec_attribute): Likewise.
19019 (rs6000_function_value): Likewise.
19020 (emit_fusion_gpr_load): Likewise.
19021 (emit_fusion_p9_load): Likewise.
19022 (emit_fusion_p9_store): Likewise.
19023 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19024 (fusion_gpr_mem_load): Likewise.
19025 (fusion_addis_mem_combo_load): Likewise.
19026 (fusion_addis_mem_combo_store): Likewise.
19027 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19028 (fusion_gpr_mem_load): Likewise.
19029 (fusion_addis_mem_combo_load): Likewise.
19030 (fusion_addis_mem_combo_store): Likewise.
19031 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19033 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19034 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19035 (rs6000_preferred_simd_mode): Likewise.
19036 (output_vec_const_move): Likewise.
19037 (rs6000_expand_vector_extract): Likewise.
19038 (rs6000_split_vec_extract_var): Likewise.
19039 (reg_offset_addressing_ok_p): Likewise.
19040 (rs6000_legitimate_offset_address_p): Likewise.
19041 (rs6000_legitimize_address): Likewise.
19042 (rs6000_emit_set_const): Likewise.
19043 (rs6000_const_vec): Likewise.
19044 (rs6000_emit_move): Likewise.
19045 (rs6000_darwin64_record_arg_recurse): Likewise.
19046 (swap_selector_for_mode): Likewise.
19047 (paired_init_builtins): Likewise.
19048 (altivec_init_builtins): Likewise.
19049 (rs6000_expand_float128_convert): Likewise.
19050 (emit_load_locked): Likewise.
19051 (emit_store_conditional): Likewise.
19052 (rs6000_output_function_epilogue): Likewise.
19053 (rs6000_handle_altivec_attribute): Likewise.
19054 (rs6000_function_value): Likewise.
19055 (emit_fusion_gpr_load): Likewise.
19056 (emit_fusion_p9_load): Likewise.
19057 (emit_fusion_p9_store): Likewise.
19058 * config/rx/rx.c (rx_gen_move_template): Likewise.
19059 (flags_from_mode): Likewise.
19060 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19061 (s390_slb_comparison): Likewise.
19062 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19063 (s390_vector_mode_supported_p): Likewise.
19064 (s390_cc_modes_compatible): Likewise.
19065 (s390_match_ccmode_set): Likewise.
19066 (s390_canonicalize_comparison): Likewise.
19067 (s390_emit_compare_and_swap): Likewise.
19068 (s390_branch_condition_mask): Likewise.
19069 (s390_rtx_costs): Likewise.
19070 (s390_secondary_reload): Likewise.
19071 (__SECONDARY_RELOAD_CASE): Likewise.
19072 (s390_expand_cs): Likewise.
19073 (s390_preferred_simd_mode): Likewise.
19074 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19075 * config/sh/sh.c (sh_print_operand): Likewise.
19076 (dump_table): Likewise.
19077 (sh_secondary_reload): Likewise.
19078 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19079 * config/sh/sh.md (casesi_worker_1): Likewise.
19080 (casesi_worker_2): Likewise.
19081 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19082 (fcc_comparison_operator): Likewise.
19083 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19084 (emit_soft_tfmode_cvt): Likewise.
19085 (sparc_preferred_simd_mode): Likewise.
19086 (output_cbranch): Likewise.
19087 (sparc_print_operand): Likewise.
19088 (sparc_expand_vec_perm_bmask): Likewise.
19089 (vector_init_bshuffle): Likewise.
19090 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19091 (spu_vector_mode_supported_p): Likewise.
19092 (spu_expand_insv): Likewise.
19093 (spu_emit_branch_or_set): Likewise.
19094 (spu_handle_vector_attribute): Likewise.
19095 (spu_builtin_splats): Likewise.
19096 (spu_builtin_extract): Likewise.
19097 (spu_builtin_promote): Likewise.
19098 (spu_expand_sign_extend): Likewise.
19099 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19100 (tilegx_simd_int): Likewise.
19101 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19102 (tilepro_simd_int): Likewise.
19103 * config/v850/v850.c (const_double_split): Likewise.
19104 (v850_print_operand): Likewise.
19105 (ep_memory_offset): Likewise.
19106 * config/vax/vax.c (vax_rtx_costs): Likewise.
19107 (vax_output_int_move): Likewise.
19108 (vax_output_int_add): Likewise.
19109 (vax_output_int_subtract): Likewise.
19110 * config/visium/predicates.md (visium_branch_operator): Likewise.
19111 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19112 (visium_print_operand_address): Likewise.
19113 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19114 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19115 (xtensa_expand_conditional_branch): Likewise.
19116 (xtensa_copy_incoming_a7): Likewise.
19117 (xtensa_output_literal): Likewise.
19118 * dfp.c (decimal_real_maxval): Likewise.
19119 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19122 Alan Hayward <alan.hayward@arm.com>
19123 David Sherwood <david.sherwood@arm.com>
19125 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19126 (mode_nunits_inline): Likewise.
19127 (mode_inner_inline): Likewise.
19128 (mode_unit_size_inline): Likewise.
19129 (mode_unit_precision_inline): Likewise.
19130 (emit_insn_modes_h): Likewise. Also emit a #define of the
19132 (emit_mode_wider): Add an E_ prefix to mode names.
19133 (emit_mode_complex): Likewise.
19134 (emit_mode_inner): Likewise.
19135 (emit_mode_adjustments): Likewise.
19136 (emit_mode_int_n): Likewise.
19137 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19138 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19139 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19140 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19141 (CRC32_BUILTIN, ENTRY): Likewise.
19142 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19143 (aarch64_pop_regs): Likewise.
19144 (aarch64_process_components): Likewise.
19145 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19146 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19147 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19148 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19149 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19150 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19152 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19153 (pcmpistr): Likewise.
19154 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19155 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19156 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19158 * config/rl78/rl78.c (mduc_regs): Likewise.
19159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19160 (htm_expand_builtin): Likewise.
19161 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19162 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19163 * config/xtensa/xtensa.c (print_operand): Likewise.
19164 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19165 (NUM_MODE_VECTOR_INT): Likewise.
19166 * genoutput.c (null_operand): Likewise.
19167 (output_operand_data): Likewise.
19168 * genrecog.c (print_parameter_value): Likewise.
19169 * lra.c (debug_operand_data): Likewise.
19171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19173 * dwarf2out.c (scompare_loc_descriptor_wide)
19174 (scompare_loc_descriptor_narrow): New functions, split out from...
19175 (scompare_loc_descriptor): ...here.
19176 * expmed.c (emit_store_flag_int): New function, split out from...
19177 (emit_store_flag): ...here.
19179 2017-08-30 Richard Biener <rguenther@suse.de>
19181 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19182 (dwarf2out_early_finish): Move setting of AT_pubnames from
19183 early debug output to early finish.
19185 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19187 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19188 and -mdata-region to the assembler.
19190 2017-08-30 Richard Biener <rguenther@suse.de>
19192 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19194 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19195 (dwarf2out_early_global_decl): Only generate a DIE for the
19196 abstract origin if it doesn't already exist or is a declaration DIE.
19197 (resolve_addr): Do not add the linkage name twice when
19198 generating a stub DIE for the DW_TAG_GNU_call_site target.
19200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19202 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19203 Use machine_mode rather than int for arg1_mode.
19205 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19208 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19209 that the second argument of the built-in functions to unpack
19210 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19211 switch statement instead a lot of if statements.
19212 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19213 Allow 64-bit values to be in Altivec registers as well as
19214 traditional floating point registers.
19215 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19217 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19219 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19220 MAX_REGS_PER_ADDRESS == 1.
19222 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19224 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19225 * config/i386/i386.c (ix86_option_override_internal): Simplify
19226 setting of opts->x_flag_entry.
19228 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19229 Jakub Jelinek <jakub@redhat.com>
19230 Richard Biener <rguenther@suse.de>
19232 PR tree-optimization/81503
19233 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19234 folded constant fits in the target type; reorder tests for clarity.
19236 2017-08-29 Martin Liska <mliska@suse.cz>
19238 * passes.def: Include pass_lower_switch.
19239 * stmt.c (dump_case_nodes): Remove and move to
19240 tree-switch-conversion.
19241 (case_values_threshold): Likewise.
19242 (expand_switch_as_decision_tree_p): Likewise.
19243 (emit_case_decision_tree): Likewise.
19244 (expand_case): Likewise.
19245 (balance_case_nodes): Likewise.
19246 (node_has_low_bound): Likewise.
19247 (node_has_high_bound): Likewise.
19248 (node_is_bounded): Likewise.
19249 (emit_case_nodes): Likewise.
19250 (struct simple_case_node): New struct.
19251 (add_case_node): Remove.
19252 (emit_case_dispatch_table): Use vector instead of case_list.
19253 (reset_out_edges_aux): Remove.
19254 (compute_cases_per_edge): Likewise.
19255 (expand_case): Build list of simple_case_node.
19256 (expand_sjlj_dispatch_table): Use it.
19257 * tree-switch-conversion.c (struct case_node): Moved from
19258 stmt.c and adjusted.
19259 (emit_case_nodes): Likewise.
19260 (node_has_low_bound): Likewise.
19261 (node_has_high_bound): Likewise.
19262 (node_is_bounded): Likewise.
19263 (case_values_threshold): Likewise.
19264 (reset_out_edges_aux): Likewise.
19265 (compute_cases_per_edge): Likewise.
19266 (add_case_node): Likewise.
19267 (dump_case_nodes): Likewise.
19268 (balance_case_nodes): Likewise.
19269 (expand_switch_as_decision_tree_p): Likewise.
19270 (emit_jump): Likewise.
19271 (emit_case_decision_tree): Likewise.
19272 (try_switch_expansion): Likewise.
19273 (do_jump_if_equal): Likewise.
19274 (emit_cmp_and_jump_insns): Likewise.
19275 (fix_phi_operands_for_edge): New function.
19276 (record_phi_operand_mapping): Likewise.
19277 (class pass_lower_switch): New pass.
19278 (pass_lower_switch::execute): New function.
19279 (make_pass_lower_switch): Likewise.
19280 (conditional_probability):
19281 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19282 * tree-pass.h: Add make_pass_lower_switch.
19284 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19287 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19290 2017-08-29 Richard Biener <rguenther@suse.de>
19292 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19293 we do not add a DW_AT_inline attribute twice.
19294 (gen_subprogram_die): Remove code setting DW_AT_inline on
19295 DECL_ABSTRACT_P nodes.
19297 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19299 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19300 calls to internal functions.
19301 (gimplify_modify_expr): Likewise.
19302 * tree-call-cdce.c (use_internal_fn): Likewise.
19303 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19304 (convert_to_divmod): Set the nothrow flag.
19305 * tree-if-conv.c (predicate_mem_writes): Likewise.
19306 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19307 (vectorizable_call): Likewise.
19308 (vectorizable_store): Likewise.
19309 (vectorizable_load): Likewise.
19310 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19311 (vect_recog_mask_conversion_pattern): Likewise.
19313 2017-08-29 Martin Liska <mliska@suse.cz>
19316 * gcc.c (driver_handle_option): Add new argument.
19317 * opts-common.c (handle_option): Pass
19318 target_option_override_hook.
19319 * opts-global.c (lang_handle_option): Add new option.
19320 (set_default_handlers): Add new argument.
19321 (decode_options): Likewise.
19322 * opts.c (target_handle_option): Likewise.
19323 (common_handle_option): Call target_option_override_hook.
19324 * opts.h (struct cl_option_handler_func): Add hook for
19325 target option override.
19326 (struct cl_option_handlers): Likewise.
19327 (set_default_handlers): Add new argument.
19328 (decode_options): Likewise.
19329 (common_handle_option): Likewise.
19330 (target_handle_option): Likewise.
19331 * toplev.c (toplev::main): Pass targetm.target_option.override
19334 2017-08-29 Richard Biener <rguenther@suse.de>
19335 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19337 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19338 life to the active subtree.
19340 2017-08-28 Jeff Law <law@redhat.com>
19342 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19343 derive_equivalences.
19344 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19345 Code moved into....
19346 (edge_info::derive_equivalences): New private member function
19348 * tree-ssa-dom.c (class edge_info): Changed from a struct
19349 to a class. Add ctor/dtor, methods and data members.
19350 (edge_info::edge_info): Renamed from allocate_edge_info.
19351 Initialize additional members.
19352 (edge_info::~edge_info): New.
19353 (free_dom_edge_info): Delete the edge info.
19354 (record_edge_info): Use new class & associated member functions.
19355 Tighten forms for testing for edge equivalences.
19356 (record_temporary_equivalences): Iterate over the simple
19357 equivalences rather than assuming there's only one per edge.
19358 (cprop_into_successor_phis): Iterate over the simple
19359 equivalences rather than assuming there's only one per edge.
19360 (optimize_stmt): Use operand_equal_p rather than pointer
19361 equality for mini-DSE code.
19363 2017-08-28 Nathan Sidwell <nathan@acm.org>
19365 * gcc.c (execute): Fold SIGPIPE handling into switch
19366 statement. Adjust internal error message.
19368 2017-08-28 Richard Biener <rguenther@suse.de>
19371 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19372 Do nothing for removed DIEs.
19374 2017-08-28 Richard Biener <rguenther@suse.de>
19376 PR tree-optimization/81977
19377 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19380 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19383 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19384 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19385 using targetm.gen_mem_thread_fence.
19387 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19390 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19391 predicate to register_operand. Reorder operands.
19392 (*btr<mode>): Ditto.
19393 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19394 (*btr<mode>_mask): Ditto.
19396 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19398 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19399 * config/rs6000/xmmintrin.h: New file.
19400 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19402 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19405 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19406 parameter and_insn and return it.
19407 (recombine_lvx_pattern): Insert a copy to ensure availability of
19408 the base register of the copied masking operation at the point of
19409 the instruction replacement.
19410 (recombine_stvx_pattern): Likewise.
19412 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19414 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19415 undocumented switches.
19416 (-mpower9-dform-vector): Likewise.
19417 (-mpower9-dform): Likewise.
19418 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19419 comments to delete references to -mpower9-dform* switches.
19420 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19421 Delete reference to -mpower9-dform* switches, test for
19422 -mpower9-vector instead.
19423 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19424 (OTHER_P9_VECTOR_MASKS): Likewise.
19425 (POWERPC_MASKS): Likewise.
19426 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19427 tests against -mpower9-dform* to -mpower9-vector. Delete code
19428 that checked for -mpower9-dform* consistancy with other options.
19429 Add test for -mpower9-misc to enable other power9 switches.
19430 (rs6000_init_hard_regno_mode_ok): Likewise.
19431 (rs6000_option_override_internal): Likewise.
19432 (rs6000_emit_prologue): Likewise.
19433 (rs6000_emit_epilogue): Likewise.
19434 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19435 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19436 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19438 (emit_fusion_p9_store): Likewise.
19439 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19440 resetting these macros if the assembler does not support ISA 3.0
19442 (TARGET_P9_DFORM_VECTOR): Likewise.
19443 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19444 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19446 2017-08-25 Alan Modra <amodra@gmail.com>
19448 PR rtl-optimization/81747
19449 * cse.c (cse_extended_basic_block): Don't attempt to record
19450 equivalences for degenerate conditional branches that branch
19451 to their fall-through.
19453 2017-08-24 Martin Sebor <msebor@redhat.com>
19455 PR middle-end/81908
19456 * gimple-fold.c (size_must_be_zero_p): New function.
19457 (gimple_fold_builtin_memory_op): Call it.
19459 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19461 * config/rs6000/mm_malloc.h: New file.
19463 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19465 PR tree-optimization/81913
19466 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19467 analysis when either IVs in condition can wrap.
19469 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19471 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19472 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19474 2017-08-24 Richard Biener <rguenther@suse.de>
19477 * targhooks.c (default_target_can_inline_p): Properly
19478 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19479 is present and always compare.
19480 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19481 infer -mfpmath=sse from TARGET_SSE_P.
19482 (ix86_can_inline_p): Properly use target_option_default_node when
19483 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19485 2017-08-24 Richard Biener <rguenther@suse.de>
19488 * dwarf2out.c (output_die): Handle flag_generate_offload like
19490 (output_comp_unit): Likewise.
19491 (gen_array_type_die): Likewise.
19492 (dwarf2out_early_finish): Likewise.
19493 (note_variable_value_in_expr): Likewise.
19494 (dwarf2out_finish): Likewise. Adjust assert.
19495 * cgraphunit.c (symbol_table::compile): Move setting of
19496 flag_generate_offload earlier ...
19497 (symbol_table::finalize_compilation_unit): ... here, before
19498 early debug finalization.
19500 2017-08-24 Richard Biener <rguenther@suse.de>
19502 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19503 and ipa-fnsummary.h.
19504 (ix86_can_inline_p): When ix86_fpmath flags do not match
19505 check whether the callee uses FP math at all.
19507 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19509 PR middle-end/81931
19510 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19511 instead of TYPE_PRECISION.
19513 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19514 Alan Hayward <alan.hayward@arm.com>
19515 David Sherwood <david.sherwood@arm.com>
19517 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19519 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19521 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19523 (ix86_frame::stack_realign_allocate): New field.
19524 (struct machine_frame_state): Modify comments.
19525 (machine_frame_state::sp_realigned_fp_end): New field.
19526 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19527 layout calculation.
19528 (sp_valid_at): Add assertion to assure no attempt to access invalid
19529 offset of a realigned stack.
19530 (fp_valid_at): Likewise.
19531 (choose_baseaddr): Modify comments.
19532 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19533 ix86_expand_prologue.
19534 (ix86_expand_prologue): Modify stack realignment and allocation.
19535 (ix86_expand_epilogue): Modify comments.
19536 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19537 avx2_runtime, avx512f, and avx512f_runtime.
19539 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19541 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19542 (mstackrealign): Do not init to -1.
19543 * config/i386/i386.c (ix86_option_override_internal):
19544 Check opts_set, not opts when setting default value of
19545 opts->x_ix86_force_align_arg_pointer.
19547 2017-08-23 Richard Biener <rguenther@suse.de>
19549 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19550 lang_hooks.decl_printable_name.
19551 * print-rtl-function.c (print_rtx_function): Likewise.
19552 * tree-pretty-print.c (dump_function_header): Likewise.
19554 2017-08-23 Richard Biener <rguenther@suse.de>
19557 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19558 -g0 at compile-time.
19560 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19562 PR middle-end/19706
19563 * doc/sourcebuild.texi (Other hardware attributes):
19566 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19568 PR middle-end/19706
19569 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19570 Fix single-use check.
19572 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19574 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19576 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19578 * doc/install.texi: Modify to add more details on running selected
19581 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19583 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19584 is combined with -mabi=ms.
19585 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19588 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19590 PR tree-optimization/81488
19591 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19592 and cached_basis fields.
19593 (MAX_SPREAD): New constant.
19594 (alloc_cand_and_find_basis): Initialize new fields.
19595 (clear_visited): New function.
19596 (create_phi_basis_1): Rename from create_phi_basis, set visited
19597 and cached_basis fields.
19598 (create_phi_basis): New wrapper function.
19599 (phi_add_costs_1): Rename from phi_add_costs, add spread
19600 parameter, set visited field, short-circuit when limits reached.
19601 (phi_add_costs): New wrapper function.
19602 (record_phi_increments_1): Rename from record_phi_increments, set
19604 (record_phi_increments): New wrapper function.
19605 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19606 (phi_incr_cost): New wrapper function.
19607 (all_phi_incrs_profitable_1): Rename from
19608 all_phi_incrs_profitable, set visited field.
19609 (all_phi_incrs_profitable): New wrapper function.
19611 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19612 Alan Hayward <alan.hayward@arm.com>
19613 David Sherwood <david.sherwood@arm.com>
19615 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19616 that takes the outer and inner modes.
19617 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19618 comparison as the canonical test for a paradoxical subreg.
19619 * combine.c (simplify_set): Use paradoxical_subreg_p.
19620 (make_extraction): Likewise.
19621 (force_to_mode): Likewise.
19622 (rtx_equal_for_field_assignment_p): Likewise.
19623 (gen_lowpart_for_combine): Likewise.
19624 (simplify_comparison): Likewise.
19625 * cse.c (equiv_constant): Likewise.
19626 * expmed.c (store_bit_field_1): Likewise.
19627 * final.c (alter_subreg): Likewise.
19628 * fwprop.c (propagate_rtx): Likewise.
19629 (forward_propagate_subreg): Likewise.
19630 * ira-conflicts.c (ira_build_conflicts): Likewise.
19631 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19632 * lra-constraints.c (curr_insn_transform): Likewise.
19633 (split_reg): Likewise.
19634 * lra-eliminations.c (move_plus_up): Likewise.
19635 (lra_eliminate_regs_1): Likewise.
19636 * recog.c (general_operand): Likewise.
19637 * ree.c (combine_reaching_defs): Likewise.
19638 * reload.c (push_reload): Likewise.
19639 (find_reloads): Likewise.
19640 * reload1.c (elimination_effects): Likewise.
19641 (compute_reload_subreg_offset): Likewise.
19642 (choose_reload_regs): Likewise.
19643 * rtlanal.c (subreg_lsb_1): Likewise.
19644 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19645 (simplify_subreg): Likewise.
19646 * var-tracking.c (track_loc_p): Likewise.
19647 * emit-rtl.c (byte_lowpart_offset): Likewise.
19648 (paradoxical_subreg_p): Delete out-of-line definition.
19650 2017-08-22 Jeff Law <law@redhat.com>
19652 PR tree-optimization/81741
19653 PR tree-optimization/71947
19654 * tree-ssa-dom.c: Include tree-inline.h.
19655 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19656 equivalences if one is more expensive to compute than the other.
19657 * tree-ssa-scopedtables.h (class const_or_copies): Make
19658 record_const_or_copy_raw method private.
19659 (class avail_exprs_stack): New method simplify_binary_operation.
19660 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19661 avail_exprs_stack::simplify_binary_operation as needed.
19662 (avail_exprs_stack::simplify_binary_operation): New function.
19664 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19666 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19667 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19668 (DOT_SYMBOLS): Likewise.
19669 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19670 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19671 (RS6000_ABI_NAME): Likewise.
19672 (TARGET_CMODEL): Likewise.
19673 (TOC_SECTION_ASM_OP): Likewise.
19674 (SET_CMODEL): New macro.
19675 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19677 2017-08-22 Richard Biener <rguenther@suse.de>
19679 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19680 to allow for free-lang-data replacements similar to verify_type_variant.
19682 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19683 Alan Hayward <alan.hayward@arm.com>
19684 David Sherwood <david.sherwood@arm.com>
19686 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19687 VOIDmode for the LABEL_REF.
19689 2017-08-22 Richard Biener <rguenther@suse.de>
19691 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19693 2017-08-22 Richard Biener <rguenther@suse.de>
19695 * common.opt (feliminate-dwarf2-dups): Ignore.
19696 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19697 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19698 same_die_p_wrap, compute_section_prefix,
19699 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19700 (comdat_symbol_id, comdat_symbol_number): Likewise.
19701 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19703 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19704 (output_die): Mark unreachable path unreachable.
19705 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19706 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19707 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19708 (dwarf2out_early_finish): Likewise.
19710 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19712 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19714 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19717 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19718 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19719 (avr_attribute_table) <io, io_low, address>: Initialize
19720 .decl_required with true.
19722 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19724 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19725 undocumented debugging options.
19726 (-mvsx-scalar-double): Likewise.
19727 (-mallow-df-permute): Likewise.
19728 (-mvectorize-builtins): Likewise.
19729 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19730 (rs6000_builtin_vectorized_function): Likewise.
19731 (rs6000_builtin_md_vectorized_function): Likewise.
19732 (rs6000_opt_vars): Likewise.
19734 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19737 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19738 (*btrq_imm): Rename from *btrq.
19739 (*btcq_imm): Rename from *btcq.
19740 (btsc): New code attribute.
19741 (*<btsc><mode>): New insn pattern.
19742 (*btr<mode>): Ditto.
19743 (*<btsc><mode>_mask): New insn_and_split pattern.
19744 (*btr<mode>_mask): Ditto.
19746 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19747 Alan Hayward <alan.hayward@arm.com>
19748 David Sherwood <david.sherwood@arm.com>
19750 * function.c (pad_below): Simplify padding calculation.
19752 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19753 Alan Hayward <alan.hayward@arm.com>
19754 David Sherwood <david.sherwood@arm.com>
19756 * target.def (function_prologue): Remove frame size argument.
19757 (function_epilogue): Likewise.
19758 * doc/tm.texi: Regenerate.
19759 * final.c (final_start_function): Update call to function_prologue.
19760 (final_end_function): Update call to function_epilogue.
19761 (default_function_pro_epilogue): Remove frame size argument.
19762 * output.h (default_function_pro_epilogue): Likewise.
19763 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19764 (arm_output_function_prologue): Likewise.
19765 * config/frv/frv.c (frv_function_prologue): Likewise.
19766 (frv_function_epilogue): Likewise.
19767 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19768 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19769 (ia64_output_function_epilogue): Likewise.
19770 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19771 (m32r_output_function_epilogue): Likewise.
19772 * config/microblaze/microblaze.c (microblaze_function_prologue)
19773 (microblaze_function_epilogue): Likewise.
19774 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19775 (mips_output_function_epilogue): Likewise.
19776 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19777 (mmix_target_asm_function_epilogue): Likewise.
19778 * config/msp430/msp430.c (msp430_start_function): Likewise.
19779 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19780 (nds32_asm_function_epilogue): Likewise.
19781 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19782 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19783 (pa_output_function_epilogue): Likewise.
19784 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19785 (rs6000_output_function_epilogue): Likewise.
19786 * config/rl78/rl78.c (rl78_start_function): Likewise.
19787 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19788 (rs6000_output_function_epilogue): Likewise.
19789 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19790 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19791 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19792 (sparc_asm_function_epilogue): Likewise.
19794 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19796 * tree.h (type_has_mode_precision_p): New function.
19797 * convert.c (convert_to_integer_1): Use it.
19798 * expr.c (expand_expr_real_2): Likewise.
19799 (expand_expr_real_1): Likewise.
19800 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19801 * match.pd: Likewise.
19802 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19803 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19804 * tree-tailcall.c (process_assignment): Likewise.
19805 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19806 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19807 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19808 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19809 (vectorizable_assignment): Likewise.
19810 (vectorizable_shift): Likewise.
19811 (vectorizable_operation): Likewise.
19812 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19814 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19816 * match.pd: Add pow (C, x) simplification.
19818 2017-08-21 Richard Biener <rguenther@suse.de>
19820 PR tree-optimization/81900
19821 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19822 for blocks with abnormal predecessors.
19823 (compute_antic): Do not set visited flag prematurely.
19825 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19828 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19830 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19832 * stor-layout.h (vector_type_mode): Move to...
19833 * tree.h (vector_type_mode): ...here.
19834 * stor-layout.c (vector_type_mode): Move to...
19835 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19837 2017-08-21 Richard Biener <rguenther@suse.de>
19839 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19840 register_external_die hooks.
19841 (debug_false_tree_charstarstar_uhwistar): Declare.
19842 (debug_nothing_tree_charstar_uhwi): Likewise.
19843 * debug.c (do_nothing_debug_hooks): Adjust.
19844 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19845 (debug_nothing_tree_charstar_uhwi): Likewise.
19846 * dbxout.c (dbx_debug_hooks): Adjust.
19847 (xcoff_debug_hooks): Likewise.
19848 * sdbout.c (sdb_debug_hooks): Likewise.
19849 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19850 * dwarf2out.c (macinfo_label_base): New global.
19851 (dwarf2out_register_external_die): New function for the
19852 register_external_die hook.
19853 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19854 (dwarf2_debug_hooks): Use them.
19855 (dwarf2_lineno_debug_hooks): Adjust.
19856 (struct die_struct): Add with_offset flag.
19857 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19858 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19859 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19860 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19861 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19862 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19863 defining section names for the early LTO debug variants.
19864 (reset_indirect_string): New helper.
19865 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19866 (print_dw_val): Add support for offsetted symbol references.
19867 (get_ultimate_context): Split out from is_cxx.
19868 (is_cxx): Use get_ultimate_context.
19869 (is_fortran): Add decl overload.
19870 (compute_comp_unit_symbol): Split out worker from
19871 compute_section_prefix.
19872 (compute_section_prefix): Call compute_comp_unit_symbol and
19873 set comdat_type_p here.
19874 (output_die): Skip DIE symbol output for the LTO added one.
19875 Handle DIE symbol references with offset.
19876 (output_comp_unit): Guard section name mangling properly.
19877 For LTO debug sections emit a symbol at the section beginning
19878 which we use to refer to its DIEs.
19879 (add_abstract_origin_attribute): For DIEs registered via
19880 dwarf2out_register_external_die directly refer to the early
19881 DIE rather than indirectly through the shadow one we created.
19882 Remove obsolete call to dwarf2out_abstract_function for
19883 non-function/block origins.
19884 (gen_array_type_die): When generating early LTO debug do
19885 not emit DW_AT_string_length.
19886 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19887 late when in LTO. As suggested place a gcc_unreachable for
19888 the DECL_ABSTRACT_P case.
19889 (gen_subprogram_die): Avoid another specification DIE
19890 for early built declarations/definitions for the late LTO case.
19891 (gen_variable_die): Add type references for late duplicated VLA dies
19893 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19894 we have the abstract instance already.
19895 (process_scope_var): Adjust decl DIE contexts in LTO which
19896 first puts them in limbo.
19897 (gen_decl_die): Do not generate type DIEs late apart from
19898 types for VLAs or for decls we do not yet have a DIE. Do not
19899 call dwarf2out_abstract_function late.
19900 (dwarf2out_early_global_decl): Make sure to create DIEs
19901 for abstract instances of a decl first.
19902 (dwarf2out_late_global_decl): Adjust comment.
19903 (output_macinfo_op): With multiple macro sections use
19904 macinfo_label_base to distinguish labels.
19905 (output_macinfo): Likewise. Update macinfo_label_base.
19906 Pass in the line info label.
19907 (note_variable_value_in_expr): When generating LTO resolve
19908 all variable values here by generating DIEs as needed.
19909 (init_sections_and_labels): Add early LTO debug flag parameter
19910 and generate different sections and names if set. Add generation
19911 counter for the labels so we can have multiple of them.
19912 (reset_dies): Helper to allow DIEs to be output multiple times.
19913 (dwarf2out_finish): When outputting DIEs to the fat part of an
19914 LTO object first reset DIEs.
19915 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19916 (modified_type_die): Check for decl_ultimate_origin being self
19918 (gen_type_die_with_usage): Likewise.
19919 (gen_typedef_die): Allow decl_ultimate_origin being self.
19920 (set_decl_abstract_flags): Remove.
19921 (set_block_abstract_flags): Likewise.
19922 (dwarf2out_abstract_function): Treat the early generated DIEs
19923 as the abstract copy and only add DW_AT_inline and
19924 DW_AT_artificial here and call set_decl_origin_self.
19925 If the DIE has an abstract origin don't do anything.
19926 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19927 if we have none yet (Go fails to build one, PR78628).
19928 (variably_modified_type_p): Prevent endless recursion for Ada
19929 cyclic pointer types.
19930 * lto-streamer-in.c: Include debug.h.
19931 (dref_queue): New global.
19932 (lto_read_tree_1): Stream in DIE references.
19933 (lto_input_tree): Register DIE references.
19934 (input_function): Stream DECL_DEBUG_ARGS.
19935 * lto-streamer-out.c: Include debug.h.
19936 (lto_write_tree_1): Output DIE references.
19937 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19938 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19939 (output_function): Stream DECL_DEBUG_ARGS.
19940 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19941 Stream DECL_ABSTRACT_ORIGIN.
19942 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19943 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19944 DECL_CONTEXT for file-scope decls.
19945 * lto-streamer.h (struct dref_entry): Declare.
19946 (dref_queue): Likewise.
19947 * cfgexpand.c (pass_expand::execute): Do not call the
19948 outlining_inline_function hook here.
19949 * lto-wrapper.c (debug_obj): New global.
19950 (tool_cleanup): Unlink it if required.
19951 (debug_objcopy): New function.
19952 (run_gcc): Handle early debug sections in the IL files by
19953 extracting them to separate files, partially linkin them and
19954 feeding the result back as result to the linker.
19955 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19956 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19957 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19958 sections into a separate segment.
19959 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19961 (darwin_asm_dwarf_section): Likewise.
19962 (darwin_asm_output_dwarf_offset): Likewise.
19963 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19965 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19966 Alan Hayward <alan.hayward@arm.com>
19967 David Sherwood <david.sherwood@arm.com>
19969 * read-md.h (md_reader::record_potential_iterator_use): Replace
19970 pointer argument with an rtx and an index.
19971 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19972 (apply_mode_iterator): Likewise.
19973 (apply_code_iterator): Likewise.
19974 (apply_int_iterator): Likewise.
19975 (apply_subst_iterator): Likewise.
19976 (record_iterator_use): Likewise.
19977 (record_attribute_use): Likewise.
19978 (md_reader::record_potential_iterator_use): Likewise. Update calls
19979 to record_iterator_use and apply_iterator.
19980 (iterator_use): Replace ptr with x and index.
19981 (attribute_use): Likewise.
19982 (apply_attribute_uses): Update calls to apply_iterator.
19983 (apply_iterators): Likewise. Update initialization of iterator_use.
19984 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19985 and record_potential_iterator_use.
19986 (rtx_reader::read_rtx_operand): Likewise.
19988 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19989 Alan Hayward <alan.hayward@arm.com>
19990 David Sherwood <david.sherwood@arm.com>
19992 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19995 2017-08-21 Richard Biener <rguenther@suse.de>
19997 PR middle-end/81884
19998 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19999 at struct end conservatively when comparing common bases.
20001 2017-08-21 Richard Biener <rguenther@suse.de>
20003 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20004 (mem_ref_in_stmt): Remove.
20005 (determine_max_movement): Use ref index to get at the reference.
20006 (invariantness_dom_walker::before_dom_children): Deal with
20007 lim data already initialized.
20008 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20010 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20012 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20013 (TARGET_ISA_ROUND): Ditto.
20014 (TARGET_ROUND): Ditto.
20015 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20016 * config/i386/i386.md: Ditto.
20017 * config/i386/sse.md: Ditto.
20018 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20019 with OPTION_MASK_ISA_SSE4_1.
20021 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20024 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20025 __builtin_ia32_lzcnt_u16.
20027 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20030 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20031 (rs6000_set_current_function): Rewrite function to use it.
20033 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20036 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20037 and TYPE_WARN_IF_NOT_ALIGN.
20038 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20039 (handle_warn_if_not_align): New.
20040 (place_union_field): Call handle_warn_if_not_align.
20041 (place_field): Call handle_warn_if_not_align.
20042 Copy TYPE_WARN_IF_NOT_ALIGN.
20043 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20044 (layout_type): Likewise.
20045 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20047 (tree_decl_common): Add warn_if_not_align.
20048 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20049 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20050 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20051 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20052 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20053 * doc/extend.texi: Document warn_if_not_aligned attribute.
20054 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20056 2017-08-17 Martin Liska <mliska@suse.cz>
20059 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20060 (get_data_dependence): Use it as pointer type.
20061 (distribute_loop): Likewise.
20063 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20065 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20066 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20067 (p8_vmrgow_<mode>_direct): New define_insn.
20068 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20069 handle endianness for vmrgew and vmrgow permute patterns.
20071 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20073 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20074 * config/rs6000/rs6000-cpus.def: Remove comment.
20075 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20076 (POWERPC_MASKS): Likewise.
20077 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20078 use of TARGET_VSX_TIMODE.
20079 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20080 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20081 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20082 (rs6000_option_override_internal): Remove dead code.
20083 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20084 (rs6000_legitimize_reload_address): Likewise.
20085 (rs6000_legitimate_address_p): Likewise.
20086 (rs6000_opt_masks): Delete "vsx-timode".
20087 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20088 from function comment.
20089 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20090 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20091 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20093 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20094 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20095 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20096 (VSX_TI): Likewise.
20098 (define_peephole2): Likewise.
20100 2017-08-17 Martin Sebor <msebor@redhat.com>
20103 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20104 past the end of an array.
20105 (test_pp_format): Add test cases.
20107 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20109 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20110 missing ECF_NOTHROW flags.
20112 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20115 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20116 operands residing in integer registers.
20117 (*vsx_le_perm_load_<mode>): Likewise.
20118 (*vsx_le_perm_store_<mode>): Likewise.
20119 (define_peephole2): Add peepholes to optimize the above.
20121 2017-08-17 Marek Polacek <polacek@redhat.com>
20123 PR middle-end/81814
20124 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20125 to mimic what shorten_compare did. Change the return type to bool.
20126 (fold_cond_expr_with_comparison): Update call to
20127 operand_equal_for_comparison_p.
20128 (fold_ternary_loc): Likewise.
20130 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20132 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20134 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20136 2017-08-17 Richard Biener <rguenther@suse.de>
20138 * tree-ssa-structalias.c (solve_graph): When propagating
20139 to successors update the graphs succ edges and avoid duplicate work.
20141 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20144 * config/i386/i386.c (ix86_option_override_internal): Save target
20145 specific options after ix86_stack_protector_guard_reg was changed.
20147 2017-08-17 Richard Biener <rguenther@suse.de>
20149 PR tree-optimization/81827
20150 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20151 (new_var_info): Initialize it conservatively.
20152 (get_call_vi): Mark register vars.
20153 (new_scalar_tmp_constraint_exp): Likewise.
20154 (handle_rhs_call): Likewise.
20155 (handle_const_call): Likewise.
20156 (create_function_info_for): Likewise.
20157 (solve_constraints): Sort varinfos to separate register from
20158 non-register vars to pack points-to solution bitmaps during
20161 2017-08-17 Marek Polacek <polacek@redhat.com>
20163 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20165 2017-08-17 Richard Biener <rguenther@suse.de>
20167 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20168 to true when overflow is undefined and we saturated the result.
20170 2017-08-17 Alan Modra <amodra@gmail.com>
20173 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20174 Don't use store multiple if only one reg needs saving.
20175 (interesting_frame_related_regno): New function.
20176 (rs6000_frame_related): Don't emit frame info for regs that
20178 (rs6000_emit_epilogue): Likewise.
20180 2017-08-16 Nathan Sidwell <nathan@acm.org>
20182 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20183 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20184 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20185 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20186 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20187 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20188 (verify_type): Adjust for TYPE_BINFO move.
20189 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20190 process TYPE_BINFO directly.
20191 (hash_tree): Likewise.
20192 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20194 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20197 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20199 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20201 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20204 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20205 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20206 Add ix86_binary_operator_ok to insn constraint.
20207 (*iordi_1_bts): Ditto.
20208 (*xordi_1_btc): Ditto.
20209 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20210 Update corresponding peephole2 pattern.
20214 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20216 PR tree-optimization/81832
20217 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20218 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20220 2017-08-16 Marek Polacek <polacek@redhat.com>
20223 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20224 regarding size_zero_node.
20226 2017-08-16 Martin Liska <mliska@suse.cz>
20229 * config.gcc: Respect previously set extra_objs in case
20232 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20234 PR tree-optimization/81835
20235 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20236 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20237 not depend on the phi.
20239 2017-08-16 Alan Modra <amodra@gmail.com>
20241 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20244 2017-08-16 Alan Modra <amodra@gmail.com>
20246 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20247 (save_reg_p): ..into this. Update all callers.
20248 (first_reg_to_save): Simplify.
20250 2017-08-16 Alan Modra <amodra@gmail.com>
20252 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20255 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20259 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20260 constant count if that count is less than 32.
20262 2017-08-15 Nathan Sidwell <nathan@acm.org>
20264 * gcc.c (execute): Emit friendlier message if inferior is killed
20265 by an external cause.
20267 2017-08-15 Richard Biener <rguenther@suse.de>
20269 PR tree-optimization/81790
20270 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20271 CONSTRUCTORs from simplifying and VN.
20273 2017-08-14 Martin Sebor <msebor@redhat.com>
20275 * builtin-attrs.def: Add comments.
20277 2017-08-14 Martin Sebor <msebor@redhat.com>
20280 * doc/extend.texi (attribute nonstring): Document new attribute.
20282 2017-08-14 Martin Sebor <msebor@redhat.com>
20285 * tree-diagnostic.c (default_tree_printer): Handle %G.
20286 * gimple-pretty-print.h (percent_G_format): Declare new function.
20287 * gimple-pretty-print.c (percent_G_format): Define.
20288 * tree-pretty-print.c (percent_K_format): Add argument.
20290 2017-08-14 Martin Sebor <msebor@redhat.com>
20292 PR translation/79998
20293 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20294 Remove a stray space.
20296 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20299 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20300 (*iordi_1_bts): Ditto.
20301 (*xordi_1_btc): Ditto.
20303 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20306 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20307 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20309 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20310 quoted strings, and make more translator-friendly.
20311 (darwin_rs6000_override_options): Likewise.
20312 (rs6000_option_override_internal): Likewise.
20313 (rs6000_return_in_memory): Fix overlong line.
20314 (init_cmulative_args): Use quoted strings, and make more
20315 translator-friendly.
20316 (rs6000_pass_by_reference): Fix overlong line.
20317 (def_builtin): Use quoted strings.
20318 (altivec_expand_predicate_builtin): Use quoted strings, and make
20319 more translator-friendly.
20320 (htm_expand_builtin): Use quoted strings.
20321 (cpu_expand_builtin): Use quoted strings, and make more
20322 translator-friendly.
20323 (altivec_expand_builtin): Likewise.
20324 (paired_expand_predicate_builtin): Likewise.
20325 (rs6000_invalid_builtin): Likewise.
20326 (builtin_function_type): Use quoted strings.
20327 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20328 more translator-friendly.
20329 (rs6000_trampoline_init): Likewise.
20330 (rs6000_handle_altivec_attribute): Likewise.
20331 (rs6000_inner_target_options): Use quoted strings.
20332 (rs6000_disable_incompatible_switches): Likewise.
20333 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20334 strings, and make more translator-friendly.
20335 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20337 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20339 PR tree-optimization/81799
20340 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20341 cond_expr to simple gimple operand.
20343 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20345 PR middle-end/46932
20346 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20348 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20352 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20353 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20354 TARGET_GASISR_PROLOGUES.
20355 * config/avr/avr.c (avr_option_override): Same.
20356 (avr_pass_pre_proep::execute): Same.
20358 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20361 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20362 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20364 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20366 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20367 define_insn_and_split. Split to a memory load from 0 in
20368 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20369 using PTR mode iterator.
20370 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20371 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20372 (*add_tp_<mode>): Redefine as define_insn_and_split.
20373 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20374 address space. Merge with *add_tp_x32 using PTR mode iterator.
20375 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20376 Split to an add with a memory load from 0 in
20377 DEFAULT_TLS_SEG_REG address space.
20379 2017-08-12 Andrew Pinski <apinski@cavium.com>
20381 * config/aarch64/aarch64-option-extensions.def (rdma):
20382 Fix feature string to what Linux prints out in /proc/cpuinfo.
20384 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20387 * dwarf2out.c (modified_type_die): For C typedef types that have
20388 an ultimate origin, process the ultimate origin instead of the
20390 (gen_typedef_die): Assert that input DECLs have no ultimate
20392 (gen_type_die_with_usage): For typedef variants that have an
20393 ultimate origin, just call gen_decl_die on the original DECL.
20394 (process_scope_var): Avoid creating DIEs for local typedefs and
20395 concrete static variables.
20397 2017-08-12 Alan Modra <amodra@gmail.com>
20401 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20402 match gnu-user.h startfile.
20403 (ENDFILE_LINUX_SPEC): Similarly.
20405 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20408 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20410 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20412 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20413 * config/aarch64/aarch64.md (mov<mode>): Change.
20414 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20415 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20416 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20418 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20420 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20421 for storage order barriers.
20423 2017-08-11 Martin Liska <mliska@suse.cz>
20426 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20427 variables of void type.
20429 2017-08-11 Martin Liska <mliska@suse.cz>
20431 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20432 TARGET_SUPPORTS_ALIASES.
20433 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20434 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20435 (optimize_weakref): Likewise.
20436 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20437 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20438 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20440 2017-08-11 Martin Liska <mliska@suse.cz>
20443 * config/i386/i386.c (make_resolver_func): Do complete
20444 refactoring of the function.
20446 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20449 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20450 * config/i386/i386.c (ix86_stack_protect_guard): Use
20451 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20452 * doc/invoke.texi (x86 Options): Document
20453 -mstack-protector-guard-symbol= option.
20455 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20457 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20458 * config/i386/i386.c (ix86_split_stack_guard): New function.
20459 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20460 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20461 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20462 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20463 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20464 (split_stack_space_check): Call ix86_split_stack_guard.
20466 2017-08-10 Martin Sebor <msebor@redhat.com>
20468 * print-tree.c (print_node): Print location using the established
20470 Replace spaces with colons.
20471 (debug_raw, debug): Ditto.
20473 2017-08-10 Martin Sebor <msebor@redhat.com>
20476 * pretty-print.c (pp_format): Correct the handling of %s precision.
20478 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20481 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20483 (ix86_finalize_stack_frame_flags): This. Also clear
20484 frame_pointer_needed if -fno-omit-frame-pointer is used without
20486 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20487 with ix86_finalize_stack_frame_flags.
20488 (ix86_expand_epilogue): Likewise.
20489 (ix86_expand_split_stack_prologue): Likewise.
20490 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20492 2017-08-10 Martin Liska <mliska@suse.cz>
20495 * c-attribs.c (handle_target_attribute):
20496 Report warning for an empty string argument of target attribute.
20498 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20501 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20503 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20504 or DECL_NONLOCAL labels.
20505 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20506 or DECL_NONLOCAL labels here.
20508 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20510 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20511 to indicate when early gimple folding has been disabled.
20512 (rs6000_gimple_fold_builtin): Add debug content.
20513 (rs6000_invalid_builtin): Fix whitespace.
20514 (rs6000_expand_builtin): Fix whitespace.
20515 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20517 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20520 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20521 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20523 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20525 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20526 (qdf24xx): Likewise.
20527 * config/aarch64/aarch64.md: Include falkor.md.
20528 * config/aarch64/falkor.md: New.
20530 2017-08-09 Marek Polacek <polacek@redhat.com>
20533 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20534 * diagnostic.c (emit_diagnostic): Add a comment.
20535 (emit_diagnostic_valist): New function.
20537 2017-08-09 Marek Polacek <polacek@redhat.com>
20540 * input.c (make_location): New overload.
20541 * input.h (make_location): Declare.
20543 2017-08-08 Alan Modra <amodra@gmail.com>
20544 H.J. Lu <hongjiu.lu@intel.com>
20547 * gcc.c (NO_PIE_SPEC): Delete.
20548 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20550 (LINK_PIE_SPEC): ..to here.
20551 (LINK_COMMAND_SPEC): Support -no-pie.
20552 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20553 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20555 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20556 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20557 (ENDFILE_CRTEND_SPEC): Similarly.
20559 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20562 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20563 (mstack-protector-guard-offset=): Ditto.
20564 * config/i386/i386.c (ix86_option_override): Handle
20565 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20567 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20568 ix86_stack_protect_guard_offset variables.
20569 (TARGET_STACK_PROTECT_GUARD): Always define.
20570 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20571 and -mstack-protector-guard-offset= options.
20573 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20575 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20576 boundary case for the last candidate.
20578 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20580 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20581 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20583 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20585 PR middle-end/19706
20586 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20587 * config/aarch64/aarch64-builtins.c
20588 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20589 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20590 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20592 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20593 Andrew Pinski <pinskia@gmail.com>
20595 PR middle-end/19706
20596 * internal-fn.def (XORSIGN): New.
20597 * optabs.def (xorsign_optab): New.
20598 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20599 (convert_expand_mult_copysign): New.
20600 (pass_optimize_widening_mul::execute): Call
20601 convert_expand_mult_copysign.
20603 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20605 PR tree-optimization/81354
20606 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20607 Insert on edges rather than explicitly creating landing pads.
20608 (analyze_candidates_and_replace): Commit edge inserts.
20610 2017-08-08 Richard Biener <rguenther@suse.de>
20612 PR middle-end/81719
20613 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20614 (expand_simple_operations): Also look through ADDR_EXPRs with
20615 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20617 2017-08-08 Richard Biener <rguenther@suse.de>
20619 PR tree-optimization/81723
20620 * tree-vect-slp.c (struct bst_traits): New hash traits.
20621 (bst_fail): New global.
20622 (vect_build_slp_tree_2): New worker, split out from ...
20623 (vect_build_slp_tree): ... this now wrapping it with using
20624 bst_fail set to cache SLP tree build fails. Properly handle
20626 (vect_analyze_slp_instance): Allocate and free bst_fail.
20628 2017-08-08 Martin Liska <mliska@suse.cz>
20631 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20632 LABEL_DECLs that can be from a different function.
20634 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20636 PR tree-optimization/81744
20637 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20638 loop's number of iterations.
20640 2017-08-08 Martin Liska <mliska@suse.cz>
20642 * asan.c: Include header files.
20643 * attribs.c (build_decl_attribute_variant): New function moved
20645 (build_type_attribute_qual_variant): Likewise.
20646 (cmp_attrib_identifiers): Likewise.
20647 (simple_cst_list_equal): Likewise.
20648 (omp_declare_simd_clauses_equal): Likewise.
20649 (attribute_value_equal): Likewise.
20650 (comp_type_attributes): Likewise.
20651 (build_type_attribute_variant): Likewise.
20652 (lookup_ident_attribute): Likewise.
20653 (remove_attribute): Likewise.
20654 (merge_attributes): Likewise.
20655 (merge_type_attributes): Likewise.
20656 (merge_decl_attributes): Likewise.
20657 (merge_dllimport_decl_attributes): Likewise.
20658 (handle_dll_attribute): Likewise.
20659 (attribute_list_equal): Likewise.
20660 (attribute_list_contained): Likewise.
20661 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20662 (lookup_attribute_by_prefix): Likewise.
20663 * bb-reorder.c: Include header files.
20664 * builtins.c: Likewise.
20665 * calls.c: Likewise.
20666 * cfgexpand.c: Likewise.
20667 * cgraph.c: Likewise.
20668 * cgraphunit.c: Likewise.
20669 * convert.c: Likewise.
20670 * dwarf2out.c: Likewise.
20671 * final.c: Likewise.
20672 * fold-const.c: Likewise.
20673 * function.c: Likewise.
20674 * gimple-expr.c: Likewise.
20675 * gimple-fold.c: Likewise.
20676 * gimple-pretty-print.c: Likewise.
20677 * gimple.c: Likewise.
20678 * gimplify.c: Likewise.
20679 * hsa-common.c: Likewise.
20680 * hsa-gen.c: Likewise.
20681 * internal-fn.c: Likewise.
20682 * ipa-chkp.c: Likewise.
20683 * ipa-cp.c: Likewise.
20684 * ipa-devirt.c: Likewise.
20685 * ipa-fnsummary.c: Likewise.
20686 * ipa-inline.c: Likewise.
20687 * ipa-visibility.c: Likewise.
20689 * lto-cgraph.c: Likewise.
20690 * omp-expand.c: Likewise.
20691 * omp-general.c: Likewise.
20692 * omp-low.c: Likewise.
20693 * omp-offload.c: Likewise.
20694 * omp-simd-clone.c: Likewise.
20695 * opts-global.c: Likewise.
20696 * passes.c: Likewise.
20697 * predict.c: Likewise.
20698 * sancov.c: Likewise.
20699 * sanopt.c: Likewise.
20700 * symtab.c: Likewise.
20701 * toplev.c: Likewise.
20702 * trans-mem.c: Likewise.
20703 * tree-chkp.c: Likewise.
20704 * tree-eh.c: Likewise.
20705 * tree-into-ssa.c: Likewise.
20706 * tree-object-size.c: Likewise.
20707 * tree-parloops.c: Likewise.
20708 * tree-profile.c: Likewise.
20709 * tree-ssa-ccp.c: Likewise.
20710 * tree-ssa-live.c: Likewise.
20711 * tree-ssa-loop.c: Likewise.
20712 * tree-ssa-sccvn.c: Likewise.
20713 * tree-ssa-structalias.c: Likewise.
20714 * tree-ssa.c: Likewise.
20715 * tree-streamer-in.c: Likewise.
20716 * tree-vectorizer.c: Likewise.
20717 * tree-vrp.c: Likewise.
20718 * tsan.c: Likewise.
20719 * ubsan.c: Likewise.
20720 * varasm.c: Likewise.
20721 * varpool.c: Likewise.
20722 * tree.c: Remove functions moved to attribs.[ch].
20723 * tree.h: Likewise.
20724 * config/aarch64/aarch64.c: Add attrs.h header file.
20725 * config/alpha/alpha.c: Likewise.
20726 * config/arc/arc.c: Likewise.
20727 * config/arm/arm.c: Likewise.
20728 * config/avr/avr.c: Likewise.
20729 * config/bfin/bfin.c: Likewise.
20730 * config/c6x/c6x.c: Likewise.
20731 * config/cr16/cr16.c: Likewise.
20732 * config/cris/cris.c: Likewise.
20733 * config/darwin.c: Likewise.
20734 * config/epiphany/epiphany.c: Likewise.
20735 * config/fr30/fr30.c: Likewise.
20736 * config/frv/frv.c: Likewise.
20737 * config/ft32/ft32.c: Likewise.
20738 * config/h8300/h8300.c: Likewise.
20739 * config/i386/winnt.c: Likewise.
20740 * config/ia64/ia64.c: Likewise.
20741 * config/iq2000/iq2000.c: Likewise.
20742 * config/lm32/lm32.c: Likewise.
20743 * config/m32c/m32c.c: Likewise.
20744 * config/m32r/m32r.c: Likewise.
20745 * config/m68k/m68k.c: Likewise.
20746 * config/mcore/mcore.c: Likewise.
20747 * config/microblaze/microblaze.c: Likewise.
20748 * config/mips/mips.c: Likewise.
20749 * config/mmix/mmix.c: Likewise.
20750 * config/mn10300/mn10300.c: Likewise.
20751 * config/moxie/moxie.c: Likewise.
20752 * config/msp430/msp430.c: Likewise.
20753 * config/nds32/nds32-isr.c: Likewise.
20754 * config/nds32/nds32.c: Likewise.
20755 * config/nios2/nios2.c: Likewise.
20756 * config/nvptx/nvptx.c: Likewise.
20757 * config/pa/pa.c: Likewise.
20758 * config/pdp11/pdp11.c: Likewise.
20759 * config/powerpcspe/powerpcspe.c: Likewise.
20760 * config/riscv/riscv.c: Likewise.
20761 * config/rl78/rl78.c: Likewise.
20762 * config/rx/rx.c: Likewise.
20763 * config/s390/s390.c: Likewise.
20764 * config/sh/sh.c: Likewise.
20765 * config/sol2.c: Likewise.
20766 * config/sparc/sparc.c: Likewise.
20767 * config/spu/spu.c: Likewise.
20768 * config/stormy16/stormy16.c: Likewise.
20769 * config/tilegx/tilegx.c: Likewise.
20770 * config/tilepro/tilepro.c: Likewise.
20771 * config/v850/v850.c: Likewise.
20772 * config/vax/vax.c: Likewise.
20773 * config/visium/visium.c: Likewise.
20774 * config/xtensa/xtensa.c: Likewise.
20776 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20779 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20780 constraints since the -mupper-regs-* switches have been
20782 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20783 into a vector from a double word element that was extracted from
20784 another vector, and eliminate extra XXPERMDI instructions.
20785 (vsx_concat_<mode>_2): Likewise.
20786 (vsx_concat_<mode>_3): Likewise.
20787 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20788 concat to allow optimizing inserts from previous extracts.
20790 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20792 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20793 memory reference to a SSP offset in TLS address space.
20794 (ix86_print_operand) <case '@'>: Remove.
20795 (ix86_print_operand_punct_valid_p): Remove '@' code.
20796 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20797 UNSPEC_SP_TLS_TEST.
20798 (stack_tls_protect_set_<mode>): Remove.
20799 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20800 (stack_tls_protect_test_<mode>): Remove.
20801 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20803 2017-08-07 Olivier Hainque <hainque@adacore.com>
20806 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20808 2017-08-07 Douglas Rupp <rupp@adacore.com>
20810 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20811 variable was referenced as multidir in command.
20813 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20816 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20819 2017-08-07 Martin Liska <mliska@suse.cz>
20821 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20822 * config/rl78/rl78.c: Add include of attribs.h.
20823 * config/sh/sh.c: Likewise.
20824 * config/v850/v850.c: Likewise.
20826 2017-08-07 Tom de Vries <tom@codesourcery.com>
20828 PR middle-end/78266
20829 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20831 2017-08-07 Martin Liska <mliska@suse.cz>
20833 * config/mips/mips.c: Include attribs.h.
20835 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20838 * doc/invoke.texi: Document change in behvaior for -Ofast for
20841 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20843 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20845 (aarch64_pop_regs): Likewise.
20846 (aarch64_gen_load_pair): Likewise.
20847 (aarch64_save_callee_saves): Likewise.
20848 (aarch64_restore_callee_saves): Likewise.
20850 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20852 * config/i386/i386.c: Revert the last change.
20854 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20857 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20859 (ix86_finalize_stack_frame_flags): This. Also clear
20860 frame_pointer_needed if -fno-omit-frame-pointer is used without
20862 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20863 with ix86_finalize_stack_frame_flags.
20864 (ix86_expand_epilogue): Likewise.
20865 (ix86_expand_split_stack_prologue): Likewise.
20867 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20870 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20871 to P_AES for Westmere.
20873 2017-08-07 Jonathan Yong <10walls@gmail.com>
20875 * config/i386/mingw.opt (fset-stack-executable): Removed.
20876 * config/i386/cygming.opt (fset-stack-executable): Moved
20878 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20880 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20882 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20884 2017-08-07 Marek Polacek <polacek@redhat.com>
20886 PR middle-end/81737
20887 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20889 2017-08-07 Martin Liska <mliska@suse.cz>
20891 * attribs.h (canonicalize_attr_name): New function.
20892 (cmp_attribs): Move from c-format.c and adjusted.
20893 (is_attribute_p): Moved from tree.h.
20894 * tree-inline.c: Add new includes.
20895 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20896 (private_is_attribute_p): Remove.
20897 (private_lookup_attribute): Likewise.
20898 (private_lookup_attribute_by_prefix): Simplify.
20899 (remove_attribute): Use is_attribute_p.
20900 * tree.h: Remove removed declarations.
20902 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20904 PR middle-end/81698
20905 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20906 instead of computing it in the function. Formatting fix.
20907 (expand_case): Don't rely on default_edge being the first edge,
20908 clear it if removing it, pass default_edge to
20909 emit_case_dispatch_table.
20910 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20913 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20915 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20916 insn in the function, emit NOP after the insn.
20918 2017-08-06 Tom de Vries <tom@codesourcery.com>
20920 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20923 2017-08-06 Tom de Vries <tom@codesourcery.com>
20925 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20928 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20930 PR tree-optimization/57371
20931 * match.pd: New pattern.
20933 2017-08-04 Marek Polacek <polacek@redhat.com>
20935 PR middle-end/81695
20936 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20937 perform the computation in offset_int.
20939 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20941 PR tree-optimization/81136
20942 * tree-vectorizer.h: Include tree-hash-traits.h.
20943 (vec_base_alignments): New typedef.
20944 (vec_info): Add a base_alignments field.
20945 (vect_record_base_alignments): Declare.
20946 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20948 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20949 (create_data_ref): Add an is_conditional_in_stmt argument.
20950 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20951 the is_conditional_in_stmt field.
20952 (data_ref_loc): Add an is_conditional_in_stmt field.
20953 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20954 (find_data_references_in_stmt): Update call to create_data_ref.
20955 (graphite_find_data_references_in_stmt): Likewise.
20956 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20957 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20958 (vect_record_base_alignment): New function.
20959 (vect_record_base_alignments): Likewise.
20960 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20961 for nested statements even if we fail to compute a misalignment.
20962 Use pooled base alignments for unconditional references.
20963 (vect_find_same_alignment_drs): Compare base addresses instead
20965 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20966 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20968 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20970 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20971 Add an explicit name for the enum. Use auto_vec for slp_instances
20972 and grouped_stores.
20973 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20975 (_bb_vec_info): Add a constructor and destructor.
20976 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20977 (destroy_loop_vec_info): Delete.
20978 (vect_destroy_datarefs): Likewise.
20979 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20980 (vec_info::vec_info): New function.
20981 (vec_info::~vec_info): Likewise.
20982 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20983 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20984 destroy_loop_vec_info.
20985 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20986 (_loop_vec_info::_loop_vec_info): ...this.
20987 (destroy_loop_vec_info): Replace with...
20988 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20989 the stmt_vec_infos. Leave handling of vec_info information to its
20990 destructor. Remove explicit vector releases.
20991 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20992 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20993 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20994 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20995 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20996 (destroy_bb_vec_info): Replace with...
20997 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20998 information to its destructor.
20999 (vect_slp_analyze_bb_1): Use new and delete instead of
21000 new_bb_vec_info and destroy_bb_vec_info.
21001 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21004 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21006 * tree-data-ref.h (subscript): Add access_fn field.
21007 (data_dependence_relation): Add could_be_independent_p.
21008 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21009 (same_access_functions): Move to tree-data-ref.c.
21010 * tree-data-ref.c (ref_contains_union_access_p): New function.
21011 (access_fn_component_p): Likewise.
21012 (access_fn_components_comparable_p): Likewise.
21013 (dr_analyze_indices): Add a reference to access_fn_component_p.
21014 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21016 (constant_access_functions): Likewise.
21017 (add_other_self_distances): Likewise.
21018 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21019 (initialize_data_dependence_relation): Use XCNEW and remove
21020 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21021 of access functions that have the same type. Allow the
21022 subsequence to end with different bases in some circumstances.
21023 Record the chosen access functions in SUB_ACCESS_FN.
21024 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21025 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21026 (subscript_dependence_tester_1): Likewise dra and drb.
21027 (build_classic_dist_vector): Update calls accordingly.
21028 (subscript_dependence_tester): Likewise.
21029 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21030 DDR_COULD_BE_INDEPENDENT_P.
21031 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21032 comp_alias_ddrs instead of may_alias_ddrs.
21033 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21035 (vect_analyze_data_ref_dependence): Use it if
21036 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21037 distance vectors if that fails.
21038 (dependence_distance_ge_vf): New function.
21039 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21040 LOOP_VINFO_MAY_ALIAS_DDRS.
21042 2017-08-04 Richard Biener <rguenther@suse.de>
21044 PR middle-end/81705
21045 * fold-const.c (fold_binary_loc): Properly restrict
21046 minus_var0 && minus_var1 case when associating undefined overflow
21049 2017-08-04 Tom de Vries <tom@codesourcery.com>
21051 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21053 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21055 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21056 Don't start diagnostic messages with a capital letter.
21057 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21059 (rs6000_invalid_builtin): Likewise.
21060 (rs6000_trampoline_init): Likewise.
21062 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21065 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21066 after setting changeable df flags.
21068 2017-08-03 Richard Biener <rguenther@suse.de>
21070 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21071 up if the use is in USE - X.
21073 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21075 * toplev.c (dumpfile.h): New include.
21076 (internal_error_reentered): New static function. Use it...
21077 (internal_error_function): ...here to handle reentered internal_error.
21079 2017-08-03 Richard Biener <rguenther@suse.de>
21081 PR middle-end/81148
21082 * fold-const.c (split_tree): Add minus_var and minus_con
21083 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21084 here but always use minus_*.
21085 (associate_trees): Assert we never associate with MINUS_EXPR
21086 and NULL first operand. Do not recurse for PLUS_EXPR operands
21087 when associating as MINUS_EXPR either.
21088 (fold_binary_loc): Track minus_var and minus_con.
21090 2017-08-03 Tom de Vries <tom@codesourcery.com>
21093 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21094 ACCEL_COMPILER, apply finish_options on
21095 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21097 2017-08-03 Tom de Vries <tom@codesourcery.com>
21100 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21101 function_entry_patch_area_size > 0.
21103 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21106 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21107 instead of 10??LU, perform unit multiplication in wide_int,
21108 don't change alloc_object_size_limit if the limit is larger
21111 PR tree-optimization/81655
21112 PR tree-optimization/81588
21113 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21114 the case when ranges[i].low and high are 1 for unsigned type with
21117 PR middle-end/81052
21118 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21119 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21121 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21123 * tree-vrp.h: Add include guard.
21125 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21128 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21129 (ud2): New insn pattern.
21130 * config/i386/i386.c (ix86_expand_epilogue):
21131 For naked functions, generate ud2 instead of trap insn.
21133 2017-08-02 Marek Polacek <polacek@redhat.com>
21136 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21138 2017-08-02 Tom de Vries <tom@codesourcery.com>
21139 Cesar Philippidis <cesar@codesourcery.com>
21141 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21142 Add missing edge probabilities.
21144 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21146 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21147 Correct endianness.
21149 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21151 PR middle-end/79499
21152 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21153 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21154 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21155 prologue_seq sequences - if any.
21157 2017-08-02 Richard Biener <rguenther@suse.de>
21159 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21160 via vectors if supported, integer extracts via punning if supported
21161 or otherwise vector extracts.
21163 2017-08-02 Richard Biener <rguenther@suse.de>
21165 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21167 (bitmap_insert_into_set): ... this.
21169 2017-08-02 Richard Biener <rguenther@suse.de>
21171 PR tree-optimization/81633
21173 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21175 PR tree-optimization/71752
21176 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21178 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21180 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21181 (machine_function::call_ms2sysv_pad_out): Remove field.
21182 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21183 (ix86_compute_frame_layout): Likewise.
21185 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21188 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21189 attribute with interrupt attribute.
21191 2017-08-01 Andrew Pinski <apinski@cavium.com>
21193 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21194 BIT_INSERT_EXPR's operand 1
21195 to see if the types precision matches.
21197 2017-08-01 Martin Liska <mliska@suse.cz>
21199 PR middle-end/70140
21200 * builtins.c (expand_builtin_memcpy_args): Remove.
21201 (expand_builtin_memcpy): Call newly added function
21202 expand_builtin_memory_copy_args.
21203 (expand_builtin_memcpy_with_bounds): Likewise.
21204 (expand_builtin_mempcpy): Remove last argument.
21205 (expand_builtin_mempcpy_with_bounds): Likewise.
21206 (expand_builtin_memory_copy_args): New function created from
21207 expand_builtin_mempcpy_args with small modifications.
21208 (expand_builtin_mempcpy_args): Remove.
21209 (expand_builtin_stpcpy): Remove unused argument.
21210 (expand_builtin): Likewise.
21211 (expand_builtin_with_bounds): Likewise.
21213 2017-08-01 Martin Liska <mliska@suse.cz>
21216 Make mempcpy more optimal (PR middle-end/70140).
21218 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21222 __builtin_vec_cmpne verify both arguments are compatible vectors
21223 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21224 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21225 move computation of aligned to after checking the argument types.
21229 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21230 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21233 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21234 Doug Rupp <rupp@adacore.com>
21235 Olivier Hainque <hainque@adacore.com>
21237 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21238 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21240 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21241 for TARGET_OS_CPP_BUILTIN.
21242 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21243 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21245 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21246 passing required abi options to the assembler for EABI configurations.
21247 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21248 of .text.hot and .text.unlikely sections for kernel modules when
21249 using ARM style exceptions.
21250 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21251 options. Add EXTRA_CC1_SPEC.
21252 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21254 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21256 (ARM_TARGET2_DWARF_FORMAT): Define.
21257 * config/arm/t-vxworks: Adjust multilib control to removal of the
21258 Diab command line options.
21260 2017-08-01 Martin Liska <mliska@suse.cz>
21262 PR gcov-profile/81561
21263 * gcov.c (unblock): Make unblocking safe as we need to preserve
21264 index correspondence of blocks and block_lists.
21266 2017-08-01 Richard Biener <rguenther@suse.de>
21268 PR tree-optimization/81181
21269 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21270 (compute_antic): ... end of iteration here.
21272 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21274 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21275 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21276 (ftree-slp-vectorize): Likewise.
21277 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21278 can no longer be set independent of flag_tree_loop_vectorize.
21279 * omp-general.c (emp_max_vf): Likewise.
21280 * opts.c (enable_fdo_optimizations): Remove references to
21281 flag_tree_vectorize, these are now implicit.
21282 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21283 and leave it for the options machinery.
21285 2017-08-01 Martin Liska <mliska@suse.cz>
21287 PR middle-end/70140
21288 * builtins.c (expand_builtin_memcpy_args): Remove.
21289 (expand_builtin_memcpy): Call newly added function
21290 expand_builtin_memory_copy_args.
21291 (expand_builtin_memcpy_with_bounds): Likewise.
21292 (expand_builtin_mempcpy): Remove last argument.
21293 (expand_builtin_mempcpy_with_bounds): Likewise.
21294 (expand_builtin_memory_copy_args): New function created from
21295 expand_builtin_mempcpy_args with small modifications.
21296 (expand_builtin_mempcpy_args): Remove.
21297 (expand_builtin_stpcpy): Remove unused argument.
21298 (expand_builtin): Likewise.
21299 (expand_builtin_with_bounds): Likewise.
21301 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21304 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21305 print "ds:" only for immediates in generic address space.
21307 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21310 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21311 (ix86_function_ok_for_sibcall): Return false for naked functions.
21313 2017-08-01 Richard Biener <rguenther@suse.de>
21315 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21316 (compute_antic): Seed worklist with exit block predecessors.
21317 * cfganal.c (dfs_find_deadend): For a cycle return the source
21318 of the edge closing it.
21320 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21322 * config/aarch64/aarch64.c
21323 (aarch64_can_const_movi_rtx_p): Move 0 check.
21325 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21327 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21328 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21330 * match.pd: Ditto in address comparison pattern.
21332 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21334 PR tree-optimization/81627
21335 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21336 closed ssa form for store-store chain.
21338 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21340 PR tree-optimization/81620
21341 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21342 for store-store chain.
21344 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21346 PR tree-optimization/81588
21347 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21348 ranges[i].in_p, invert comparison code ccode. For >/>=,
21349 swap rhs1 and rhs2 and comparison code unconditionally,
21350 for </<= don't do that. Don't swap rhs1/rhs2 again if
21351 ranges[i].in_p, instead invert comparison code ccode if
21352 opcode or oe->rank is BIT_IOR_EXPR.
21355 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21356 a direct optab to conversion optab.
21357 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21358 with GET_MODE_INNER as last argument instead of optab_handler.
21359 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21360 vector extraction if possible and optab is available.
21361 * expr.c (store_constructor): Use convert_optab_handler instead
21362 of optab_handler. Use vector initialization from smaller
21363 vectors if possible and optab is available.
21364 * tree-vect-stmts.c (vectorizable_load): Likewise.
21365 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21366 now have two modes.
21367 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21368 of vec_init from half-sized vectors with the same element mode.
21369 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21370 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21371 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21372 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21373 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21374 after mode in gen_vec_extract* calls.
21375 (vec_extract<mode>): Renamed to ...
21376 (vec_extract<mode><ssescalarmodelower>): ... this.
21377 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21378 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21379 element mode after mode in gen_vec_init* calls.
21380 (VEC_INIT_HALF_MODE): New mode iterator.
21381 (vec_init<mode>): Renamed to ...
21382 (vec_init<mode><ssescalarmodelower>): ... this.
21383 (vec_init<mode><ssehalfvecmodelower>): New expander.
21384 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21385 (vec_extractv2sfsf): ... this.
21386 (vec_initv2sf): Renamed to ...
21387 (vec_initv2sfsf): ... this.
21388 (vec_extractv2si): Renamed to ...
21389 (vec_extractv2sisi): ... this.
21390 (vec_initv2si): Renamed to ...
21391 (vec_initv2sisi): ... this.
21392 (vec_extractv4hi): Renamed to ...
21393 (vec_extractv4hihi): ... this.
21394 (vec_initv4hi): Renamed to ...
21395 (vec_initv4hihi): ... this.
21396 (vec_extractv8qi): Renamed to ...
21397 (vec_extractv8qiqi): ... this.
21398 (vec_initv8qi): Renamed to ...
21399 (vec_initv8qiqi): ... this.
21400 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21401 (vec_init<mode>): Renamed to ...
21402 (vec_init<mode><VEC_base_l>): ... this.
21403 (vec_extract<mode>): Renamed to ...
21404 (vec_extract<mode><VEC_base_l>): ... this.
21405 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21406 (vec_initv2sfsf): ... this.
21407 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21408 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21409 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21410 element mode after mode in gen_vec_init* calls.
21411 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21412 (vec_init<mode><Vel>): ... this.
21413 (vec_extract<mode>): Renamed to ...
21414 (vec_extract<mode><Vel>): ... this.
21415 * config/aarch64/iterators.md (Vel): New mode attribute.
21416 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21417 Add element mode after mode in gen_vec_extract* calls.
21418 * config/s390/vector.md (non_vec_l): New mode attribute.
21419 (vec_extract<mode>): Renamed to ...
21420 (vec_extract<mode><non_vec_l>): ... this.
21421 (vec_init<mode>): Renamed to ...
21422 (vec_init<mode><non_vec_l>): ... this.
21423 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21424 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21426 * config/arm/iterators.md (V_elem_l): New mode attribute.
21427 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21428 (vec_extract<mode><V_elem_l>): ... this.
21429 (vec_extractv2di): Renamed to ...
21430 (vec_extractv2didi): ... this.
21431 (vec_init<mode>): Renamed to ...
21432 (vec_init<mode><V_elem_l>): ... this.
21433 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21434 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21435 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21436 Add element mode after gen_vec_extract* calls.
21437 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21438 (vec_init<mode><unitmode>): ... this.
21439 (vec_extract<mode>): Renamed to ...
21440 (vec_extract<mode><unitmode>): ... this.
21441 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21442 (vec_init<mode><unitmode>): ... this.
21443 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21444 (vec_initv2sfsf): ... this.
21445 (vec_extractv2sf): Renamed to ...
21446 (vec_extractv2sfsf): ... this.
21447 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21448 Add element mode after gen_vec_extract* calls.
21449 * config/mips/mips.md (unitmode): New mode iterator.
21450 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21451 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21452 * config/spu/spu.md (inner_l): New mode attribute.
21453 (vec_init<mode>): Renamed to ...
21454 (vec_init<mode><inner_l>): ... this.
21455 (vec_extract<mode>): Renamed to ...
21456 (vec_extract<mode><inner_l>): ... this.
21457 * config/sparc/sparc.md (veltmode): New mode iterator.
21458 (vec_init<VMALL:mode>): Renamed to ...
21459 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21460 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21461 (vec_initv2sisi): ... this.
21462 (vec_initv2sf): Renamed to ...
21463 (vec_initv2sfsf): ... this.
21464 (vec_extractv2sf): Renamed to ...
21465 (vec_extractv2sfsf): ... this.
21466 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21467 (vec_init<mode>): Renamed to ...
21468 (vec_init<mode><VEC_base_l>): ... this.
21469 (vec_extract<mode>): Renamed to ...
21470 (vec_extract<mode><VEC_base_l>): ... this.
21471 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21472 (vec_initv2sfsf): ... this.
21473 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21474 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21475 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21476 gen_vec_init* calls.
21478 2017-08-01 Richard Biener <rguenther@suse.de>
21480 PR tree-optimization/81297
21481 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21482 TREE_OVERFLOW from INTEGER_CSTs.
21484 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21486 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21488 2017-07-31 Carl Love <cel@us.ibm.com>
21490 * config/rs6000/rs6000-c: Add support for built-in functions
21491 vector signed char vec_xl_be (signed long long, signed char *);
21492 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21493 vector signed int vec_xl_be (signed long long, signed int *);
21494 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21495 vector signed long long vec_xl_be (signed long long, signed long long *);
21496 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21497 vector signed short vec_xl_be (signed long long, signed short *);
21498 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21499 vector double vec_xl_be (signed long long, double *);
21500 vector float vec_xl_be (signed long long, float *);
21501 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21502 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21503 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21505 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21506 (altivec_expand_builtin): Add switch statement to call
21507 altivec_expand_xl_be for each builtin.
21508 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21509 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21510 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21511 __builtin_vsx_le_be_v16qi.
21512 * doc/extend.texi: Update the built-in documentation file for the
21513 new built-in functions.
21515 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21518 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21520 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21522 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21524 * config.gcc: Add z14.
21525 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21526 CPU model numbers for z13s and z14.
21527 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21529 * config/s390/s390-opts.h (enum processor_type): Rename
21530 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21531 * config/s390/s390.c (processor_table): Add field for CPU name to
21532 be passed to Binutils.
21533 (s390_asm_output_machine_for_arch): Use the new field in
21534 processor_table for Binutils.
21535 (s390_expand_builtin): Replace arch12 with z14.
21536 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21537 (s390_get_sched_attrmask): Likewise.
21538 (s390_get_unit_mask): Likewise.
21539 * config/s390/s390.opt: Add z14 to processor_type enum.
21541 2017-07-31 Martin Jambor <mjambor@suse.cz>
21544 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21545 regardless of optimization level.
21547 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21548 Martin Liska <mliska@suse.cz>
21550 * predict.def: Remove old comment and adjust probability.
21551 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21552 PREDICT statements.
21554 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21557 * config/i386/i386.c (ix86_function_naked): New function.
21558 (ix86_can_use_return_insn_p): Return false for naked functions.
21559 (ix86_expand_prologue): Skip prologue for naked functions.
21560 (ix86_expand_epilogue): Skip epilogue for naked functions
21561 and emit trap instruction.
21562 (ix86_warn_func_return): New function.
21563 (ix86_attribute_table): Add "naked" attribute specification.
21564 (TARGET_WARN_FUNC_RETURN): Define.
21565 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21567 2017-07-31 Martin Liska <mliska@suse.cz>
21569 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21570 (dump_gimple_bb_header): Always dump BB info.
21571 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21573 2017-07-31 Martin Liska <mliska@suse.cz>
21576 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21577 also with current_function_decl non-null equality.
21579 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21582 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21583 change type to the element type, instead add eltype variable and
21584 use it where we are interested in the element type.
21586 PR tree-optimization/81603
21587 * ipa-polymorphic-call.c
21588 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21589 offset arithmetic in offset_int, bail out if the resulting bit offset
21590 doesn't fit into shwi.
21592 2017-07-31 Martin Liska <mliska@suse.cz>
21594 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21595 (gimplify_save_expr): Fix comment.
21597 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21600 * config/i386/i386.c (ix86_function_arg): Update arguments for
21602 (ix86_compute_frame_layout): Set the initial stack offset to
21603 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21604 INCOMING_FRAME_SP_OFFSET.
21605 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21606 stack before exception handler returns.
21607 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21608 the 'ERROR_CODE' for exception handler.
21610 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21612 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21613 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21614 (ASM_OUTPUT_REG_POP): Ditto.
21615 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21616 instead of asm_fprintf to output pure string.
21618 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21620 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21621 to imported_module_or_decl hook.
21622 (debug_nothing_tree_tree_tree_bool): Remove.
21623 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21624 * debug.c (do_nothing_debug_hooks): Use
21625 debug_nothing_tree_tree_tree_bool_bool instead of
21626 debug_nothing_tree_tree_tree_bool.
21627 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21628 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21629 * sdbout.c (sdb_debug_hooks): Likewise.
21630 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21631 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21633 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21634 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21635 attribute, don't add anything.
21637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21639 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21640 (fold_build2_stat_loc): Likewise.
21641 (fold_build3_stat_loc): Likewise.
21642 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21643 (fold_build1_loc): Remove macro.
21644 (fold_build2_loc): Likewise.
21645 (fold_build3_loc): Likewise.
21647 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21649 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21650 (gimple_build_debug_bind_source_stat): Likewise.
21651 * gimple.h (gimple_build_debug_bind): Remove macro.
21652 (gimple_build_debug_bind_source): Likewise.
21654 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21656 * bitmap.c (bitmap_alloc): Adjust.
21657 (bitmap_gc_alloc): Likewise.
21658 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21662 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21663 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21664 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21665 (bitmap_gc_alloc_stat): Likewise.
21666 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21668 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21670 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21671 * rtl.h (shallow_copy_rtx): Remove macro.
21673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21675 * emit-rtl.c (gen_raw_REG): Adjust.
21676 * gengenrtl.c (gendef): Likewise.
21677 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21678 * rtl.h (rtx_alloc): Remove macro.
21680 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21682 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21683 (build_tree_list_stat): Likewise.
21684 * tree.h (build_tree_list): Remove macro.
21685 (build_tree_list_vec): Likewise.
21687 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21689 * tree.c (make_vector_stat): Remove _stat from name.
21690 (build_vector_stat): Likewise.
21691 * tree.h (make_vector_stat): Remove macro.
21692 (build_vector_stat): Likewise.
21694 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21696 * tree.h (build_var_debug_value): Remove prototype.
21698 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21700 * tree.c (tree_cons_stat): Remove _stat from name.
21701 * tree.h (tree_cons): Remove macro.
21703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21705 * tree.c (build_vl_exp_stat): Remove _stat from name.
21706 * tree.h (build_vl_exp): Remove macro.
21708 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21710 * tree.c (build_decl_stat): Remove _stat from name.
21711 * tree.h (build_decl): Remove macro.
21713 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21715 * gimple.c (gimple_build_with_ops_stat): Adjust.
21716 (gimple_alloc_stat): Remove _stat from name.
21717 * gimple.h (gimple_alloc): Remove macro.
21719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21721 * tree.c (make_tree_vec_stat): Remove _stat from name.
21722 (grow_tree_vec_stat): Likewise.
21723 * tree.h (make_tree_vec_stat): Adjust prototype.
21724 (grow_tree_vec_stat): Likewise.
21725 (make_tree_vec): Remove macro.
21726 (grow_tree_vec): Likewise.
21728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21730 * fold-const.c (fold_build1_stat_loc): Adjust.
21731 (fold_build2_stat_loc): Likewise.
21732 (fold_build3_stat_loc): Likewise.
21733 * tree.c (build0_stat): Remove _stat from name.
21734 (build1_stat): Likewise.
21735 (build2_stat): Likewise.
21736 (build3_stat): Likewise.
21737 (build4_stat): Likewise.
21738 (build5_stat): Likewise.
21739 * tree.h (build1_loc): Remove macro, and rename _stat function
21741 (build2_loc): Likewise.
21742 (build3_loc): Likewise.
21743 (build4_loc): Likewise.
21744 (build5_loc): Likewise.
21746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21748 * tree.c (make_int_cst_stat): Remove _stat from name.
21749 * tree.h (make_int_cst_stat): Adjust prototype.
21750 (make_int_cst): Remove macro.
21752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21754 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21755 * tree.h (make_tree_binfo_stat): Adjust prototype.
21756 (make_tree_binfo): Remove.
21758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21760 * tree.c (copy_node_stat): Rename to copy_node.
21761 (build_distinct_type_copy): Adjust.
21762 * tree.h (copy_node_stat): Adjust prototype.
21763 (copy_node): Remove macro.
21765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21767 * tree.c (make_node_stat): rename to make_node.
21768 (build_tree_list_stat): Adjust.
21769 (build0_stat): Likewise.
21770 (build2_stat): Likewise.
21771 (build3_stat): Likewise.
21772 (build4_stat): Likewise.
21773 (build5_stat): Likewise.
21774 (build_decl_stat): Likewise.
21775 * tree.h (make_node_stat): Adjust prototype.
21776 (make_node): remove macro.
21778 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21780 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21781 (PPC_FEATURE2_SCV): Likewise.
21782 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21784 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21786 * config/aarch64/aarch64.c
21787 (aarch64_internal_mov_immediate): Add new special pattern.
21788 * config/aarch64/aarch64.md (*movdi_aarch64):
21789 Add reg/32bit const mov case.
21791 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21792 Richard Sandiford <richard.sandiford@linaro.org>
21794 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21795 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21796 Add integer and movi cases.
21797 (movi-split-hf-df-sf split, fp16): New.
21798 (enabled): Added TARGET_FP_F16INST.
21799 * config/aarch64/iterators.md (GPF_HF): New.
21800 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21802 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21804 * config/aarch64/aarch64.c
21805 (aarch64_simd_container_mode): Add prototype.
21806 (aarch64_expand_mov_immediate): Add HI support.
21807 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21808 (aarch64_can_const_movi_rtx_p): New.
21809 (aarch64_preferred_reload_class):
21810 Remove restrictions of using FP registers for certain SIMD operations.
21811 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21812 (aarch64_valid_floating_const): Add integer move validation.
21813 (aarch64_simd_imm_scalar_p): Remove.
21814 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21815 (aarch64_legitimate_constant_p): Expand list of supported cases.
21816 * config/aarch64/aarch64-protos.h
21817 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21818 (aarch64_reinterpret_float_as_int): New.
21819 (aarch64_simd_imm_scalar_p): Remove.
21820 * config/aarch64/constraints.md (Uvi): New.
21821 (Dd): Split into Ds and new Dd.
21822 * config/aarch64/aarch64.md (*movsi_aarch64):
21824 (*movdi_aarch64): Add SIMD mov case.
21826 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21828 * tree-predcom.c: (struct chain): Handle store-store chain in which
21829 stores for elimination only store loop invariant values.
21830 (execute_pred_commoning_chain): Ditto.
21831 (prepare_initializers_chain_store_elim): Ditto.
21832 (prepare_finalizers): Ditto.
21833 (is_inv_store_elimination_chain): New function.
21834 (initialize_root_vars_store_elim_1): New function.
21836 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21838 * tree-predcom.c: Revise general description of the pass.
21839 (enum chain_type): New enum type for store elimination.
21840 (struct chain): New field supporting store elimination.
21841 (struct component): Ditto.
21842 (dump_chain): Dump store-stores chain.
21843 (release_chain): Release resources.
21844 (split_data_refs_to_components): Compute and create component
21845 contains only stores for elimination.
21846 (get_chain_last_ref_at): New function.
21847 (make_invariant_chain): Initialization.
21848 (make_rooted_chain): Specify chain type in parameter and record it.
21849 (add_looparound_copies): Skip for store-stores chain.
21850 (determine_roots_comp): Compute type of chain and pass it to
21852 (initialize_root_vars_store_elim_2): New function.
21853 (finalize_eliminated_stores): New function.
21854 (remove_stmt): Handle store for elimination.
21855 (execute_pred_commoning_chain): Execute predictive commoning on
21856 store-store chains.
21857 (determine_unroll_factor): Skip unroll for store-stores chain.
21858 (prepare_initializers_chain_store_elim): New function.
21859 (prepare_initializers_chain): Hanlde store-store chain.
21860 (prepare_finalizers_chain, prepare_finalizers): New function.
21861 (tree_predictive_commoning_loop): Return integer value indicating
21862 if loop is unrolled or lcssa form is corrupted.
21863 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21865 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21867 * tree-predcom.c (initialize_root): Delete.
21868 (execute_pred_commoning_chain): Initialize root vars and replace
21869 reference of non-combined chain directly, rather than call above
21872 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21874 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21875 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21877 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21879 * tree-predcom.c (struct chain): New field init_seq.
21880 (release_chain): Release init_seq.
21881 (prepare_initializers_chain): Record intialization stmts in above
21883 (insert_init_seqs): New function.
21884 (tree_predictive_commoning_loop): Call insert_init_seqs.
21886 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21888 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21890 2017-07-28 Richard Biener <rguenther@suse.de>
21892 * match.pd: Remove superfluous :c.
21893 * genmatch.c (simplify::id): Add member.
21894 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21896 (current_id): New global.
21897 (dt_node::parent): Move from ...
21898 (dt_operand::parent): ... here. Add for_id member.
21899 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21900 (decision_tree::find_node): Relax order requirement when
21901 merging DT_TRUE nodes to ones inbetween the current simplify
21902 and the one we try to merge with. Add diagnostic whenever
21903 we need to enforce pattern order by not merging.
21904 (decision_tree::insert): Set current_id.
21905 (decision_tree::print_node): Dump parent node and for_id.
21906 (parser::last_id): Add member.
21907 (parser::push_simplify): Assign unique id.
21908 (parser::parser): Initialize last_id.
21910 2017-07-28 Martin Liska <mliska@suse.cz>
21913 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21914 gimple_build_debug_bind.
21916 2017-07-28 Richard Biener <rguenther@suse.de>
21918 PR tree-optimization/81502
21919 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21921 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21923 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21924 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21925 for BIT_FIELD_REF args.
21926 * fold-const.c (make_bit_field_ref): Likewise.
21927 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21929 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21932 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21933 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21934 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21935 Or it into SANITIZER_UNDEFINED.
21936 * ubsan.c: Include gimple-fold.h and varasm.h.
21937 (ubsan_expand_ptr_ifn): New function.
21938 (instrument_pointer_overflow): New function.
21939 (maybe_instrument_pointer_overflow): New function.
21940 (instrument_object_size): Formatting fix.
21941 (pass_ubsan::execute): Call instrument_pointer_overflow
21942 and maybe_instrument_pointer_overflow.
21943 * internal-fn.c (expand_UBSAN_PTR): New function.
21944 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21945 * sanitizer.def (__ubsan_handle_pointer_overflow,
21946 __ubsan_handle_pointer_overflow_abort): New builtins.
21947 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21948 * internal-fn.def (UBSAN_PTR): New internal function.
21949 * opts.c (sanitizer_opts): Add pointer-overflow.
21950 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21951 * fold-const.c (build_range_check): Compute pointer range check in
21952 integral type if pointer arithmetics would be needed. Formatting
21955 2017-07-28 Martin Liska <mliska@suse.cz>
21958 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21959 parameters that are of a variable-length.
21961 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21963 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21965 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21966 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21967 (CRT_CALL_STATIC_FUNCTION): Likewise.
21968 (ASM_DEFAULT_SPEC): New define.
21969 (ASM_SPEC32): Likewise.
21970 (ASM_SPEC64): Likewise.
21971 (ASM_SPEC_COMMON): Likewise.
21972 (ASM_SPEC): Likewise.
21973 (INVALID_64BIT): Likewise.
21974 (LINK_OS_DEFAULT_SPEC): Likewise.
21975 (LINK_OS_SPEC32): Likewise.
21976 (LINK_OS_SPEC64): Likewise.
21977 (POWERPC_LINUX): Likewise.
21978 (PTRDIFF_TYPE): Likewise.
21979 (RESTORE_FP_PREFIX): Likewise.
21980 (RESTORE_FP_SUFFIX): Likewise.
21981 (SAVE_FP_PREFIX): Likewise.
21982 (SAVE_FP_SUFFIX): Likewise.
21983 (SIZE_TYPE): Likewise.
21984 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21985 (TARGET_64BIT): Likewise.
21986 (TARGET_64BIT): Likewise.
21987 (TARGET_AIX): Likewise.
21988 (WCHAR_TYPE_SIZE): Likewise.
21989 (WCHAR_TYPE): Undefine.
21990 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21991 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21992 (CPP_OS_RTEMS_SPEC): Delete.
21993 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21994 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21996 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21998 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22000 PR tree-optimization/81578
22001 * tree-parloops.c (build_new_reduction): Bail out if
22002 reduction_code isn't one of the standard OpenMP reductions.
22003 Move the details printing after that decision.
22005 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22007 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22008 related to reload_in_progress.
22009 (splat_input_operand): Likewise.
22010 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22012 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22014 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22015 (TARGET_INSTANTIATE_DECLS): Likewise.
22016 (legitimate_indexed_address_p): Delete reload_in_progress code.
22017 (rs6000_debug_legitimate_address_p): Likewise.
22018 (rs6000_eliminate_indexed_memrefs): Likewise.
22019 (rs6000_emit_le_vsx_store): Likewise.
22020 (rs6000_emit_move_si_sf_subreg): Likewise.
22021 (rs6000_emit_move): Likewise.
22022 (register_to_reg_type): Likewise.
22023 (rs6000_pre_atomic_barrier): Likewise.
22024 (rs6000_machopic_legitimize_pic_address): Likewise.
22025 (rs6000_allocate_stack_temp): Likewise.
22026 (rs6000_address_for_fpconvert): Likewise.
22027 (rs6000_address_for_altivec): Likewise.
22028 (rs6000_secondary_memory_needed_rtx): Delete function.
22029 (rs6000_check_sdmode): Likewise.
22030 (rs6000_alloc_sdmode_stack_slot): Likewise.
22031 (rs6000_instantiate_decls): Likewise.
22032 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22033 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22034 Delete reload_in_progress.
22035 (*vec_reload_and_plus_<mptrsize>): Likewise.
22036 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22037 (vsx_div_v2di): Likewise.
22038 (vsx_udiv_v2di): Likewise.
22040 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22042 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22043 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22044 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22045 (rs6000_debug_reg_global): Delete print of LRA status.
22046 (rs6000_option_override_internal): Delete dead LRA related code.
22047 (rs6000_lra_p): Delete function.
22048 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22050 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22052 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22053 * config/riscv/rtems.h: New file.
22055 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22056 Sudakshina Das <sudi.das@arm.com>
22058 * config/aarch64/aarch64.md
22059 (define_split for and<mode>3nr_compare): Move
22060 non aarch64_logical_operand to a register.
22061 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22062 register immediate operand to a register.
22063 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22065 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22067 PR middle-end/81564
22068 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22070 2017-07-27 Richard Biener <rguenther@suse.de>
22072 PR tree-optimization/81573
22073 PR tree-optimization/81494
22074 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22075 multi defuse cycle case.
22077 2017-07-27 Richard Biener <rguenther@suse.de>
22079 PR tree-optimization/81571
22080 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22083 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22085 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22086 earlier and only if MASK_FPU is set. Adjust formatting.
22088 2017-07-27 Martin Liska <mliska@suse.cz>
22090 * opt-functions.awk: Add validation of value of Init.
22091 * optc-gen.awk: Pass new argument.
22093 2017-07-27 Martin Liska <mliska@suse.cz>
22095 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22096 Fix wrong condition.
22098 2017-07-27 Martin Liska <mliska@suse.cz>
22100 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22101 BBs and edges seen by autoFDO.
22103 2017-07-27 Richard Biener <rguenther@suse.de>
22105 PR tree-optimization/81502
22106 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22107 with incompatible but same sized type.
22108 (execute_update_addresses_taken): Likewise.
22110 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22112 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22113 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22115 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22118 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22119 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22120 Change s_operand to memory_operand.
22122 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22124 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22125 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22126 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22127 Emit instructions rather than returning an expression. Handle TFmode
22128 and KFmode by casting to TImode.
22129 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22130 (rs6000_emit_le_vsx_store): Likewise.
22131 * config/rs6000/vsx.md (VSX_TI): New iterator.
22132 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22133 (*vsx_le_undo_permute_<mode>): Likewise.
22134 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22135 emit the split sequence.
22136 (*vsx_le_perm_store_<mode>): Likewise.
22138 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22140 PR tree-optimization/81555
22141 PR tree-optimization/81556
22142 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22143 if true, force CHANGED for the recursive invocation.
22144 (reassociate_bb): Remember original length of ops array, pass
22145 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22147 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22148 attributes for noipa attribute. For naked attribute use
22149 lookup_attribute first before lookup_attribute_spec.
22150 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22152 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22153 for functions with noipa attribute.
22154 (cgraph_externally_visible_p): Return true for functions with noipa
22156 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22157 for functions with noipa attribute.
22158 * doc/extend.texi: Document noipa function attribute.
22159 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22160 also for functions with noipa attribute.
22161 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22163 2017-07-26 Andrew Pinski <apinski@cavium.com>
22165 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22166 vec_unalign_load_cost and vec_unalign_store_cost.
22168 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22170 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22171 -mvsx-small-integer option.
22172 (ISA_3_0_MASKS_IEEE): Likewise.
22173 (OTHER_VSX_VECTOR_MASKS): Likewise.
22174 (POWERPC_MASKS): Likewise.
22175 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22176 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22177 code, only testing for DImode being allowed in non-VSX floating
22179 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22180 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22182 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22183 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22184 TARGET_P8_VECTOR test.
22185 (rs6000_secondary_reload_simple_move): Likewise.
22186 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22187 since TARGET_P9_VECTOR was already tested.
22188 (rs6000_opt_masks): Remove -mvsx-small-integer.
22189 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22190 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22192 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22193 test for TARGET_VEXTRACTUB was used, and that uses
22195 (p9 extract splitter): Likewise.
22196 (vsx_extract_<mode>_di_p9): Likewise.
22197 (vsx_extract_<mode>_store_p9): Likewise.
22198 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22199 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22200 the elimination of TARGET_VSX_SMALL_INTEGER.
22201 (vsx_extract_<mode>_p8): Likewise.
22202 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22203 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22204 (vsx_set_<mode>_p9): Likewise.
22205 (vsx_set_v4sf_p9): Likewise.
22206 (vsx_set_v4sf_p9_zero): Likewise.
22207 (vsx_insert_extract_v4sf_p9): Likewise.
22208 (vsx_insert_extract_v4sf_p9_2): Likewise.
22209 * config/rs6000/rs6000.md (sign extend splitter): Change
22210 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22211 (floatsi<mode>2_lfiwax_mem): Likewise.
22212 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22213 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22214 since a test for TARGET_P9_VECTOR was used.
22215 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22216 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22217 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22218 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22219 TARGET_P8_VECTOR test.
22220 (fix_trunc<mode>si2_stfiwx): Likewise.
22221 (fix_trunc<mode>si2_internal): Likewise.
22222 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22223 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22225 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22226 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22227 TARGET_P8_VECTOR test.
22228 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22229 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22230 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22232 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22233 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22234 since a test for TARGET_P9_VECTOR was used.
22235 (splitter for loading small constants): Likewise.
22237 2017-07-26 Andrew Pinski <apinski@cavium.com>
22239 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22242 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22245 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22246 (fp_valid_at): Likewise.
22248 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22250 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22251 (qdf24xx_addrcost_table): Likewise.
22252 (cortexa57_tunings): Update to use generic_branch_cost.
22253 (cortexa72_tunings): Likewise.
22254 (cortexa73_tunings): Likewise.
22255 (qdf24xx_tunings): Likewise.
22257 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22259 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22260 (thunderx2t99_branch_cost): Likewise.
22261 (cortexa35_tunings): Update to use generic_branch_cost.
22262 (cortexa53_tunings): Likewise.
22263 (cortexa57_tunings): Likewise.
22264 (cortexa72_tunings): Likewise.
22265 (cortexa73_tunings): Likewise.
22266 (thunderx2t99_tunings): Likewise.
22268 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22270 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22271 (sparc_option_override): Honour MASK_FSMULD.
22272 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22273 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22274 * config/sparc/sparc.opt (mfsmuld): New option.
22275 * doc/invoke.texi (mfsmuld): Document option.
22277 2017-07-26 Marek Polacek <polacek@redhat.com>
22279 PR middle-end/70992
22280 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22282 2017-07-26 Richard Biener <rguenther@suse.de>
22284 * gimple-match-head.c (do_valueize): Return OP if valueize
22286 (get_def): New helper to get at the def stmt of a SSA name
22287 if valueize allows.
22288 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22289 do_valueize to get at the def stmt.
22290 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22292 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22294 PR middle-end/46932
22295 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22297 2017-07-26 Martin Liska <mliska@suse.cz>
22300 * function.c (expand_function_start): Make expansion of
22301 nonlocal_goto_save_area after parm_birth_insn.
22303 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22305 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22306 from all CPU target flags enable members.
22308 2017-07-26 Richard Biener <rguenther@suse.de>
22310 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22311 (decision_tree::gen): Make 'type' const.
22312 (write_predicate): Likewise.
22314 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22316 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22317 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22318 (rs6000_option_override_internal): Likewise.
22319 (rs6000_expand_vector_set): Likewise.
22320 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22321 (TARGET_UPPER_REGS_SF): Likewise.
22322 (TARGET_UPPER_REGS_DI): Likewise.
22323 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22324 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22325 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22326 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22327 (Splitters for DI constants in Altivec registers): Likewise.
22328 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22329 (vsx_set_v4sf_p9): Likewise.
22330 (vsx_set_v4sf_p9_zero): Likewise.
22331 (vsx_insert_extract_v4sf_p9): Likewise.
22332 (vsx_insert_extract_v4sf_p9_2): Likewise.
22334 2017-07-25 Carl Love <cel@us.ibm.com>
22336 * doc/extend.texi: Update the built-in documentation file for the
22337 existing built-in functions
22338 vector signed char vec_cnttz (vector signed char);
22339 vector unsigned char vec_cnttz (vector unsigned char);
22340 vector signed short vec_cnttz (vector signed short);
22341 vector unsigned short vec_cnttz (vector unsigned short);
22342 vector signed int vec_cnttz (vector signed int);
22343 vector unsigned int vec_cnttz (vector unsigned int);
22344 vector signed long long vec_cnttz (vector signed long long);
22345 vector unsigned long long vec_cnttz (vector unsigned long long);
22347 2017-07-25 Andrew Pinski <apinski@cavium.com>
22349 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22350 accesses where the use is for the first operand of a BIT_INSERT.
22352 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22355 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22356 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22358 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22360 * config/i386/gstabs.h: Delete.
22361 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22363 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22365 * config/i386/i386.c (ix86_decompose_address): Do not check for
22366 register RTX when looking at index_reg or base_reg.
22367 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22369 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22371 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22372 to update EH info here.
22374 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22376 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22378 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22380 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22382 2017-07-25 Torsten Duwe <duwe@suse.de>
22384 * common.opt: Introduce -fpatchable-function-entry
22385 command line option, and its variables function_entry_patch_area_size
22386 and function_entry_patch_area_start.
22387 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22388 including a two-value parser.
22389 * target.def (print_patchable_function_entry): New target hook.
22390 * targhooks.h (default_print_patchable_function_entry): New function.
22391 * targhooks.c (default_print_patchable_function_entry): Likewise.
22392 * toplev.c (process_options): Switch off IPA-RA if
22393 patchable function entries are being generated.
22394 * varasm.c (assemble_start_function): Look at the
22395 patchable-function-entry command line switch and current
22396 function attributes and maybe generate NOP instructions by
22397 calling the print_patchable_function_entry hook.
22398 * doc/extend.texi: Document patchable_function_entry attribute.
22399 * doc/invoke.texi: Document -fpatchable_function_entry
22400 command line option.
22401 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22403 * doc/tm.texi: Re-generate.
22405 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22408 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22409 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22411 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22413 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22415 2017-07-25 Richard Biener <rguenther@suse.de>
22417 PR tree-optimization/81455
22418 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22419 not walk in cycles when looking for guards.
22421 2017-07-25 Richard Biener <rguenther@suse.de>
22423 PR tree-optimization/81529
22424 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22425 when optimizing backedge uses.
22427 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22429 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22431 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22432 to dl_section_ref. On AIX, append an expression to subtract
22433 the size of the section length to dl_section_ref.
22435 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22437 * configure.ac: If any of the config.* scripts fail, exit 1.
22438 * configure: Regenerate.
22440 2017-07-25 Richard Biener <rguenther@suse.de>
22442 PR middle-end/81546
22443 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22444 of immediate uses, be more verbose on errors.
22446 2017-07-25 Richard Biener <rguenther@suse.de>
22448 PR tree-optimization/81510
22449 * tree-vect-loop.c (vect_is_simple_reduction): When the
22450 reduction stmt is not inside the loop bail out.
22452 2017-07-25 Richard Biener <rguenther@suse.de>
22454 PR tree-optimization/81303
22455 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22456 profitability check against LOOP_VINFO_NITERSM1.
22458 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22460 * domwalk.c (cmp_bb_postorder): Simplify.
22461 (sort_bbs_postorder): New function. Use it...
22462 (dom_walker::walk): ...here to optimize common cases.
22464 2017-07-25 Martin Liska <mliska@suse.cz>
22467 * ipa-visibility.c (function_and_variable_visibility): Make the
22468 redirection just on target that supports aliasing.
22469 Fix GNU coding style.
22471 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22474 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22476 * config/arm/rtems.h: Likewise.
22477 * config/bfin/rtems.h: Likewise.
22478 * config/i386/rtemself.h: Likewise.
22479 * config/lm32/rtems.h: Likewise.
22480 * config/m32c/rtems.h: Likewise.
22481 * config/m68k/rtemself.h: Likewise.
22482 * config/microblaze/rtems.h: Likewise.
22483 * config/mips/rtems.h: Likewise.
22484 * config/moxie/rtems.h: Likewise.
22485 * config/nios2/rtems.h: Likewise.
22486 * config/powerpcspe/rtems.h: Likewise.
22487 * config/rs6000/rtems.h: Likewise.
22488 * config/rtems.h: Likewise.
22489 * config/sh/rtems.h: Likewise.
22490 * config/sh/rtemself.h: Likewise.
22491 * config/sparc/rtemself.h: Likewise.
22493 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22496 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22497 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22498 * tree-ssa-structalias.c (alias_get_name): Same.
22500 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22503 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22504 instructions if no du chain is found.
22506 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22508 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22510 2017-07-25 Richard Biener <rguenther@suse.de>
22512 PR middle-end/81505
22513 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22516 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22518 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22519 upper-regs options.
22520 (ISA_2_7_MASKS_SERVER): Likewise.
22521 (ISA_3_0_MASKS_IEEE): Likewise.
22522 (OTHER_P8_VECTOR_MASKS): Likewise.
22523 (OTHER_VSX_VECTOR_MASKS): Likewise.
22524 (POWERPC_MASKS): Likewise.
22525 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22526 duplicate list of options.
22527 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22528 explicit -mupper-regs options.
22529 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22530 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22531 alias for -mupper-regs-df.
22532 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22533 (rs6000_init_hard_regno_mode_ok): Likewise.
22534 (rs6000_option_override_internal): Likewise.
22535 (rs6000_opt_masks): Likewise.
22536 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22537 options in terms of whether -mvsx or -mpower8-vector was used.
22538 (TARGET_UPPER_REGS_DI): Likewise.
22539 (TARGET_UPPER_REGS_SF): Likewise.
22540 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22541 -mupper-regs-* options.
22543 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22545 * passes.c (emergency_dump_function): Print some empty lines and a
22546 header before the RTL dump.
22548 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22550 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22552 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22555 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22556 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22558 2017-07-24 Carl Love <cel@us.ibm.com>
22560 * config/rs6000/rs6000-c.c: Add support for built-in functions
22561 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22562 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22563 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22564 vec_extract_fp_from_shortl): Add defines for the two builtins.
22565 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22566 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22568 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22569 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22570 * doc/extend.texi: Update the built-in documentation file for the
22571 new built-in function.
22573 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22576 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22578 * doc/generic.texi: Likewise.
22579 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22580 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22582 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22584 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22585 (aarch64_mls_elt_merge<mode>): Likewise.
22587 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22589 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22590 having __cxa_atexit.
22592 2017-07-23 Michael Collison <michael.collison@arm.com>
22594 * config/arm/arm.c (arm_option_override): Deprecate
22595 use of -mstructure-size-boundary.
22596 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22597 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22599 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22602 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22603 Reduce cost estimate for direct moves.
22605 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22608 * config/i386/i386.c (ix86_option_override_internal): Disable
22609 BMI, BMI2 and TBM instructions for -m16.
22611 2017-07-21 Carl Love <cel@us.ibm.com>
22613 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22614 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22615 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22616 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22617 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22618 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22619 VMULOSW): New enum "unspec" values.
22620 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22621 altivec_vmulosw): New patterns.
22622 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22623 VMULOSW): Add definitions.
22625 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22627 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22628 (qdf24xx): Likewise.
22629 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22630 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22631 (AARCH64_FL_V8_1): Renumber.
22632 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22633 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22634 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22635 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22636 rdma to feature modifiers list.
22638 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22640 PR middle-end/56727
22641 * ipa-visibility (function_and_variable_visibility): Convert
22642 recursive PLT call to direct call if appropriate.
22644 2017-07-21 Andrew Pinski <apinski@cavium.com>
22646 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22647 operand 1 to see if the types precision matches.
22648 * fold-const.c (operand_equal_p): Likewise.
22650 2017-07-21 Richard Biener <rguenther@suse.de>
22652 PR tree-optimization/81303
22653 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22654 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22655 (vect_peeling_hash_get_lowest_cost): Adjust.
22656 (vect_enhance_data_refs_alignment): Likewise. Use
22657 vect_get_peeling_costs_all_drs to compute the penalty for no
22658 peeling to match up costs.
22660 2017-07-21 Richard Biener <rguenther@suse.de>
22662 PR tree-optimization/81500
22663 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22664 we didn't identify a reduction path.
22666 2017-07-21 Tom de Vries <tom@codesourcery.com>
22667 Cesar Philippidis <cesar@codesourcery.com>
22669 PR gcov-profile/81442
22670 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22673 2017-07-21 Tom de Vries <tom@codesourcery.com>
22676 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22678 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22679 nvptx_override_options_after_change.
22681 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22683 * dwarf2out.c (output_file_names): Avoid double testing for
22684 dwarf_version >= 5.
22686 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22688 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22690 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22692 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22694 (try_crossjump_to_edge): Do not punt on partitioned functions.
22696 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22698 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22699 Put all BBs reachable only via paths crossing cold region to cold
22701 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22703 2016-07-21 Richard Biener <rguenther@suse.de>
22705 PR tree-optimization/81303
22706 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22707 into account prologue and epilogue iterations when raising
22708 min_profitable_iters to sth at least covering one vector iteration.
22710 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22712 * config/arm/arm.c (arm_test_cpu_arch_dat):
22715 2017-07-20 Nathan Sidwell <nathan@acm.org>
22717 Remove TYPE_METHODS.
22718 * tree.h (TYPE_METHODS): Delete.
22719 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22720 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22721 (dbxout_type_methods): Scan TYPE_FIELDS.
22722 (dbxout_type): Don't check TYPE_METHODS here.
22723 * function.c (use_register_for_decl): Always ignore register for
22724 class types when not optimizing.
22725 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22726 * tree.c (free_lang_data_in_type): Stitch out member functions and
22727 templates from TYPE_FIELDS.
22728 (build_distinct_type_copy, verify_type_variant,
22729 verify_type): Member fns are on TYPE_FIELDS.
22730 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22731 * tree-pretty-print.c (dump_generic_node): Likewise.
22733 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22736 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22737 V2TImode and V4TImode.
22738 (ix86_expand_vector_extract): Likewise.
22739 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22740 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22741 (ssescalarmode): Handle V4TImode and V2TImode.
22742 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22743 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22744 (VEXTRACTI128_MODE): New mode iterator.
22745 (splitter for *vec_extractv?ti first element): New.
22746 (VEC_INIT_MODE): New mode iterator.
22747 (vec_init<mode>): Consolidate 3 expanders into one using
22748 VEC_INIT_MODE mode iterator.
22750 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22752 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22753 non_spilled_static_chain_regno_p.
22755 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22757 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22759 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22761 * bb-reorder.c (connect_traces): Allow copying of blocks within
22764 2017-07-20 Richard Biener <rguenther@suse.de>
22766 * gimple.h (gimple_phi_result): Add gphi * overload.
22767 (gimple_phi_result_ptr): Likewise.
22768 (gimple_phi_arg): Likewise. Adjust index assert to only
22769 allow actual argument accesses rather than all slots available
22771 (gimple_phi_arg_def): Add gphi * overload.
22772 * tree-phinodes.c (make_phi_node): Initialize only actual
22774 (resize_phi_node): Clear memory not covered by old node,
22775 do not initialize excess argument slots.
22776 (reserve_phi_args_for_new_edge): Initialize new argument slot
22779 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22781 PR tree-optimization/81388
22783 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22785 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22786 by removing computation of may_be_zero.
22788 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22789 Tom de Vries <tom@codesourcery.com>
22791 PR middle-end/81030
22792 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22793 when gimple level profile disagrees with what RTL expander did.
22795 2017-07-20 Richard Biener <rguenther@suse.de>
22797 PR tree-optimization/61171
22798 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22799 (vect_analyze_stmt): Add slp instance parameter.
22800 (vectorizable_reduction): Likewise.
22801 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22802 (vect_is_simple_reduction): Deal with chains not detected
22803 as SLP reduction chain, specifically not properly associated
22804 chains containing a mix of plus/minus.
22805 (get_reduction_op): Remove.
22806 (get_initial_defs_for_reduction): Simplify, pass in whether
22807 this is a reduction chain, pass in the SLP node for the PHIs.
22808 (vect_create_epilog_for_reduction): Get the SLP instance as
22810 (vectorizable_reduction): Get the SLP instance as arg.
22811 During analysis remember the SLP node with the PHIs in the
22812 instance. Simplify getting at the vectorized reduction PHIs.
22813 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22814 through SLP instance.
22815 (vect_slp_analyze_operations): Likewise.
22816 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22817 (vect_transform_stmt): Likewise.
22819 2017-07-20 Tom de Vries <tom@codesourcery.com>
22821 PR tree-optimization/81489
22822 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22823 read of phi arg location to before loop that modifies phi.
22825 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22827 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22830 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22832 PR middle-end/81331
22833 * except.c (execute): Fix ordering issue.
22835 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22837 PR rtl-optimization/81423
22838 * combine.c (make_compound_operation_int): Don't try to optimize
22839 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22841 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22843 PR rtl-optimization/81423
22844 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22845 with a constant that is -1 in the truncated to mode.
22847 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22849 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22850 (determine_unlikely_bbs): ... here.
22851 * predict.h (propagate_unlikely_bbs_forward): Declare.
22852 * cfgexpand.c (pass_expand::execute): Use it.
22853 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22855 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22856 propagate_unlikely_bbs_forward.
22858 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22860 PR middle-end/81331
22861 * except.c (maybe_add_nop_after_section_switch): New function.
22864 2017-07-19 Tom de Vries <tom@codesourcery.com>
22866 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22868 2017-07-19 Tom de Vries <tom@codesourcery.com>
22870 * gimple.h (gimple_phi_arg): Make assert more strict.
22872 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22874 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22875 * config/rs6000/mmintrin.h: New file.
22876 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22878 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22880 PR tree-optimization/81346
22881 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22883 2017-07-19 Tom de Vries <tom@codesourcery.com>
22885 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22887 2017-07-19 Tom de Vries <tom@codesourcery.com>
22889 * config/nvptx/nvptx-modes.def: Add V2DImode.
22890 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22891 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22892 (nvptx_output_mov_insn): Handle lack of mov.b128.
22893 (nvptx_print_operand): Handle 'H' and 'L' codes.
22894 (nvptx_vector_mode_supported): Allow V2DImode.
22895 (nvptx_preferred_simd_mode): New function.
22896 (nvptx_data_alignment): New function.
22897 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22898 nvptx_preferred_simd_mode.
22899 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22901 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22903 2017-07-19 Tom de Vries <tom@codesourcery.com>
22905 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22906 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22907 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22908 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22909 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22910 (mov<VECIM>_insn): New define_insn.
22911 (define_expand "mov<VECIM>): New define_expand.
22913 2017-07-19 Tom de Vries <tom@codesourcery.com>
22915 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22917 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22919 PR tree-optimization/81346
22920 * fold-const.h (fold_div_compare, range_check_type): Declare.
22921 * fold-const.c (range_check_type): New function.
22922 (build_range_check): Use range_check_type.
22923 (fold_div_compare): No longer static, rewritten into
22924 a match.pd helper function.
22925 (fold_comparison): Don't call fold_div_compare here.
22926 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22927 as helper function.
22929 2017-07-19 Nathan Sidwell <nathan@acm.org>
22931 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22932 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22933 * tree.c (find_decls_types_r, verify_type): Use
22934 TYPE_{MIN,MAX}_VALUE_RAW.
22935 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22936 (hash_tree): Likewise.
22937 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22939 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22942 2017-07-18 Tom de Vries <tom@codesourcery.com>
22944 PR middle-end/81464
22945 * omp-expand.c (expand_omp_for_static_chunk): Handle
22946 equal-argument loop exit phi.
22948 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22951 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22952 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22953 operand 2 predicate.
22954 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22955 operand 2 predicate.
22956 (ror,rol -> rorx splitters): Use const_int_operand as
22957 operand 2 predicate.
22959 2017-06-18 Richard Biener <rguenther@suse.de>
22961 PR tree-optimization/81410
22962 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22963 the gap in the ! slp_perm SLP case after each group.
22965 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22967 PR middle-end/81463
22968 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22971 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22973 PR middle-end/81462
22974 * predict.c (set_even_probabilities): Cleanup; do not affect
22975 probabilities that are already known.
22976 (combine_predictions_for_bb): Call even when count is set.
22978 2017-07-18 Nathan Sidwell <nathan@acm.org>
22980 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22983 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22986 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22987 optimization for loop niter analysis.
22989 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22992 * config/avr/avr.c (avr_optimize_casesi): Don't use
22993 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22995 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22997 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22998 body_cost_vec from _vect_peel_extended_info.
22999 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23000 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23003 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23005 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23007 2017-07-18 Richard Biener <rguenther@suse.de>
23009 PR tree-optimization/80620
23010 PR tree-optimization/81403
23011 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23012 info when re-using a VN table entry.
23014 2017-07-18 Richard Biener <rguenther@suse.de>
23016 PR tree-optimization/81418
23017 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23018 vectype_in. Verify that with lane-reducing reduction operations
23019 we have a single def-use cycle.
23021 2017-07-17 Carl Love <cel@us.ibm.com>
23023 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23025 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23026 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23027 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23028 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23029 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23030 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23031 VMULOSW): New enum "unspec" values.
23032 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23033 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23034 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23035 altivec_vmulosw): New patterns.
23036 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23037 VMULOSW): Add definitions.
23039 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23041 * config/alpha/alpha.c: Include predict.h.
23043 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23045 * tree-vrp.c (compare_assert_loc): Fix comparison function
23046 to return predictable results.
23048 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23050 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23052 (subdi3): Likewise.
23053 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23054 * doc/invoke.texi (mexpand-adddi): Update text.
23056 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23058 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23059 that also clobbers the CC register. The old expand code is moved
23061 (*arc_clzsi2): ... here.
23062 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23063 the CC register. The old expand code is moved to ...
23064 (arc_ctzsi2): ... here.
23066 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23068 * config/arc/arc.opt (mindexed-loads): Use initial value
23069 TARGET_INDEXED_LOADS_DEFAULT.
23070 (mauto-modify-reg): Use initial value
23071 TARGET_AUTO_MODIFY_REG_DEFAULT.
23072 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23073 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23074 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23075 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23077 2017-07-17 Martin Liska <mliska@suse.cz>
23080 * opts.c (finish_options): Do not allow -fgnu-tm
23081 w/ -fsanitize={kernel-,}address. Say sorry.
23083 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23086 * tree-loop-distribution.c (classify_partition): Only assert on
23087 numer of iterations.
23088 (merge_dep_scc_partitions): Delete prameter. Update function call.
23089 (distribute_loop): Remove code handling loop with unknown niters.
23090 (pass_loop_distribution::execute): Skip loop with unknown niters.
23092 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23095 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23096 function sort_partitions_by_post_order.
23098 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23100 PR tree-optimization/81374
23101 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23102 the max index of basic blocks, rather than number of basic blocks.
23104 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23106 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23108 (arc_legitimate_pic_operand_p): Likewise.
23109 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23111 (arc_needs_pcl_p): Likewise.
23112 (arc_legitimate_pc_offset_p): Likewise.
23113 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23114 function is also used in constrains.md.
23115 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23116 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23117 PLUS. Only return true/false in known cases, otherwise assert.
23118 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23119 is already called in arc_legitimate_constant_p.
23120 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23122 (LEGITIMATE_PIC_OPERAND_P): Use
23123 arc_raw_symbolic_reference_mentioned_p function.
23124 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23129 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23130 Andrew Burgess <andrew.burgess@embecosm.com>
23132 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23133 (arc_return_address_register): New function.
23134 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23135 (arc_handle_fndecl_attribute): Add naked attribute.
23136 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23137 (TARGET_WARN_FUNC_RETURN): Likewise.
23138 (arc_allocate_stack_slots_for_args): New function.
23139 (arc_warn_func_return): Likewise.
23140 (machine_function): Change type fn_type.
23141 (arc_compute_function_type): Consider new naked function type,
23142 change function return type.
23143 (arc_must_save_register): Adapt to handle new
23144 arc_compute_function_type's return type.
23145 (arc_expand_prologue): Likewise.
23146 (arc_expand_epilogue): Likewise.
23147 (arc_return_address_regs): Delete.
23148 (arc_return_address_register): New function.
23149 (arc_epilogue_uses): Use above function.
23150 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23151 (arc_function_type): Change encoding, add naked type.
23152 (ARC_INTERRUPT_P): Change to handle the new encoding.
23153 (ARC_FAST_INTERRUPT_P): Likewise.
23154 (ARC_NORMAL_P): Define.
23155 (ARC_NAKED_P): Likewise.
23156 (arc_compute_function_type): Delete prototype.
23157 * config/arc/arc.md (in_ret_delay_slot): Use
23158 arc_return_address_register function.
23159 (simple_return): Likewise.
23160 (p_return_i): Likewise.
23162 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23164 PR tree-optimization/81428
23165 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23166 can't be built for those types.
23168 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23170 Remove stuff dead since r239246.
23172 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23173 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23174 (avr_inform_devices): Remove dead stuff.
23176 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23178 * config/arm/arm_neon.h: Fix softp typo.
23180 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23182 PR tree-optimization/81365
23183 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23184 aggregate moves onto bb predecessor edges, make sure there are no
23185 loads that could alias the lhs in between the start of bb and the
23188 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23191 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23192 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23193 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23195 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23197 PR tree-optimization/81396
23198 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23199 (init_symbolic_number): Initialize it to 1.
23200 (perform_symbolic_merge): Add n_ops from both operands into the new
23202 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23203 without base_addr as useless if they need more than one operation.
23204 (bswap_replace): Handle !bswap case for NULL base_addr.
23206 2017-07-17 Tom de Vries <tom@codesourcery.com>
23209 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23212 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23214 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23215 conditional builtin define __FIX_LEON3FT_B2BST.
23217 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23219 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23220 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23223 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23225 PR rtl-optimization/81424
23226 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23227 to remove potential trapping from operands if -fnon-call-exceptions.
23229 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23231 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23232 profile_proability for scalling.
23233 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23235 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23237 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23239 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23241 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23242 fixpoint arithmetics.
23244 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23246 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23247 fixpoint arithmetics.
23249 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23251 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23252 fixpoint arithmetics.
23254 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23256 * profile-count.h (profile_probability::from_reg_br_prob_note,
23257 profile_probability::to_reg_br_prob_note): New functions.
23258 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23259 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23260 * predict.c (probability_reliable_p): Update.
23261 (edge_probability_reliable_p): Update.
23262 (br_prob_note_reliable_p): Update.
23263 (invert_br_probabilities): Update.
23264 (add_reg_br_prob_note): New function.
23265 (combine_predictions_for_insn): Update.
23266 * asan.c (asan_clear_shadow): Update.
23267 * cfgbuild.c (compute_outgoing_frequencies): Update.
23268 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23269 (update_br_prob_note): Update.
23270 (rtl_verify_edges): Update.
23271 (purge_dead_edges): Update.
23272 (fixup_reorder_chain): Update.
23273 * emit-rtl.c (try_split): Update.
23274 * ifcvt.c (cond_exec_process_insns): Update.
23275 (cond_exec_process_if_block): Update.
23276 (dead_or_predicable): Update.
23277 * internal-fn.c (expand_addsub_overflow): Update.
23278 (expand_neg_overflow): Update.
23279 (expand_mul_overflow): Update.
23280 * loop-doloop.c (doloop_modify): Update.
23281 * loop-unroll.c (compare_and_jump_seq): Update.
23282 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23283 * predict.h: Update.
23284 * reorg.c (mostly_true_jump): Update.
23286 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23287 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23288 * config/arc/arc.c: (emit_unlikely_jump): Update.
23289 * config/arm/arm.c: (emit_unlikely_jump): Update.
23290 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23291 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23292 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23293 (ix86_print_operand): Update.
23294 (ix86_split_fp_branch): Update.
23295 (predict_jump): Update.
23296 * config/ia64/ia64.c (ia64_print_operand): Update.
23297 * config/mmix/mmix.c (mmix_print_operand): Update.
23298 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23299 (rs6000_expand_split_stack_prologue): Update.
23300 * config/rs6000/rs6000.c: Update.
23301 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23302 (s390_expand_vec_movstr): Update.
23303 (s390_expand_cs_tdsi): Update.
23304 (s390_expand_split_stack_prologue): Update.
23305 * config/sh/sh.c (sh_print_operand): Update.
23306 (expand_cbranchsi4): Update.
23307 (expand_cbranchdi4): Update.
23308 * config/sparc/sparc.c (output_v9branch): Update.
23309 * config/spu/spu.c (get_branch_target): Update.
23310 (ea_load_store_inline): Update.
23311 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23312 * config/tilepro/tilepro.c: Update.
23314 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23316 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23317 (gimplify_save_expr): Likewise.
23319 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23321 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23323 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23325 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23327 * timevar.def (TV_IPA_FNSUMMARY): Define.
23329 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23331 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23332 to back store errata sensitive sequence from being generated.
23333 (sqrtdf2_fix): Likewise.
23335 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23337 * tree-ssa-threadupdate.c (compute_path_counts,
23338 update_joiner_offpath_counts): Use profile_probability.
23340 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23343 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23345 * config/arm/arm-c.c (arm_cpu_builtins): Define
23346 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23348 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23350 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23351 array entries to represent __ieee128 versions of the
23352 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23353 scalar_extract_sig, and scalar_insert_exp built-in functions.
23354 (altivec_resolve_overloaded_builtin): Add special case handling
23355 for the __builtin_scalar_insert_exp function, as represented by
23356 the P9V_BUILTIN_VEC_VSIEDP constant.
23357 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23358 exponent support for __ieee128 argument.
23359 (VSESQP): Add scalar extract signature support for __ieee128
23361 (VSTDCNQP): Add scalar test negative support for __ieee128
23363 (VSIEQP): Add scalar insert exponent support for __int128 argument
23364 with __ieee128 result.
23365 (VSIEQPF): Add scalar insert exponent support for __ieee128
23366 argument with __ieee128 result.
23367 (VSTDCQP): Add scalar test data class support for __ieee128
23369 (VSTDCNQP): Add overload support for scalar test negative with
23370 __ieee128 argument.
23371 (VSTDCQP): Add overload support for scalar test data class
23372 __ieee128 argument.
23373 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23374 UNSPEC_VSX_SXSIGDP.
23375 (UNSPEC_VSX_SIEXPQP): New constant.
23376 (xsxexpqp): New insn for VSX scalar extract exponent quad
23378 (xsxsigqp): New insn for VSX scalar extract significand quad
23380 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23381 precision with floating point argument.
23382 (xststdcqp): New expand for VSX scalar test data class quad
23384 (xststdcnegqp): New expand for VSX scalar test negative quad
23386 (xststdcqp): New insn to match expansions for VSX scalar test data
23387 class quad precision and VSX scalar test negative quad precision.
23388 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23389 special case operand checking to enforce that second operand of
23390 VSX scalar test data class with quad precision argument is a 7-bit
23392 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23393 prototypes and descriptions of __ieee128 versions of
23394 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23395 scalar_test_data_class, and scalar_test_neg built-in functions.
23397 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23399 PR tree-optimization/81162
23400 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23401 replace a negate with an add.
23403 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23405 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23407 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23409 * config/arm/arm-c.c (arm_cpu_builtins): Define
23410 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23412 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23414 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23415 (armv8-r): Set ARM Cortex-R52 as default CPU.
23416 * config/arm/arm-tables.opt: Regenerate.
23417 * config/arm/arm-tune.md: Regenerate.
23418 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23420 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23421 extension for -mcpu=cortex-r52.
23423 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23425 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23426 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23427 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23428 (fp-armv8): Define it as FP_ARMv8 only.
23429 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23430 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23432 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23433 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23434 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23435 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23436 than TARGET_FPU_ARMV8.
23437 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23438 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23439 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23441 * config/arm/neon.md (neon_vrint): Likewise.
23442 (neon_vcvt): Likewise.
23443 (neon_<fmaxmin_op><mode>): Likewise.
23444 (<fmaxmin><mode>3): Likewise.
23445 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23446 * config/arm/predicates.md (arm_cond_move_operator): Check against
23447 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23449 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23451 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23452 to top of function.
23454 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23456 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23457 loop in comment with memset.
23459 2017-07-14 Martin Liska <mliska@suse.cz>
23461 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23462 * dwarf2out.c (is_java): Remove the function.
23463 (output_pubname): Remove usage of the function.
23464 (lower_bound_default): Remove usage of DW_LANG_Java.
23465 (gen_compile_unit_die): Likewise.
23466 * gcc.c: Remove compiler defaults for .java and .zip files.
23467 * gimple-expr.c (remove_suffix): Change as there's no longer
23468 extension than 4-letter one.
23469 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23470 (gimplify_save_expr): Likewise.
23471 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23472 as it's possible even for other languages than Java.
23473 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23474 * lto-opts.c (lto_write_options): Remove reference to Java.
23475 * opts.c (strip_off_ending): Update file extension handling.
23476 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23477 * tree-eh.c (lower_resx): Likewise.
23478 * tree.c (free_lang_data_in_type): Remove dead code.
23479 (find_decls_types_r): Likewise.
23480 (build_common_builtin_nodes): Remove Java from a comment.
23481 (verify_type): Remove dead code.
23482 * varasm.c (assemble_external): Remove Java from a comment.
23484 2017-07-14 Martin Liska <mliska@suse.cz>
23486 * opts.c (finish_options): Add quotes.
23487 (common_handle_option): Likewise.
23489 2017-07-14 Martin Liska <mliska@suse.cz>
23491 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23492 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23493 Remove N_SO_PASCAL.
23494 * dwarf2out.c (lower_bound_default): Do not handle
23496 (gen_compile_unit_die): Likewise.
23497 * gcc.c: Remove default extension binding for GNU Pascal.
23498 * stmt.c: Remove Pascal language from a comment.
23499 * xcoffout.c: Likewise.
23501 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23504 * diagnostic-show-locus.c (fixit_cmp): New function.
23505 (layout::layout): Sort m_fixit_hints.
23506 (column_range::column_range): Assert that the values are valid.
23507 (struct char_span): New struct.
23508 (correction::overwrite): New method.
23509 (struct source_line): New struct.
23510 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23511 calls in terms of classes source_line and char_span, and
23512 correction::overwrite.
23513 (selftest::test_overlapped_fixit_printing_2): New function.
23514 (selftest::diagnostic_show_locus_c_tests): Call it.
23516 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23518 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23519 early if there is no lhs.
23521 2017-07-13 Martin Liska <mliska@suse.cz>
23523 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23524 (gen_reference_type_die): Likewise.
23525 * stor-layout.c: Remove Pascal-related comment.
23527 2017-07-13 Martin Liska <mliska@suse.cz>
23529 * opts.c (finish_options): Add quotes to error messages.
23530 (parse_sanitizer_options): Likewise.
23532 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23534 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23536 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23538 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23540 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23542 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23544 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23546 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23549 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23550 provides the hardware capability bits, define the macro
23551 __BUILTIN_CPU_SUPPORTS__.
23552 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23553 if GLIBC does not provide the hardware capability bits. Add a
23554 gcc_unreachable call if the built-in cpu function is neither
23555 __builtin_cpu_is nor __builtin_cpu_supports.
23556 (rs6000_get_function_versions_dispatcher): Change the warning
23557 that an old GLIBC is used which does not export the capability
23558 bits to be an error.
23559 * doc/extend.texi (target_clones attribute): Document the
23560 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23561 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23562 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23563 the macros defined by GCC if the newer GLIBC is available.
23565 2017-07-12 Jeff Law <law@redhat.com>
23567 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23568 remaining includes slightly.
23569 * config/riscv/riscv-builtins.c: Include profile-count.h.
23571 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23574 * config/avr/avr.c (avr_set_current_function): In diagnostic
23575 messages: Quote keywords and (parts of) identifiers.
23576 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23579 2017-07-12 Carl Love <cel@us.ibm.com>
23581 * config/rs6000/rs6000-c.c: Add support for built-in functions
23582 vector bool char vec_revb (vector bool char);
23583 vector bool short vec_revb (vector short char);
23584 vector bool int vec_revb (vector bool int);
23585 vector bool long long vec_revb (vector bool long long);
23586 * doc/extend.texi: Update the built-in documentation file for the
23587 new built-in functions.
23589 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23591 * config/s390/s390.md: Remove movcc splitter.
23593 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23595 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23596 load/store on condition.
23598 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23601 * config/avr/avr.c (avr_encode_section_info)
23602 [progmem && !TREE_READONLY]: Error if progmem object needs
23605 2017-07-11 Michael Collison <michael.collison@arm.com>
23607 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23610 2017-07-11 Carl Love <cel@us.ibm.com>
23612 * config/rs6000/rs6000-c.c: Add support for builtins
23613 vector unsigned int vec_parity_lsbb (vector signed int);
23614 vector unsigned int vec_parity_lsbb (vector unsigned int);
23615 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23616 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23617 vector unsigned long long vec_parity_lsbb (vector signed long long);
23618 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23619 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23620 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23621 * doc/extend.texi: Update the built-in documentation file for the
23622 new built-in functions.
23624 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23626 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23627 (layout::m_primary_loc): New field.
23628 (layout::layout): Initialize new field. Move location filtering
23629 logic from here to...
23630 (layout::maybe_add_location_range): ...this new method. Add
23631 support for filtering to just the lines already specified by other
23633 (layout::will_show_line_p): New method.
23634 (gcc_rich_location::add_location_if_nearby): New method.
23635 (selftest::test_add_location_if_nearby): New test function.
23636 (selftest::diagnostic_show_locus_c_tests): Call it.
23637 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23640 2017-07-11 Tom de Vries <tom@codesourcery.com>
23642 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23643 (bb_first_real_insn): New function.
23644 (nvptx_single): Add extra initialization of broadcasted condition
23647 2017-07-11 Nathan Sidwell <nathan@acm.org>
23649 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23651 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23653 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23654 Binutils doc as TEXI will mess them up.
23655 * doc/invoke.texi (AVR Options): Same here.
23657 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23659 * config/sparc/sparc.opt (mfix-ut700): New option.
23660 (mfix-gr712rc): Likewise.
23661 (sparc_fix_b2bst): New variable.
23662 * doc/invoke.texi (SPARC options): Document them.
23663 (ARM options): Fix warnings.
23664 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23665 instructions to prevent sequences that can trigger the store-store
23666 errata for certain LEON3FT processors.
23667 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23668 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23669 * config/sparc/sparc.md (fix_b2bst): New attribute.
23670 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23672 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23675 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23677 (*rsqrtsf2_sse): Ditto.
23679 (div<mode>3): Macroize insn from divdf3 and divsf3
23680 using MODEF mode iterator.
23682 2017-07-10 Martin Sebor <msebor@redhat.com>
23684 PR tree-optimization/80397
23685 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23686 instead of testing for equality to INTEGER_TYPE.
23688 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23690 * config.gcc: Remove uclibc from arc target spec.
23692 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23694 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23696 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23699 * lto-wrapper.c (remove_option): New function.
23700 (merge_and_complain): Merge PIC/PIE options more realistically.
23702 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23704 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23708 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23709 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23710 * config.in: Regenerate.
23711 * configure: Regenerate.
23712 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23713 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23714 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23715 (TARGET_GASISR_PROLOGUES): ...target mask.
23716 * common/config/avr/avr-common.c
23717 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23718 Set -mgas-isr-prologues.
23719 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23720 INSERT_PASS_BEFORE for it.
23721 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23722 * config/avr/avr.c (avr_option_override)
23723 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23724 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23725 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23726 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23727 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23729 (make_avr_pass_pre_proep): New function.
23730 (emit_push_sfr) <treg>: Add argument to function and use it
23731 instead of TMP_REG.
23732 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23733 and set machine->gasisr.yes.
23734 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23735 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23736 __gcc_isr.n_pushed to .L__stack_usage.
23737 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23738 (avr_asm_final_postscan_insn): ...this new static function.
23739 * config/avr/avr.h (machine_function)
23740 <is_no_gccisr, use_L__stack_usage>: New fields.
23741 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23742 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23743 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23744 (gasisr, *gasisr): New expander and insn.
23745 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23746 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23747 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23749 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23751 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23754 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23756 Move jump-tables out of .text again.
23759 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23760 (ASM_OUTPUT_ADDR_VEC): New function.
23761 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23762 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23763 INSN_ADDRESSes as asm comment.
23764 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23765 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23766 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23767 * config/avr/avr.md (*tablejump): Adjust comment.
23768 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23769 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23771 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23772 (avr_output_addr_vec): New proto.
23773 (avr_log_t) <insn_addresses>: New field.
23775 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23778 * config/i386/i386.c (ix86_function_arg_advance): Set
23779 outgoing_args_on_stack to true if there are outgoing arguments
23781 (ix86_function_arg): Likewise.
23782 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23783 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23784 * config/i386/i386.h (machine_function): Add
23785 outgoing_args_on_stack.
23787 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23789 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23790 supporting pthreds.
23791 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23793 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23795 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23796 (REAL_H): Remove $(MACHMODE_H).
23797 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23799 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23800 $(MACHMODE_H) and double-int.h.
23801 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23803 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23806 2017-07-07 Andrew Pinski <apinski@cavium.com>
23808 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23809 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23811 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23813 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23814 Add warning if GCC was not configured to link against a GLIBC that
23815 exports the hardware capability bits.
23816 (make_resolver_func): Make resolver function private and not a
23817 COMDAT function. Create the name with clone_function_name instead
23818 of make_unique_name.
23821 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23822 correct operand in doing the split.
23824 2017-07-07 Carl Love <cel@us.ibm.com>
23826 * config/rs6000/rs6000-c: Add support for built-in function
23827 vector unsigned short vec_pack_to_short_fp32 (vector float,
23829 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23830 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23831 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23832 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23833 (convert_4f32_8i16): Add define_expand.
23834 * doc/extend.texi: Update the built-in documentation file for the
23835 new built-in function.
23837 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23839 * config/sparc/m8.md: New file.
23840 * config/sparc/sparc.md: Include m8.md.
23842 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23844 * config/sparc/sparc.opt: New option -mvis4b.
23845 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23846 (sparc_option_override): Handle VIS4B.
23847 (enum sparc_builtins): Define
23848 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23849 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23850 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23851 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23852 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23853 (check_constant_argument): New function.
23854 (sparc_vis_init_builtins): Define builtins
23855 __builtin_vis_dictunpack{8,16,32},
23856 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23857 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23858 __builtin_vis_fpcmpde{8,16,32}shl and
23859 __builtin_vis_fpcmpur{8,16,32}shl.
23860 (sparc_expand_builtin): Check that the constant operands to
23861 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23862 constant and in range.
23863 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23865 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23866 (SPARC_IMM5_P): Likewise.
23867 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23868 (enabled): Handle vis4b.
23869 (UNSPEC_DICTUNPACK): New unspec.
23870 (UNSPEC_FPCMPSHL): Likewise.
23871 (UNSPEC_FPUCMPSHL): Likewise.
23872 (UNSPEC_FPCMPDESHL): Likewise.
23873 (UNSPEC_FPCMPURSHL): Likewise.
23874 (cpu_feature): New CPU feature `vis4b'.
23875 (dictunpack{8,16,32}): New insns.
23876 (FPCSMODE): New mode iterator.
23877 (fpcscond): New code iterator.
23878 (fpcsucond): Likewise.
23879 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23880 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23881 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23882 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23883 * config/sparc/constraints.md: Define constraints `q' for unsigned
23884 2-bit integer constants and `t' for unsigned 5-bit integer
23886 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23888 (imm5_operand_dictunpack16): Likewise.
23889 (imm5_operand_dictunpack32): Likewise.
23890 (imm2_operand): Likewise.
23891 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23892 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23893 ditunpack* and fpcmp*shl builtins.
23895 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23897 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23898 * config.in: Add HAVE_AS_SPARC6 define.
23899 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23901 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23903 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23904 (CPP_CPU_SPEC): Handle m8.
23905 (ASM_CPU_SPEC): Likewise.
23906 * config/sparc/sparc-opts.h (enum processor_type): Add
23908 * config/sparc/sparc.c (m8_costs): New struct.
23909 (sparc_option_override): Handle TARGET_CPU_m8.
23910 (sparc32_initialize_trampoline): Likewise.
23911 (sparc64_initialize_trampoline): Likewise.
23912 (sparc_issue_rate): Likewise.
23913 (sparc_register_move_cost): Likewise.
23914 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23915 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23916 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23917 (CPP_CPU_SPEC): Handle M8.
23918 (ASM_CPU_SPEC): Likewise.
23919 (AS_M8_FLAG): Define.
23920 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23921 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23922 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23924 * configure: Regenerate.
23925 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23928 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23930 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23932 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23933 ("*movdi_insn_sp32"): Do not set v3pipe.
23934 ("*movsi_insn"): Likewise.
23935 ("*movdi_insn_sp64"): Likewise.
23936 ("*movsf_insn"): Likewise.
23937 ("*movdf_insn_sp32"): Likewise.
23938 ("*movdf_insn_sp64"): Likewise.
23939 ("*zero_extendsidi2_insn_sp64"): Likewise.
23940 ("*sign_extendsidi2_insn"): Likewise.
23941 ("*mov<VM32:mode>_insn"): Likewise.
23942 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23943 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23944 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23945 ("<vlop:code><VL:mode>3"): Likewise.
23946 ("*not_<vlop:code><VL:mode>3"): Likewise.
23947 ("*nand<VL:mode>_vis"): Likewise.
23948 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23949 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23950 ("one_cmpl<VL:mode>2"): Likewise.
23951 ("faligndata<VM64:mode>_vis"): Likewise.
23952 ("alignaddrsi_vis"): Likewise.
23953 ("alignaddrdi_vis"): Likweise.
23954 ("alignaddrlsi_vis"): Likewise.
23955 ("alignaddrldi_vis"): Likewise.
23956 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23957 ("bmaskdi_vis"): Likewise.
23958 ("bmasksi_vis"): Likewise.
23959 ("bshuffle<VM64:mode>_vis"): Likewise.
23960 ("cmask8<P:mode>_vis"): Likewise.
23961 ("cmask16<P:mode>_vis"): Likewise.
23962 ("cmask32<P:mode>_vis"): Likewise.
23963 ("pdistn<P:mode>_vis"): Likewise.
23964 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23966 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23968 * config/sparc/sparc.md ("subtype"): New insn attribute.
23969 ("*wrgsr_sp64"): Set insn subtype.
23970 ("*rdgsr_sp64"): Likewise.
23971 ("alignaddrsi_vis"): Likewise.
23972 ("alignaddrdi_vis"): Likewise.
23973 ("alignaddrlsi_vis"): Likewise.
23974 ("alignaddrldi_vis"): Likewise.
23975 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23976 ("fexpand_vis"): Likewise.
23977 ("fpmerge_vis"): Likewise.
23978 ("faligndata<VM64:mode>_vis"): Likewise.
23979 ("bshuffle<VM64:mode>_vis"): Likewise.
23980 ("cmask8<P:mode>_vis"): Likewise.
23981 ("cmask16<P:mode>_vis"): Likewise.
23982 ("cmask32<P:mode>_vis"): Likewise.
23983 ("fchksm16_vis"): Likewise.
23984 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23985 ("fmean16_vis"): Likewise.
23986 ("fp<plusminus_insn>64_vis"): Likewise.
23987 ("<plusminus_insn>v8qi3"): Likewise.
23988 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23989 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23990 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23991 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23992 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23993 ("*movqi_insn"): Likewise.
23994 ("*movhi_insn"): Likewise.
23995 ("*movsi_insn"): Likewise.
23996 ("movsi_pic_gotdata_op"): Likewise.
23997 ("*movdi_insn_sp32"): Likewise.
23998 ("*movdi_insn_sp64"): Likewise.
23999 ("movdi_pic_gotdata_op"): Likewise.
24000 ("*movsf_insn"): Likewise.
24001 ("*movdf_insn_sp32"): Likewise.
24002 ("*movdf_insn_sp64"): Likewise.
24003 ("*zero_extendhisi2_insn"): Likewise.
24004 ("*zero_extendqihi2_insn"): Likewise.
24005 ("*zero_extendqisi2_insn"): Likewise.
24006 ("*zero_extendqidi2_insn"): Likewise.
24007 ("*zero_extendhidi2_insn"): Likewise.
24008 ("*zero_extendsidi2_insn_sp64"): Likewise.
24009 ("ldfsr"): Likewise.
24010 ("prefetch_64"): Likewise.
24011 ("prefetch_32"): Likewise.
24012 ("tie_ld32"): Likewise.
24013 ("tie_ld64"): Likewise.
24014 ("*tldo_ldub_sp32"): Likewise.
24015 ("*tldo_ldub1_sp32"): Likewise.
24016 ("*tldo_ldub2_sp32"): Likewise.
24017 ("*tldo_ldub_sp64"): Likewise.
24018 ("*tldo_ldub1_sp64"): Likewise.
24019 ("*tldo_ldub2_sp64"): Likewise.
24020 ("*tldo_ldub3_sp64"): Likewise.
24021 ("*tldo_lduh_sp32"): Likewise.
24022 ("*tldo_lduh1_sp32"): Likewise.
24023 ("*tldo_lduh_sp64"): Likewise.
24024 ("*tldo_lduh1_sp64"): Likewise.
24025 ("*tldo_lduh2_sp64"): Likewise.
24026 ("*tldo_lduw_sp32"): Likewise.
24027 ("*tldo_lduw_sp64"): Likewise.
24028 ("*tldo_lduw1_sp64"): Likewise.
24029 ("*tldo_ldx_sp64"): Likewise.
24030 ("*mov<VM32:mode>_insn"): Likewise.
24031 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24032 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24034 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24036 * config/sparc/sparc.md ("type"): New insn type viscmp.
24037 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24039 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24040 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24041 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24042 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24044 ("n7_vis_logical_11cycle"): Likewise.
24045 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24046 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24047 * config/sparc/niagara.md ("niag_vis"): Likewise.
24048 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24049 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24051 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24053 * config/sparc/sparc.md: New instruction type `bmask'.
24054 (bmaskdi_vis): Use the `bmask' type.
24055 (bmasksi_vis): Likewise.
24056 * config/sparc/ultra3.md (us3_array): Likewise.
24057 * config/sparc/niagara7.md (n7_array): Likewise.
24058 * config/sparc/niagara4.md (n4_array): Likewise.
24059 * config/sparc/niagara2.md (niag2_vis): Likewise.
24060 (niag3_vis): Likewise.
24061 * config/sparc/niagara.md (niag_vis): Likewise.
24063 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24065 * ipa-comdats.c: Remove optimize check from gate.
24066 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24067 for functions not optimized.
24068 (ipa_fn_summary_read): Skip optimize check.
24069 (ipa_fn_summary_write): Likewise.
24070 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24072 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24074 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24075 (check_callers): Give up on caller which is not optimized.
24076 (inline_small_functions): Likewise.
24077 (ipa_inline): Do not give up when not optimizing.
24078 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24079 away unoptimizes cdtors.
24080 (whole_program_function_and_variable_visibility): Do
24081 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24082 * ipa.c (process_references): Do not check optimize.
24083 (symbol_table::remove_unreachable_nodes): Update optimize check.
24084 (set_writeonly_bit): Update optimize check.
24085 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24086 (pass_ipa_single_use::gate): Remove.
24088 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24090 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24091 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24092 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24093 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24094 permute_load, permute_store, adjust_extract, adjust_splat,
24095 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24096 replace_swap_with_copy, dump_swap_insn_table,
24097 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24098 recombine_lvx_pattern, recombine_stvx_pattern,
24099 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24100 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24101 to file rs6000-p8swap.c.
24102 * config/rs6000/rs6000-p8swap.c: New file.
24103 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24104 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24105 and rs6000*-*-* targets.
24107 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24109 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24111 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24113 * lto-wrapper.c (merge_and_complain): Do not merge
24114 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24115 fsigned_zeros, ftrapping_math, fwrapv.
24116 (append_compiler_options): Do not track these options.
24117 (append_linker_options): Likewie
24119 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24121 * cgraphunit.c (cgraph_node::finalize_function): When
24122 !flag_toplevel_reorde set no_reorder flag.
24123 (varpool_node::finalize_decl): Likewise.
24124 (symbol_table::compile): Drop no toplevel reorder path.
24126 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24128 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24129 edges; zero probability is not better than uninitialized.
24131 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24133 * asan.h (asan_sanitize_allocas_p): Declare.
24134 * asan.c (asan_sanitize_allocas_p): New function.
24135 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24136 (handle_builtin_alloca): Likewise.
24137 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24138 if !asan_sanitize_allocas_p.
24139 * params.def (asan-instrument-allocas): Add new option.
24140 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24141 * opts.c (common_handle_option): Disable allocas sanitization for
24144 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24146 * asan.c: Include gimple-fold.h.
24147 (get_last_alloca_addr): New function.
24148 (handle_builtin_stackrestore): Likewise.
24149 (handle_builtin_alloca): Likewise.
24150 (asan_emit_allocas_unpoison): Likewise.
24151 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24152 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24153 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24154 (instrument_builtin_call): Pass gimple iterator to
24155 get_mem_refs_of_builtin_call.
24156 (last_alloca_addr): New global.
24157 * asan.h (asan_emit_allocas_unpoison): Declare.
24158 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24159 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24160 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24161 if function calls alloca.
24162 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24163 * gimple-fold.h (replace_call_with_value): Declare.
24164 * internal-fn.c: Include asan.h.
24165 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24166 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24168 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24170 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24171 (C_SELFTEST_FLAGS): New.
24172 (CPP_SELFTEST_FLAGS): New.
24173 (SELFTEST_DEPS): New, from deps of s-selftest.
24174 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24175 (CPP_SELFTEST_DEPS): New.
24176 (selftest): Add dependency on s-selftest-c++.
24177 (s-selftest): Rename to...
24178 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24179 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24180 than SELFTEST_FLAGS.
24181 (selftest-gdb): Rename to...
24182 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24184 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24185 (selftest-valgrind): Rename to...
24186 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24188 (selftest-valgrind): Reintroduce as an alias for
24189 selftest-c-valgrind.
24190 (s-selftest-c++): New.
24191 (selftest-c++-gdb): New.
24192 (selftest-c++-valgrind): New.
24194 2017-07-06 Olivier Hainque <hainque@adacore.com>
24196 * gcc.c (process_command): When deciding if undefined variables
24197 should be ignored when processing specs, accept "gcc -v" as well.
24199 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24201 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24202 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24204 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24206 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24207 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24208 * config/arm/arm-tables.opt: Regenerate.
24209 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24211 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24213 2017-07-06 Carl Love <cel@us.ibm.com>
24215 * ChangeLog: Clean up from mid air collision
24217 2017-07-06 Carl Love <cel@us.ibm.com>
24219 * config/rs6000/rs6000-c.c: Add support for built-in functions
24220 vector signed int vec_subc (vector signed int, vector signed int);
24221 vector signed __int128 vec_subc (vector signed __int128,
24222 vector signed __int128);
24223 vector unsigned __int128 vec_subc (vector unsigned __int128,
24224 vector unsigned __int128);
24225 vector signed int vec_sube (vector signed int, vector signed int,
24226 vector signed int);
24227 vector unsigned int vec_sube (vector unsigned int,
24228 vector unsigned int,
24229 vector unsigned int);
24230 vector signed __int128 vec_sube (vector signed __int128,
24231 vector signed __int128,
24232 vector signed__int128);
24233 vector unsigned __int128 vec_sube (vector unsigned __int128,
24234 vector unsigned __int128,
24235 vector unsigned __int128);
24236 vector signed int vec_subec (vector signed int, vector signed int,
24237 vector signed int);
24238 vector unsigned int vec_subec (vector unsigned int,
24239 vector unsigned int,
24240 vector unsigned int);
24241 vector signed __int128 vec_subec (vector signed __int128,
24242 vector signed __int128,
24243 vector signed__int128);
24244 vector unsigned __int128 vec_subec (vector unsigned __int128,
24245 vector unsigned __int128,
24246 vector unsigned __int128);
24247 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24248 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24249 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24250 BU_ALTIVEC_OVERLOAD_X definitions.
24251 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24252 * doc/extend.texi: Update the built-in documentation file for the new
24253 built-in functions.
24255 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24258 * diagnostic-show-locus.c (layout::layout): Use start and finish
24259 spelling location for the start and finish of each range.
24260 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24261 Add unused aspect param.
24262 * input.c (expand_location_1): Add "aspect" param, and use it
24263 to access the correct part of the location.
24264 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24266 (expand_location_to_spelling_point): Likewise.
24267 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24268 param, and pass it to expand_location_1.
24270 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24272 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24273 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24274 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24275 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24276 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24277 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24278 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24279 _mm_maskz_getmant_ss): New intrinsics.
24280 (__builtin_ia32_getexpss128_mask): Changed to ...
24281 __builtin_ia32_getexpss128_round ... this.
24282 (__builtin_ia32_getexpsd128_mask): Changed to ...
24283 __builtin_ia32_getexpsd128_round ... this.
24284 * config/i386/i386-builtin-types.def
24285 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24286 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24287 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24288 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24289 __builtin_ia32_getmantss_mask_round): New builtins.
24290 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24291 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24292 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24293 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24294 * config/i386/sse.md
24295 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24296 avx512f_sgetexp<mode><mask_scalar_name>
24297 <round_saeonly_scalar_name> ... this.
24298 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24299 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24300 vgetexp<ssescalarmodesuffix>
24301 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24302 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24303 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24304 avx512f_vgetmant<mode><mask_scalar_name>
24305 <round_saeonly_scalar_name> ... this.
24306 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24307 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24308 vgetmant<ssescalarmodesuffix>
24309 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24310 %0<mask_scalar_operand4>, %1, %2
24311 <round_saeonly_scalar_mask_op4>, %3} ... this.
24312 * config/i386/subst.md (mask_scalar_operand4,
24313 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24314 round_saeonly_scalar_nimm_predicate): New subst attributes.
24316 2017-07-06 Julia Koval <julia.koval@intel.com>
24318 * config/i386/i386.c (ix86_erase_embedded_rounding):
24319 Remove code for old rounding pattern.
24321 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24323 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24325 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24327 * doc/sourcebuild.texi (Test Directives, Variants of
24328 dg-require-support): Add documentation for dg-require-stack-check.
24330 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24332 * config/i386/subst.md (mask_scalar, round_scalar,
24333 round_saeonly_scalar): New meta-templates.
24334 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24335 round_scalar_mask_operand3, round_scalar_mask_op3,
24336 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24337 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24338 round_saeonly_scalar_constraint,
24339 round_saeonly_scalar_prefix): New subst attribute.
24340 * config/i386/sse.md
24341 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24342 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24343 <round_scalar_name> ... this.
24344 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24345 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24346 <round_scalar_name> ... this.
24347 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24348 <sse>_vm<code><mode>3<mask_scalar_name>
24349 <round_saeonly_scalar_name> ... this.
24350 (v<plusminus_mnemonic><ssescalarmodesuffix>
24351 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24352 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24353 v<plusminus_mnemonic><ssescalarmodesuffix>
24354 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24355 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24356 (v<multdiv_mnemonic><ssescalarmodesuffix>
24357 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24358 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24359 v<multdiv_mnemonic><ssescalarmodesuffix>
24360 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24361 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24362 (v<maxmin_float><ssescalarmodesuffix>
24363 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24364 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24365 v<maxmin_float><ssescalarmodesuffix>
24366 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24367 %0<mask_scalar_operand3>, %1, %<iptr>2
24368 <round_saeonly_scalar_mask_op3>} ... this.
24370 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24372 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24373 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24375 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24376 Alan Hayward <alan.hayward@arm.com>
24377 David Sherwood <david.sherwood@arm.com>
24379 * combine.c (simplify_if_then_else): Remove "enum" before
24381 * compare-elim.c (can_eliminate_compare): Likewise.
24382 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24384 (aarch64_lookup_simd_builtin_type): Likewise.
24385 (aarch64_simd_builtin_type): Likewise.
24386 (aarch64_init_simd_builtin_types): Likewise.
24387 (aarch64_simd_expand_args): Likewise.
24388 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24390 (aarch64_reverse_mask): Likewise.
24391 (aarch64_simd_emit_reg_reg_move): Likewise.
24392 (aarch64_gen_adjusted_ldpstp): Likewise.
24393 (aarch64_ccmp_mode_to_code): Likewise.
24394 (aarch64_operands_ok_for_ldpstp): Likewise.
24395 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24396 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24398 (aarch64_min_divisions_for_recip_mul): Likewise.
24399 (aarch64_reassociation_width): Likewise.
24400 (aarch64_get_condition_code_1): Likewise.
24401 (aarch64_simd_emit_reg_reg_move): Likewise.
24402 (aarch64_simd_attr_length_rglist): Likewise.
24403 (aarch64_reverse_mask): Likewise.
24404 (aarch64_operands_ok_for_ldpstp): Likewise.
24405 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24406 (aarch64_gen_adjusted_ldpstp): Likewise.
24407 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24409 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24410 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24411 (arm_lookup_simd_builtin_type): Likewise.
24412 (arm_simd_builtin_type): Likewise.
24413 (arm_init_simd_builtin_types): Likewise.
24414 (arm_expand_builtin_args): Likewise.
24415 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24416 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24417 (ft32_setup_incoming_varargs): Likewise.
24418 (ft32_function_arg): Likewise.
24419 (ft32_function_arg_advance): Likewise.
24420 (ft32_pass_by_reference): Likewise.
24421 (ft32_arg_partial_bytes): Likewise.
24422 (ft32_valid_pointer_mode): Likewise.
24423 (ft32_addr_space_pointer_mode): Likewise.
24424 (ft32_addr_space_legitimate_address_p): Likewise.
24425 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24427 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24428 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24429 (iamcu_alignment): Likewise.
24430 (canonicalize_vector_int_perm): Likewise.
24431 (ix86_noce_conversion_profitable_p): Likewise.
24432 (ix86_mpx_bound_mode): Likewise.
24433 (ix86_operands_ok_for_move_multiple): Likewise.
24434 * config/microblaze/microblaze-protos.h
24435 (microblaze_expand_conditional_branch_reg): Likewise.
24436 * config/microblaze/microblaze.c
24437 (microblaze_expand_conditional_branch_reg): Likewise.
24438 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24440 (rs6000_reassociation_width): Likewise.
24441 (rs6000_invalid_binary_op): Likewise.
24442 (fusion_p9_p): Likewise.
24443 (emit_fusion_p9_load): Likewise.
24444 (emit_fusion_p9_store): Likewise.
24445 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24447 (riscv_hard_regno_mode_ok_p): Likewise.
24448 (riscv_address_insns): Likewise.
24449 (riscv_split_symbol): Likewise.
24450 (riscv_legitimize_move): Likewise.
24451 (riscv_function_value): Likewise.
24452 (riscv_hard_regno_nregs): Likewise.
24453 (riscv_expand_builtin): Likewise.
24454 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24455 (riscv_build_integer): Likewise.
24456 (riscv_split_integer): Likewise.
24457 (riscv_legitimate_constant_p): Likewise.
24458 (riscv_cannot_force_const_mem): Likewise.
24459 (riscv_regno_mode_ok_for_base_p): Likewise.
24460 (riscv_valid_base_register_p): Likewise.
24461 (riscv_valid_offset_p): Likewise.
24462 (riscv_valid_lo_sum_p): Likewise.
24463 (riscv_classify_address): Likewise.
24464 (riscv_legitimate_address_p): Likewise.
24465 (riscv_address_insns): Likewise.
24466 (riscv_load_store_insns): Likewise.
24467 (riscv_force_binary): Likewise.
24468 (riscv_split_symbol): Likewise.
24469 (riscv_force_address): Likewise.
24470 (riscv_legitimize_address): Likewise.
24471 (riscv_move_integer): Likewise.
24472 (riscv_legitimize_const_move): Likewise.
24473 (riscv_legitimize_move): Likewise.
24474 (riscv_address_cost): Likewise.
24475 (riscv_subword): Likewise.
24476 (riscv_output_move): Likewise.
24477 (riscv_canonicalize_int_order_test): Likewise.
24478 (riscv_emit_int_order_test): Likewise.
24479 (riscv_function_arg_boundary): Likewise.
24480 (riscv_pass_mode_in_fpr_p): Likewise.
24481 (riscv_pass_fpr_single): Likewise.
24482 (riscv_pass_fpr_pair): Likewise.
24483 (riscv_get_arg_info): Likewise.
24484 (riscv_function_arg): Likewise.
24485 (riscv_function_arg_advance): Likewise.
24486 (riscv_arg_partial_bytes): Likewise.
24487 (riscv_function_value): Likewise.
24488 (riscv_pass_by_reference): Likewise.
24489 (riscv_setup_incoming_varargs): Likewise.
24490 (riscv_print_operand): Likewise.
24491 (riscv_elf_select_rtx_section): Likewise.
24492 (riscv_save_restore_reg): Likewise.
24493 (riscv_for_each_saved_reg): Likewise.
24494 (riscv_register_move_cost): Likewise.
24495 (riscv_hard_regno_mode_ok_p): Likewise.
24496 (riscv_hard_regno_nregs): Likewise.
24497 (riscv_class_max_nregs): Likewise.
24498 (riscv_memory_move_cost): Likewise.
24499 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24500 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24501 (rl78_addr_space_address_mode): Likewise.
24502 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24504 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24505 (rs6000_reassociation_width): Likewise.
24506 (rs6000_invalid_binary_op): Likewise.
24507 (fusion_p9_p): Likewise.
24508 (emit_fusion_p9_load): Likewise.
24509 (emit_fusion_p9_store): Likewise.
24510 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24511 (ok_for_simple_move_operands): Likewise.
24512 (ok_for_simple_move_strict_operands): Likewise.
24513 (ok_for_simple_arith_logic_operands): Likewise.
24514 (visium_legitimize_reload_address): Likewise.
24515 (visium_select_cc_mode): Likewise.
24516 (output_cbranch): Likewise.
24517 (visium_split_double_move): Likewise.
24518 (visium_expand_copysign): Likewise.
24519 (visium_expand_int_cstore): Likewise.
24520 (visium_expand_fp_cstore): Likewise.
24521 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24522 (visium_function_arg): Likewise.
24523 (visium_function_arg_advance): Likewise.
24524 (visium_libcall_value): Likewise.
24525 (visium_setup_incoming_varargs): Likewise.
24526 (visium_legitimate_constant_p): Likewise.
24527 (visium_legitimate_address_p): Likewise.
24528 (visium_legitimize_address): Likewise.
24529 (visium_secondary_reload): Likewise.
24530 (visium_register_move_cost): Likewise.
24531 (visium_memory_move_cost): Likewise.
24532 (prepare_move_operands): Likewise.
24533 (ok_for_simple_move_operands): Likewise.
24534 (ok_for_simple_move_strict_operands): Likewise.
24535 (ok_for_simple_arith_logic_operands): Likewise.
24536 (visium_function_value_1): Likewise.
24537 (rtx_ok_for_offset_p): Likewise.
24538 (visium_legitimize_reload_address): Likewise.
24539 (visium_split_double_move): Likewise.
24540 (visium_expand_copysign): Likewise.
24541 (visium_expand_int_cstore): Likewise.
24542 (visium_expand_fp_cstore): Likewise.
24543 (visium_split_cstore): Likewise.
24544 (visium_select_cc_mode): Likewise.
24545 (visium_split_cbranch): Likewise.
24546 (output_cbranch): Likewise.
24547 (visium_print_operand_address): Likewise.
24548 * expmed.c (flip_storage_order): Likewise.
24549 * expmed.h (emit_cstore): Likewise.
24550 (flip_storage_order): Likewise.
24551 * genrecog.c (validate_pattern): Likewise.
24552 * hsa-gen.c (gen_hsa_addr): Likewise.
24553 * internal-fn.c (expand_arith_overflow): Likewise.
24554 * ira-color.c (allocno_copy_cost_saving): Likewise.
24555 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24556 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24557 (process_invariant_for_inheritance): Likewise.
24558 * lra-eliminations.c (move_plus_up): Likewise.
24559 * omp-low.c (lower_oacc_reductions): Likewise.
24560 * simplify-rtx.c (simplify_subreg): Likewise.
24561 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24562 (TARGET_CHKP_BOUND_MODE): Likewise..
24563 * targhooks.c (default_chkp_bound_mode): Likewise.
24564 (default_setup_incoming_vararg_bounds): Likewise.
24565 * targhooks.h (default_chkp_bound_mode): Likewise.
24566 (default_setup_incoming_vararg_bounds): Likewise.
24567 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24568 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24569 (have_whole_vector_shift): Likewise.
24570 * tree-vect-stmts.c (vectorizable_load): Likewise.
24571 * doc/tm.texi: Regenerate.
24573 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24575 Graceful degrade if Binutils PR21472 is not available.
24578 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24579 .rodata in flash test fails.
24580 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24581 * confgure: Regenerate.
24582 * config.in: Regenerate.
24583 * config/avr/avr.c (avr_asm_named_section)
24584 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24585 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24586 (avr_asm_init_sections): Same.
24588 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24590 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24591 (fma<VH:mode>4_intrinsic): Likewise.
24592 (*fmsub<VCVTF:mode>4): Likewise.
24593 (*fmsub<VH:mode>4_intrinsic): Likewise.
24595 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24598 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24599 Don't depend on "optimize > 0".
24600 (out_movhi_r_mr, out_movqi_mr_r): Same.
24601 (out_movhi_mr_r, out_movqi_r_mr): Same.
24602 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24603 io_address_operand on "optimize > 0".
24605 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24607 * tree-loop-distribution.c: Add general explanantion on the pass.
24608 (generate_loops_for_partition): Mark distributed loop.
24609 (pg_add_dependence_edges): New parameter. Handle alias data
24610 dependence specially and record it in the parameter if asked.
24611 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24612 (init_partition_graph_vertices, add_partition_graph_edge): New.
24613 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24614 (free_partition_graph_vdata, build_partition_graph): New.
24615 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24616 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24617 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24618 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24619 (version_for_distribution_p, finalize_partitions): New.
24620 (distribute_loop): Handle alias data dependence specially. Factor
24621 out loop fusion code as functions and call these functions.
24623 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24625 * tree-loop-distribution.c (classify_partition): New parameter and
24626 better handle reduction statement.
24627 (rdg_build_partitions): Revise comment.
24628 (distribute_loop): Compute statements in all partitions and pass it
24629 to classify_partition.
24631 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24633 * tree-loop-distribution.c (enum partition_type): New.
24634 (struct partition): New field type.
24635 (partition_merge_into): Add parameter. Update partition type.
24636 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24637 (build_rdg_partition_for_vertex): Compute partition type.
24638 (rdg_build_partitions): Dump partition type.
24639 (distribute_loop): Update calls to partition_merge_into.
24641 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24643 * tree-loop-distribution.c (struct ddr_hasher): New.
24644 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24646 (classify_partition): Call get_data_dependence.
24647 (pg_add_dependence_edges): Ditto.
24648 (distribute_loop): Release data dependence hash table.
24650 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24652 * tree-loop-distribution.c (ref_base_address): Delete.
24653 (similar_memory_accesses): Rename ...
24654 (share_memory_accesses): ... to this. Check if partitions access
24655 the same memory reference.
24656 (distribute_loop): Call share_memory_accesses.
24658 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24660 * tree-loop-distribution.c (struct partition): New field recording
24661 its data reference.
24662 (partition_alloc, partition_free): Init and release data refs.
24663 (partition_merge_into): Merge data refs.
24664 (build_rdg_partition_for_vertex): Collect data refs for partition.
24665 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24667 (distribute_loop): Remve data refs from vertice data of partition
24670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24672 * tree-loop-distribution.c (params.h): Include header file.
24673 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24674 (datarefs_vec): New global var.
24675 (create_rdg_vertices): Use datarefs_vec directly.
24676 (free_rdg): Don't free data references.
24677 (build_rdg): Update use. Don't free data references.
24678 (distribute_loop): Compute global variable for data references.
24679 Bail out if there are too many data references.
24681 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24683 * tree-loop-distribution.c (loop_nest): New global var.
24684 (build_rdg): Use loop directly, rather than loop nest.
24685 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24687 (distribute_loop): Compute global variable loop nest. Update use.
24689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24691 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24692 (partition_merge_into): New parameter. Dump reason for fusion.
24693 (distribute_loop): Update use of partition_merge_into.
24695 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24697 * tree-loop-distribution.c (bb_top_order_index): New.
24698 (bb_top_order_index_size, bb_top_order_cmp): New.
24699 (stmts_from_loop): Use topological order.
24700 (pass_loop_distribution::execute): Compute and release topological
24701 order for basic blocks.
24703 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24705 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24708 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24710 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24711 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24712 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24713 * internal-fn.def (LOOP_DIST_ALIAS): New.
24714 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24715 (fold_loop_internal_call): ... this.
24716 (vect_loop_dist_alias_call): New function.
24717 (set_uid_loop_bbs): Call fold_loop_internal_call.
24718 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24721 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24724 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24725 Require dead FLAGS_REG at the beginning of a peephole.
24727 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24730 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24731 arguments in the call to __builtin_ia32_sbb_u32.
24732 (_subborrow_u64): Swap _X and _Y arguments in the call to
24733 __builtin_ia32_sbb_u64.
24735 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24738 * tree-vrp.c (compare_assert_loc): Turn into a function template
24739 with stable template parameter. Only test if a->e is NULL,
24740 !a->e == !b->e has been verified already. Use e == NULL or
24741 e != NULL instead of e or ! e tests. If stable is true, don't use
24742 iterative_hash_expr, on the other side allow a or b or both NULL
24743 and sort the NULLs last.
24744 (process_assert_insertions): Sort using compare_assert_loc<false>
24745 instead of compare_assert_loc, later sort using
24746 compare_assert_loc<true> before calling process_assert_insertions_for
24747 in a loop. Use break instead of continue once seen NULL pointer.
24749 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24751 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24752 Cortex-R7 and Cortex-R8 processors.
24754 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24756 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24757 uninitialized while src is not.
24759 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24761 * common/config/arm/arm-common.c: Adjust include path for
24763 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24764 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24765 (arm-cpu-data.h): Likewise.
24766 (arm-cpu-cdata.h): Likewise.
24767 * config/arm/arm-cpu.h: Delete.
24768 * config/arm/arm-cpu-cdata.h: Delete.
24769 * config/arm/arm-cpu-data.h: Delete.
24771 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24773 * config/arm/arm-cpus.in (cortex-a55): New.
24774 (cortex-a75): Likewise.
24775 (cortex-a75.cortex-a55): Likewise.
24776 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24778 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24779 * config/arm/arm-cpu-cdata.h: Regenerate.
24780 * config/arm/arm-cpu-data.h: Regenerate.
24781 * config/arm/arm-cpu.h: Regenerate.
24782 * config/arm/arm-tables.opt: Regenerate.
24783 * config/arm/arm-tune.md: Regenerate.
24785 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24787 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24789 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24791 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24794 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24796 PR tree-optimization/81292
24797 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24798 full_string_p, also call adjust_related_strinfos if the adjustment
24799 is simple, otherwise invalidate related strinfos.
24801 2017-07-04 Martin Liska <mliska@suse.cz>
24804 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24805 newly created variable as DECL_IGNORED_P.
24807 2017-07-04 Martin Liska <mliska@suse.cz>
24810 * ipa-inline.c (inline_small_functions):
24811 Use xstrdup_for_dump.
24813 2017-07-04 Tom de Vries <tom@codesourcery.com>
24815 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24817 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24820 * config/darwin.c (darwin_function_switched_text_sections):
24823 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24825 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24827 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24829 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24833 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24834 min_profitable_iters, and th as inclusive lower bounds.
24835 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24836 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24837 for min_profitable_iters and min_profitable_estimate.
24838 (vect_transform_loop): Treat th as an inclusive lower bound.
24839 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24841 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24844 * config/darwin.c (darwin_function_switched_text_sections):
24845 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24846 in two pieces, and suppress the use of buf.
24848 2017-07-03 Nathan Sidwell <nathan@acm.org>
24850 * hash-table.h (hash_table_mod1): Fix indentation.
24852 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24854 PR middle-end/81290
24855 * predict.c (force_edge_cold): Be more careful about propagation
24857 * profile-count.h (profile_probability::guessed,
24858 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24860 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24862 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24864 * doc/invoke.texi (rcpc architecture extension): Document it.
24866 2017-07-03 Richard Biener <rguenther@suse.de>
24868 PR tree-optimization/60510
24869 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24870 the scalar reduction PHI and use it.
24871 (vectorizable_reduction): Properly guard the single_defuse_cycle
24872 path for non-SLP reduction chains where we cannot use it.
24873 Rework reduc_def/index and vector type deduction. Rework
24874 vector operand gathering during reduction op code-gen.
24875 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24876 chains dissolve the chain and leave it to non-SLP reduction
24879 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24881 * tree-data-ref.h (dr_alignment): Declare.
24882 * tree-data-ref.c (dr_alignment): New function.
24883 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24884 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24886 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24888 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24890 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24891 and base_misalignment fields.
24892 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24893 * tree-data-ref.c: Include builtins.h.
24894 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24895 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24896 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24897 * tree-vect-data-refs.c: Include tree-cfg.h.
24898 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24899 fields instead of calculating an alignment here.
24900 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24901 innermost_loop_behavior fields.
24903 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24905 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24907 (DR_STEP_ALIGNMENT): New macro.
24908 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24909 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24910 (create_data_ref): Print it.
24911 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24912 to tell whether the step preserves vector (mis)alignment.
24913 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24914 Move the check for an integer step and generalise to all INTEGER_CST.
24915 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24916 Print the outer step alignment.
24918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24920 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24921 with offset_alignment.
24922 (DR_ALIGNED_TO): Delete.
24923 (DR_OFFSET_ALIGNMENT): New macro.
24924 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24925 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24926 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24927 (create_data_ref): Likewise.
24928 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24929 (vect_analyze_data_refs): Likewise.
24930 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24931 creating dummy innermost behavior.
24933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24935 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24936 with a "innermost_loop_behavior *" and refeence tree.
24937 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24938 (create_data_ref): Update call accordingly.
24939 * tree-predcom.c (find_looparound_phi): Likewise.
24941 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24943 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24944 fields with dr_wrt_vec_loop.
24945 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24946 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24947 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24948 (vect_dr_behavior): New function.
24949 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24950 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24951 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24952 track whether the step preserves the misalignment.
24953 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24954 Use vect_dr_behavior.
24955 (vect_setup_realignment): Update call accordingly.
24956 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24957 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24958 call to vect_create_addr_base_for_vector_ref.
24959 (vect_create_cond_for_align_checks): Likewise.
24960 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24961 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24962 (vect_recog_mask_conversion_pattern): Likewise.
24963 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24964 (new_stmt_vec_info): Remove redundant zeroing.
24966 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24968 * common/config/arm/arm-common.c (arm_be8_option): New function.
24969 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24970 (ISA_ARMv6): Add isa_bit_be8.
24971 * config/arm/arm.h (arm_be8_option): Add prototype.
24972 (BE8_SPEC_FUNCTION): New define.
24973 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24974 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24975 (mlittle-endian): Similarly.
24976 (mbe8, mbe32): New options.
24977 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24978 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24982 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24986 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24987 (cleanup_tree_cfg_bb): Use it.
24988 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24990 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24992 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24995 * loop-doloop.c (add_test): Update profile.
24997 2017-07-03 Martin Liska <mliska@suse.cz>
25000 * sanopt.c (rewrite_usage_of_param): New function.
25001 (sanitize_rewrite_addressable_params): Likewise.
25002 (pass_sanopt::execute): Call rewrite_usage_of_param.
25004 2017-07-03 Richard Biener <rguenther@suse.de>
25006 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25007 back to using VIEW_CONVERT_EXPR.
25009 2017-07-03 Martin Liska <mliska@suse.cz>
25012 * doc/extend.texi: Document when a resolver function is
25013 generated for target_clones.
25015 2017-07-03 Martin Liska <mliska@suse.cz>
25017 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25018 and shadow memory of auto variables which are subject of
25019 use-after-scope sanitization.
25020 (asan_expand_mark_ifn): Add do set only when is_poison.
25022 2016-07-03 Richard Biener <rguenther@suse.de>
25024 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25026 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25027 (vectorizable_reduction): Perform reduction PHI creation when
25028 visiting a reduction PHI and adjust and simplify code generation
25029 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25030 (vect_transform_loop): Visit reduction PHIs.
25031 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25032 defs into the SLP tree.
25033 (vect_build_slp_tree): Reduction defs terminate the recursion.
25034 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25036 (vect_get_vec_defs_for_stmt_copy): Export.
25037 (vect_get_vec_defs): Likewise.
25038 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25040 (vect_get_vec_defs_for_stmt_copy): Declare.
25041 (vect_get_vec_defs): Likewise.
25043 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25045 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25046 parameter with a "loop" parameter and use it instead of the
25047 loop containing DR_STMT. Don't check simple_iv when doing
25048 BB analysis. Describe the two analysis modes in the comment.
25050 2017-07-03 Tom de Vries <tom@codesourcery.com>
25052 PR tree-optimization/69468
25053 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25054 (find_same_succ_bb): Handle ignore_edge_flags.
25056 2017-07-03 Tom de Vries <tom@codesourcery.com>
25058 PR tree-optimization/81192
25059 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25061 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25063 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25065 2017-07-03 Tom de Vries <tom@codesourcery.com>
25067 PR tree-optimization/81192
25068 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25069 BB_SAME_SUCC (bb) == NULL.
25071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25073 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25076 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25078 * dumpfile.c: Include profile-count.h
25079 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25081 (insert_cond_bb): Update profile.
25082 * tree-cfg.h (insert_cond_bb): Update prototype.
25083 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25084 * tree-dump.c: Do not include tree-cfg.
25086 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25088 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25090 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25092 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25095 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25097 * tree-complex.c (expand_complex_div_wide): update profile.
25099 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25100 Alan Hayward <alan.hayward@arm.com>
25101 David Sherwood <david.sherwood@arm.com>
25103 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25104 (CORETYPES_H): New define.
25105 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25106 (insn-modes-inline.h, s-modes-inline-h): New rules.
25107 (generated_files): Add insn-modes-inline.h.
25108 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25109 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25110 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25111 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25112 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25113 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25114 (build/gencodes.o, build/genconditions.o): Likewise.
25115 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25116 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25117 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25118 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25119 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25120 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25121 * coretypes.h: Include everything up to real.h for generators.
25122 Include insn-modes.h first. Include wide-int-print.h after
25123 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25124 * machmode.h: Don't include insn-modes.h here.
25125 * function-tests.c: Remove includes of signop.h, machmode.h,
25126 double-int.h and wide-int.h.
25128 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25130 * optc-save-gen.awk: Likewise.
25131 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25132 * godump.c: Remove include of wide-int-print.h.
25133 * pretty-print.h: Likewise.
25134 * wide-int-print.cc: Likewise.
25135 * wide-int.cc: Likewise.
25136 * hash-map-tests.c: Remove include of signop.h.
25137 * hash-set-tests.c: Likewise.
25138 * rtl-tests.c: Likewise.
25139 * mkconfig.sh: Remove include of machmode.h.
25140 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25142 (emit_insn_modes_inline_h): ...this new function. Emit the code
25143 into an insn-modes-inline.h header file, adding appropriate
25144 include guards and end comments.
25145 (emit_insn_modes_c_header): Remove include of machmode.h.
25146 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25148 (main): Handle -i flag and call emit_insn_modes_inline_h when
25151 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25153 * tree-ssa-strlen.c (strinfo): Rename the length field to
25154 nonzero_chars. Add a full_string_p field.
25155 (compare_nonzero_chars, zero_length_string_p): New functions.
25156 (get_addr_stridx): Add an offset_out parameter.
25157 Use compare_nonzero_chars.
25158 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25159 (new_strinfo): Update after above changes to strinfo.
25160 (set_endptr_and_length): Set full_string_p.
25161 (get_string_length): Update after above changes to strinfo.
25162 (unshare_strinfo): Update call to new_strinfo.
25163 (maybe_invalidate): Likewise.
25164 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25165 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25166 as a uhwi instead of an shwi. Update after above changes to
25167 strinfo and new_strinfo.
25168 (zero_length_string): Assert that chainsi contains full strings.
25169 Use zero_length_string_p. Update call to new_strinfo.
25170 (adjust_related_strinfos): Update after above changes to strinfo.
25171 Copy full_string_p from origsi.
25172 (adjust_last_stmt): Use zero_length_string_p.
25173 (handle_builtin_strlen): Update after above changes to strinfo and
25174 new_strinfo. Install the lhs as the string length if the previous
25175 entry didn't describe a full string.
25176 (handle_builtin_strchr): Update after above changes to strinfo
25178 (handle_builtin_strcpy): Likewise.
25179 (handle_builtin_strcat): Likewise.
25180 (handle_builtin_malloc): Likewise.
25181 (handle_pointer_plus): Likewise.
25182 (handle_builtin_memcpy): Likewise. Track nonzero characters
25183 that aren't necessarily followed by a nul terminator.
25184 (handle_char_store): Likewise.
25186 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25188 PR tree-optimization/80769
25189 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25190 for malloc and calloc. Document the new invariant that all related
25191 strinfos have delayed lengths or none do.
25192 (verify_related_strinfos): Move earlier in file.
25193 (set_endptr_and_length): New function, split out from...
25194 (get_string_length): ...here. Also set the lengths of related
25196 (zero_length_string): Assert that chainsi has known (rather than
25198 (adjust_related_strinfos): Likewise.
25200 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25202 PR tree-optimization/81136
25203 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25204 assert that two references with the same misalignment have the same
25205 compile-time misalignment if those compile-time misalignments
25208 2017-07-01 Andi Kleen <ak@linux.intel.com>
25210 * print-tree.c (print_node): Print all attributes.
25212 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25214 * cfg.c (scale_bbs_frequencies): New function.
25215 * cfg.h (scale_bbs_frequencies): Declare it.
25216 * cfgloopanal.c (single_likely_exit): Cleanup.
25217 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25219 (scale_loop_profile): Likewise.
25220 (loop_version): Likewise.
25221 (create_empty_loop_on_edge): Update.
25222 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25223 scale_loop_frequencies, scale_loop_profile, loopify,
25224 loop_version): Update prototypes.
25225 * modulo-sched.c (sms_schedule): Update.
25226 * predict.c (unlikely_executed_edge_p): Also check probability.
25227 (probably_never_executed_edge_p): Fix typo.
25228 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25229 * tree-parloops.c (gen_parallel_loop): Update.
25230 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25231 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25232 * tree-ssa-loop-split.c (split_loop): Update.
25233 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25234 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25235 (vect_loop_versioning): Update.
25236 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25238 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25240 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25242 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25244 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25245 to keep profile consistent.
25247 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25249 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25250 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25251 * profile-count.h (max_safe_multiplier): Make unsigned.
25252 (profile_count::guessed_zero): New.
25254 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25256 * bb-reorder.c (fix_up_crossing_landing_pad,
25257 fix_crossing_conditional_branches): Use make_single_succ_edge
25258 to keep profile consistent.
25260 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25262 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25265 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25268 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25269 the right scopes, make sure cond_jump isn't preserved between multiple
25270 iterations. Search for fallthru edge whenever there are 3+ edges and
25271 use find_fallthru_edge for it.
25273 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25275 Patch by Alexander Monakov <amonakov@ispras.ru>
25276 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25277 probabilities consistently.
25279 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25281 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25282 emit_cmp_and_jump_insns.
25284 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25287 * tree-inline.c (expand_call_inline): Combine profile statuses.
25289 2017-06-30 Andrew Pinski <apinski@cavium.com>
25291 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25292 fold_stmt returned true.
25294 2017-06-30 Nathan Sidwell <nathan@acm.org>
25296 * ggc.h (empty_string): Delete.
25297 * cfgexpand.c (expand_asm_stmt): Use plain "".
25298 * optabs.c (expand_asm_memory_barrier): Likewise.
25299 * stringpool.c (empty_string): Delete.
25300 (digit_vector, digit_string): Delete.
25301 (ggc_alloc_string): Use plain "", don't optimize single digit
25302 strings. Use ggc_alloc_atomic.
25304 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25306 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25307 comparison set and one other set, use the cost of the non-comparison
25310 2017-06-30 Nathan Sidwell <nathan@acm.org>
25312 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25315 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25317 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25318 loops. Remove now unneeded calls to gimple_switch_set_label() that
25319 just set removed labels to NULL_TREE.
25321 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25323 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25324 (set_range_info): ...here. Only call set_range_info_raw if domain
25326 (set_nonzero_bits): Call set_range_info_raw.
25327 * tree-ssanames.h (set_range_info_raw): New.
25329 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25332 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25333 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25334 of nonimmediate_operand and <store_mask_constraint> instead of m
25335 for the input operand. For V8FI iterator, always split if input
25336 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25337 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25338 <store_mask_predicate> instead of register_operand and
25339 <store_mask_constraint> instead of v for the input operand. Make
25340 sure both operands aren't MEMs for if not <mask_applied>.
25342 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25344 * lto-wrapper.c (copy_file) Close both file descriptors before
25347 2017-06-30 Martin Liska <mliska@suse.cz>
25350 * multiple_target.c (create_dispatcher_calls): Make ifunc
25351 also for function that don't have calls or are not referenced.
25353 2017-06-30 Richard Biener <rguenther@suse.de>
25355 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25356 analyze the first scalar stmt. Move vector type computation
25357 for the BB case here from ...
25358 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25359 live operation processing in the SLP case properly.
25361 2017-06-30 Richard Biener <rguenther@suse.de>
25363 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25365 2017-06-30 Martin Liska <mliska@suse.cz>
25368 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25369 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25371 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25373 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25376 2017-06-30 Richard Biener <rguenther@suse.de>
25378 PR tree-optimization/81249
25379 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25380 condition reduction result to original scalar type.
25382 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25384 * profile-count.h (enum profile_quality): Fix typos and whitespace
25387 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25389 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25390 type for branch probabilities.
25392 2017-06-29 Julian Brown <julian@codesourcery.com>
25393 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25395 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25396 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25397 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25398 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25400 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25402 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25403 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25404 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25405 CC usage from generic code to here.
25406 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25407 CC usage into the target macros.
25409 2017-06-29 Maya Rashish <coypu@sdf.org>
25411 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25414 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25416 * arm/arm-builtins.c: Include profile-count.h
25417 * except.c (sjlj_emit_function_enter): Use
25418 profile_probability::unlikely.
25420 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25422 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25423 and tocrel_offset be pointer args rather than implicitly using
25425 (legitimate_constant_pool_address_p, rs6000_emit_move,
25426 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25427 tocrel_offset and use in toc_relative_expr_p call.
25428 (print_operand, print_operand_address): Use static tocrel_base_oac
25429 and tocrel_offset_oac.
25430 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25433 2017-06-29 Maya Rashish <coypu@sdf.org>
25435 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25437 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25439 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25440 objects, take into account only the alignment of 'op0' and 'mode1' if
25443 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25445 * ccmp.c (ccmp_tree_comparison_p): New function.
25446 (ccmp_candidate_p): Update to use above function.
25447 (get_compare_parts): New function.
25448 (expand_ccmp_next): Update to use new functions.
25449 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25451 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25452 take mode as argument.
25453 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25454 * expr.c (expand_expr_real_1): Pass mode as argument.
25456 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25458 * combine.c (combine_instructions): Print insns to dump_file, together
25461 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25463 * asan.c (asan_emit_stack_protection): Update.
25464 (create_cond_insert_point): Update.
25465 * auto-profile.c (afdo_propagate_circuit): Update.
25466 * basic-block.h (struct edge_def): Turn probability to
25467 profile_probability.
25468 (EDGE_FREQUENCY): Update.
25469 * bb-reorder.c (find_traces_1_round): Update.
25470 (better_edge_p): Update.
25471 (sanitize_hot_paths): Update.
25472 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25473 (make_single_succ_edge): Update.
25474 (check_bb_profile): Update.
25475 (dump_edge_info): Update.
25476 (update_bb_profile_for_threading): Update.
25477 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25478 probabilitycount to 0.
25479 * cfgbuild.c (compute_outgoing_frequencies): Update.
25480 * cfgcleanup.c (try_forward_edges): Update.
25481 (outgoing_edges_match): Update.
25482 (try_crossjump_to_edge): Update.
25483 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25484 (expand_gimple_tailcall): Update.
25485 (construct_init_block): Use make_single_succ_edge.
25486 (construct_exit_block): Use make_single_succ_edge.
25487 * cfghooks.c (verify_flow_info): Update.
25488 (redirect_edge_succ_nodup): Update.
25489 (split_edge): Update.
25490 (account_profile_record): Update.
25491 * cfgloopanal.c (single_likely_exit): Update.
25492 * cfgloopmanip.c (scale_loop_profile): Update.
25493 (set_zero_probability): Remove.
25494 (duplicate_loop_to_header_edge): Update.
25495 * cfgloopmanip.h (loop_version): Update prototype.
25496 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25497 (force_nonfallthru_and_redirect): Update.
25498 (update_br_prob_note): Update.
25499 (rtl_verify_edges): Update.
25500 (purge_dead_edges): Update.
25501 (rtl_lv_add_condition_to_bb): Update.
25502 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25503 * cgraphunit.c (init_lowered_empty_function): Update.
25504 (cgraph_node::expand_thunk): Update.
25505 * cilk-common.c: Include profile-count.h
25506 * dojump.c (inv): Remove.
25507 (jumpifnot): Update.
25508 (jumpifnot_1): Update.
25509 (do_jump_1): Update.
25511 (do_jump_by_parts_greater_rtx): Update.
25512 (do_compare_rtx_and_jump): Update.
25513 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25514 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25515 * dwarf2cfi.c: Include profile-count.h
25516 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25517 (sjlj_emit_dispatch_table): Likewise.
25518 * explow.c: Include profile-count.h
25519 * expmed.c (emit_store_flag_force): Update.
25520 (do_cmp_and_jump): Update.
25521 * expr.c (compare_by_pieces_d::generate): Update.
25522 (compare_by_pieces_d::finish_mode): Update.
25523 (emit_block_move_via_loop): Update.
25524 (store_expr_with_bounds): Update.
25525 (store_constructor): Update.
25526 (expand_expr_real_2): Update.
25527 (expand_expr_real_1): Update.
25528 * expr.h (try_casesi, try_tablejump): Update prototypes.
25529 * gimple-pretty-print.c (dump_probability): Update.
25530 (dump_profile): New.
25531 (dump_gimple_label): Update.
25532 (dump_gimple_bb_header): Update.
25533 * graph.c (draw_cfg_node_succ_edges): Update.
25534 * hsa-gen.c (convert_switch_statements): Update.
25535 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25536 (find_if_case_1): Update.
25537 (find_if_case_2): Update.
25538 * internal-fn.c (expand_arith_overflow_result_store): Update.
25539 (expand_addsub_overflow): Update.
25540 (expand_neg_overflow): Update.
25541 (expand_mul_overflow): Update.
25542 (expand_vector_ubsan_overflow): Update.
25543 * ipa-cp.c (good_cloning_opportunity_p): Update.
25544 * ipa-split.c (split_function): Use make_single_succ_edge.
25545 * ipa-utils.c (ipa_merge_profiles): Update.
25546 * loop-doloop.c (add_test): Update.
25547 (doloop_modify): Update.
25548 * loop-unroll.c (compare_and_jump_seq): Update.
25549 (unroll_loop_runtime_iterations): Update.
25550 * lra-constraints.c (lra_inheritance): Update.
25551 * lto-streamer-in.c (input_cfg): Update.
25552 * lto-streamer-out.c (output_cfg): Update.
25553 * mcf.c (adjust_cfg_counts): Update.
25554 * modulo-sched.c (sms_schedule): Update.
25555 * omp-expand.c (expand_omp_for_init_counts): Update.
25556 (extract_omp_for_update_vars): Update.
25557 (expand_omp_ordered_sink): Update.
25558 (expand_omp_for_ordered_loops): Update.
25559 (expand_omp_for_generic): Update.
25560 (expand_omp_for_static_nochunk): Update.
25561 (expand_omp_for_static_chunk): Update.
25562 (expand_cilk_for): Update.
25563 (expand_omp_simd): Update.
25564 (expand_omp_taskloop_for_outer): Update.
25565 (expand_omp_taskloop_for_inner): Update.
25566 * omp-simd-clone.c (simd_clone_adjust): Update.
25567 * optabs.c (expand_doubleword_shift): Update.
25568 (expand_abs): Update.
25569 (emit_cmp_and_jump_insn_1): Update.
25570 (expand_compare_and_swap_loop): Update.
25571 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25572 * predict.c (predictable_edge_p): Update.
25573 (edge_probability_reliable_p): Update.
25574 (set_even_probabilities): Update.
25575 (combine_predictions_for_insn): Update.
25576 (combine_predictions_for_bb): Update.
25577 (propagate_freq): Update.
25578 (estimate_bb_frequencies): Update.
25579 (force_edge_cold): Update.
25580 * profile-count.c (profile_count::dump): Add missing space into dump.
25581 (profile_count::debug): Add newline.
25582 (profile_count::differs_from_p): Explicitly convert to unsigned.
25583 (profile_count::stream_in): Update.
25584 (profile_probability::dump): New member function.
25585 (profile_probability::debug): New member function.
25586 (profile_probability::differs_from_p): New member function.
25587 (profile_probability::differs_lot_from_p): New member function.
25588 (profile_probability::stream_in): New member function.
25589 (profile_probability::stream_out): New member function.
25590 * profile-count.h (profile_count_quality): Rename to ...
25591 (profile_quality): ... this one.
25592 (profile_probability): New.
25593 (profile_count): Update.
25594 * profile.c (compute_branch_probabilities): Update.
25595 * recog.c (peep2_attempt): Update.
25596 * sched-ebb.c (schedule_ebbs): Update.
25597 * sched-rgn.c (find_single_block_region): Update.
25598 (compute_dom_prob_ps): Update.
25599 (schedule_region): Update.
25600 * sel-sched-ir.c (compute_succs_info): Update.
25601 * stmt.c (struct case_node): Update.
25602 (do_jump_if_equal): Update.
25603 (get_outgoing_edge_probs): Update.
25604 (conditional_probability): Update.
25605 (emit_case_dispatch_table): Update.
25606 (expand_case): Update.
25607 (expand_sjlj_dispatch_table): Update.
25608 (emit_case_nodes): Update.
25609 * targhooks.c: Update.
25610 * tracer.c (better_p): Update.
25611 (find_best_successor): Update.
25612 * trans-mem.c (expand_transaction): Update.
25613 * tree-call-cdce.c: Update.
25614 * tree-cfg.c (gimple_split_edge): Upate.
25615 (move_sese_region_to_fn): Upate.
25616 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25617 * tree-eh.c (lower_resx): Upate.
25618 (cleanup_empty_eh_move_lp): Upate.
25619 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25620 * tree-inline.c (copy_edges_for_bb): Update.
25621 (copy_cfg_body): Update.
25622 * tree-parloops.c (gen_parallel_loop): Update.
25623 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25624 (gimple_gen_time_profiler): Update.
25625 * tree-ssa-dce.c (remove_dead_stmt): Update.
25626 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25627 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25628 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25629 (unloop_loops): Update.
25630 (try_peel_loop): Update.
25631 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25632 * tree-ssa-loop-split.c (connect_loops): Update.
25633 (split_loop): Update.
25634 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25635 (hoist_guard): Update.
25636 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25637 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25638 (value_replacement): Update.
25639 * tree-ssa-reassoc.c (branch_fixup): Update.
25640 * tree-ssa-tail-merge.c (replace_block_by): Update.
25641 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25642 (create_edge_and_update_destination_phis): Update.
25643 (compute_path_counts): Update.
25644 (recompute_probabilities): Update.
25645 (update_joiner_offpath_counts): Update.
25646 (freqs_to_counts_path): Update.
25647 (duplicate_thread_path): Update.
25648 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25649 (struct switch_conv_info): Update.
25650 (gen_inbound_check): Update.
25651 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25652 (vect_do_peeling): Update.
25653 (vect_loop_versioning): Update.
25654 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25655 (optimize_mask_stores): Update.
25656 * ubsan.c (ubsan_expand_null_ifn): Update.
25657 * value-prof.c (gimple_divmod_fixed_value): Update.
25658 (gimple_divmod_fixed_value_transform): Update.
25659 (gimple_mod_pow2): Update.
25660 (gimple_mod_pow2_value_transform): Update.
25661 (gimple_mod_subtract): Update.
25662 (gimple_mod_subtract_transform): Update.
25663 (gimple_ic): Update.
25664 (gimple_stringop_fixed_value): Update.
25665 (gimple_stringops_transform): Update.
25666 * value-prof.h: Update.
25668 2017-06-29 Carl Love <cel@us.ibm.com>
25670 * config/rs6000/rs6000-c.c: Add support for built-in functions
25671 vector signed int vec_signed (vector float);
25672 vector signed long long vec_signed (vector double);
25673 vector signed int vec_signed2 (vector double, vector double);
25674 vector signed int vec_signede (vector double);
25675 vector signed int vec_signedo (vector double);
25676 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25677 instruction generator.
25678 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25679 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25680 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25682 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25683 vunsignede_v2df): Add define_expands.
25684 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25685 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25686 VEC_UNSIGNEDO): Add definitions.
25687 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25688 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25689 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25690 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25691 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25692 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25693 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25694 vec_unsignede, vec_unsignedo): Add builtin defines.
25695 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25697 * doc/extend.texi: Update the built-in documentation file for the
25698 new built-in functions.
25700 2017-06-29 Richard Biener <rguenther@suse.de>
25702 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25703 reduction chains to LOOP_VINFO_REDUCTIONS.
25704 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25705 SLP reductions after processing reduction chains.
25707 2017-06-29 Nathan Sidwell <nathan@acm.org>
25709 * builtins.c (fold_builtin_FUNCTION): Use
25710 lang_hooks.decl_printable_name.
25712 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25714 PR middle-end/81194
25715 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25716 with only one label.
25717 * stmt.c (expand_case): Assert NCASES is greater than one.
25719 2017-06-29 Richard Biener <rguenther@suse.de>
25721 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25723 (group_case_labels): Likewise.
25724 (find_taken_edge): Push sanity checking on val to workers...
25725 (find_taken_edge_cond_expr): ... here
25726 (find_taken_edge_switch_expr): ... and here, handle cases
25727 with just a default label.
25728 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25729 (group_case_labels): Likewise.
25730 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25731 group_case_labels does anything cleanup the CFG again.
25733 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25735 PR tree-optimization/81196
25736 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25737 exit condition comparing two IVs.
25739 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25741 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25742 profile to the dummy entry at the end of the list of architectures.
25743 * config/arm/arm-cpu-cdata.h: Regenerated.
25745 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25746 Michael Collison <michael.collison@arm.com>
25749 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25751 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25752 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25753 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25754 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25755 of shift when the shift amount is masked with constant equal to
25756 the size of the mode.
25757 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25760 2017-06-29 Martin Liska <mliska@suse.cz>
25762 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25764 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25767 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25770 (edge_growth_cache_entry::edge_growth_cache_entry): New
25772 (reset_edge_growth_cache): Update to use constructor.
25774 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25776 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25777 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25778 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25780 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25782 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25783 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25785 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25787 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25788 (*-linux-uclibc*): Add t-uclibc tmake_file.
25789 * config/t-musl: New.
25790 * config/t-uclibc: New.
25792 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25794 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25796 (gen_comm_data): Emit architectural setting of arch_prof.
25797 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25799 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25800 (armv8-m.base, armv8-m.main): Likewise.
25801 * arm-protos.h (arm_build_target): Add profile field.
25802 (arch_option): Likewise.
25803 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25805 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25806 arm_active_target.profile.
25808 2017-06-28 Richard Biener <rguenther@suse.de>
25810 PR middle-end/81227
25811 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25812 TYPE_OVERFLOW_WRAPS.
25813 * match.pd (negate_expr_p): Likewise.
25814 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25815 fold_build2, not fold_binary.
25817 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25819 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25820 Convert memory address to Pmode.
25821 (aarch64_print_operand): Assert MEM operands are always Pmode.
25823 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25826 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25827 Remove redundant if.
25828 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25829 * config/arm/aarch-common-protos.h
25830 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25831 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25833 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25836 * multiple_target.c (create_dispatcher_calls): Set the default
25837 clone to be static, not public.
25839 2017-06-28 Richard Biener <rguenther@suse.de>
25841 * tree-vect-loop.c (vectorizable_reduction): Move special
25842 cond reduction IV var creation ...
25843 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25844 parameter. Use STMT_VINFO_VECTYPE.
25845 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25848 2017-06-28 Martin Liska <mliska@suse.cz>
25851 * ipa-visibility.c (non_local_p): Handle visibility.
25853 2017-06-28 Martin Liska <mliska@suse.cz>
25856 * common.opt: Add IntegerRange to various options.
25857 * opt-functions.awk (integer_range_info): New function.
25858 * optc-gen.awk: Add integer_range_info to cl_options struct.
25859 * opts-common.c (decode_cmdline_option): Handle
25860 CL_ERR_INT_RANGE_ARG.
25861 (cmdline_handle_error): Likewise.
25862 * opts.c (print_filtered_help): Show valid interval in
25863 when --help is provided.
25864 * opts.h (struct cl_option): Add range_min and range_max fields.
25865 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25867 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25869 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25870 (x * C EQ/NE y * C): New transformation.
25872 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25874 * genmultilib (combination_space): Accept '+' in option names.
25876 2017-06-28 Martin Liska <mliska@suse.cz>
25879 * asan.c (instrument_derefs): Bail out inner references
25880 that are hard register variables.
25882 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25885 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25886 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25888 2017-06-28 Richard Biener <rguenther@suse.de>
25890 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25891 (vect_get_slp_defs): Adjust.
25892 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25894 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25896 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25897 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25898 (vectorizable_reduction): Adjust.
25899 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25901 (vect_get_slp_defs): Likewise.
25902 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25903 (vectorizable_bswap): Adjust.
25904 (vectorizable_call): Likewise.
25905 (vectorizable_conversion): Likewise.
25906 (vectorizable_assignment): Likewise.
25907 (vectorizable_shift): Likewise.
25908 (vectorizable_operation): Likewise.
25909 (vectorizable_store): Likewise.
25910 (vectorizable_condition): Likewise.
25911 (vectorizable_comparison): Likewise.
25913 2017-06-28 Michael Collison <michael.collison@arm.com>
25916 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25918 (arm_gen_movmemqi): Removed unused variable 'i'.
25919 Convert 'for' loop into 'while' loop.
25920 (arm_expand_prologue): Remove last unnecessary set of insn.
25921 (thumb_pop): Remove unused variable 'pushed_words'.
25922 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25924 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25926 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25927 * config/s390/s390-protos.h: Add prototype of
25928 s390_rel_address_ok_p.
25929 * config/s390/s390.c (s390_got_symbol): New function.
25930 (s390_rel_address_ok_p): New function.
25931 (legitimize_pic_address): Use s390_rel_address_ok_p.
25932 (s390_load_got): Use s390_got_symbol.
25933 (s390_option_override): Issue error if
25934 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25935 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25937 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25939 2017-06-27 Andrew Pinski <apinski@cavium.com>
25941 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25942 (X * copysign (1.0, X)): New pattern.
25943 (X * copysign (1.0, -X)): New pattern.
25944 (copysign (-1.0, CST)): New pattern.
25946 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25948 * genmultilib (combination_space): Remove variable.
25949 Validate reuse rules against regular expression for any sequence
25950 of multilib options in any order.
25952 2017-06-27 Michael Collison <michael.collison@arm.com>
25954 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25955 call aarch64_split_simd_combine.
25956 * (aarch64_combine_internal<mode>): Delete pattern.
25957 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25958 Allow register and subreg operands.
25960 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25962 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25963 specific need, just fallback on defaults.
25964 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25966 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25967 Olivier Hainque <hainque@adacore.com>
25969 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25971 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25972 targets. Pick a default if no particular attempt applied.
25973 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25976 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25978 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25979 (x86_64-wrs-vxworks7): Likewise.
25981 2017-06-27 Marek Polacek <polacek@redhat.com>
25984 * ubsan.c (instrument_null): Check get_base_address's result for null.
25986 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25988 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25990 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25992 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25993 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25994 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25995 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25996 New function types.
25997 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25998 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25999 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26000 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26001 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26002 BUILT_IN_FEUPDATEENV): New builtins.
26003 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26004 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26005 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26006 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26008 (builtin_structptr_types): Adjust size.
26009 * tree.c (builtin_structptr_types): Add four entries.
26011 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26012 Olivier Hainque <hainque@adacore.com>
26014 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26015 (TLS_SYM): New local macro, forcing reference to __tls__ on
26016 link command lines for VxWorks 7 RTPs, triggering initialization
26018 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26019 OS features TLS support, true for RTPs on VxWorks 7.
26020 * config/vxworks.c (vxworks_override_options): Setup emutls
26023 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26025 * predict.c (test_prediction_value_range): Use -1U instead of -1
26026 to avoid narrowing conversion warning.
26027 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26028 to avoid narrowing conversion warning.
26029 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26031 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26033 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26035 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26036 64bit configurations.
26037 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26038 (SIZE_TYPE): Likewise.
26039 * config/vxworks.c (vxworks_emutls_var_fields): Use
26040 long_unsigned_type_node instead of unsigned_type_node as the offset
26041 field type, which is "pointer" mode in emutls.c.
26043 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26046 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26048 PR middle-end/81207
26049 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26050 gimple_vuse copying separately from gimple_vdef copying.
26052 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26054 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26057 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26058 Olivier Hainque <hainque@adacore.com>
26060 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26061 for all vxworks7 targets.
26062 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26063 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26064 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26065 variations for VX6/VX7 and 32/64bits later on in ...
26066 (VXWORKS_LIB_SPEC): Leverage new macros.
26067 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26068 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26070 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26072 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26073 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26075 2017-06-26 Carl Love <cel@us.ibm.com>
26077 * config/rs6000/rs6000-c.c: Add support for built-in functions
26078 vector bool char vec_reve (vector bool char);
26079 vector signed char vec_reve (vector signed char);
26080 vector unsigned char vec_reve (vector unsigned char);
26081 vector bool int vec_reve (vector bool int);
26082 vector signed int vec_reve (vector signed int);
26083 vector unsigned int vec_reve (vector unsigned int);
26084 vector bool long long vec_reve (vector bool long long);
26085 vector signed long long vec_reve (vector signed long long);
26086 vector unsigned long long vec_reve (vector unsigned long long);
26087 vector bool short vec_reve (vector bool short);
26088 vector signed short vec_reve (vector signed short);
26089 vector double vec_reve (vector double);
26090 vector float vec_reve (vector float);
26091 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26092 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26093 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26094 (altivec_vreve): New pattern.
26095 * config/rs6000/altivec.h (vec_reve): New define.
26096 * doc/extend.texi (vec_rev): Update the built-in documentation file
26097 for the new built-in functions.
26099 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26101 PR tree-optimization/71815
26102 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26104 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26106 (slsr_process_phi): Likewise.
26107 (replace_uncond_cands_and_profitable_phis): Don't replace a
26108 multiply candidate with a stride of 1 (copy or cast).
26109 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26111 (lowest_cost_path): Likewise.
26112 (total_savings): Likewise.
26114 2017-06-26 Richard Biener <rguenther@suse.de>
26117 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26118 Use def_builtin_pure for all gather builtins.
26120 2017-06-26 Richard Biener <rguenther@suse.de>
26122 PR tree-optimization/81203
26123 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26124 non-dominating BBs.
26126 2017-06-26 Marek Polacek <polacek@redhat.com>
26129 * doc/invoke.texi: Document -Wmultistatement-macros.
26131 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26133 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26134 arm_neon_ok_no_float_abi effective target.
26136 2017-06-26 Richard Biener <rguenther@suse.de>
26138 PR tree-optimization/80928
26139 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26140 (copy_bbs): Set BB_DUPLICATED flag early.
26141 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26143 (execute_on_shrinking_pred): Likewise.
26144 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26145 BB_DUPLICATED blocks.
26146 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26147 iterate over all PHIs considering removal of *gsi.
26149 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26151 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26154 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26156 * config/rs6000/rs6000-string.c: (expand_block_clear,
26157 do_load_for_compare, select_block_compare_mode,
26158 compute_current_alignment, expand_block_compare,
26159 expand_strncmp_align_check, expand_strn_compare,
26160 expand_block_move, rs6000_output_load_multiple)
26161 Move functions related to string/block move/compare
26162 to a separate file.
26163 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26164 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26165 for this function which is now used in two files.
26166 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26167 * config.gcc: Add rs6000-string.o to extra_objs for
26168 targets powerpc*-*-* and rs6000*-*-*.
26170 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26173 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26174 32-bit, since indexed is not valid for DImode.
26175 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26176 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26177 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26178 (define_peephole2 for Altivec d-form store): Likewise.
26181 * multiple_target.c (create_dispatcher_calls): Only create the
26182 dispatcher call if the function is the default clone of a
26183 versioned function.
26185 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26187 PR middle-end/80902
26188 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26189 a call, force the call to not be a tail call.
26191 2017-06-23 Jeff Law <law@redhat.com>
26193 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26196 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26198 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26199 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26200 handling for early expansion of vector shifts (sl,sr,sra,rl).
26201 (builtin_function_type): Add vector shift right instructions
26202 to the unsigned argument list.
26204 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26206 rtl-optimizatoin/79286
26207 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26208 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26209 trap. PIC register plus a const unspec without offset can never trap.
26211 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26213 * tree.h (builtin_structptr_type): New type.
26214 (builtin_structptr_types): Declare new array.
26215 * tree.c (builtin_structptr_types): New array.
26216 (free_lang_data, build_common_tree_nodes): Use it.
26218 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26221 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26224 2017-06-22 Matt Turner <mattst88@gmail.com>
26226 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26227 Lake models to skylake case. Assume skylake for unknown
26228 models with clflushopt.
26230 2017-06-22 Jeff Law <law@redhat.com>
26232 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26233 frame sizes that do not satisfy aarch64_uimm12_shift.
26235 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26237 * profile-count.h (apply_probability,
26238 apply_scale, probability_in): Fix checks for zero.
26240 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26242 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26243 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26245 2016-06-22 Richard Biener <rguenther@suse.de>
26247 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26248 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26249 REDUC_MAX_EXPR support.
26250 (vectorizable_reduction): Likewise.
26251 (vect_create_epilog_for_reduction): Likewise.
26253 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26255 * match.pd (A / (1 << B) -> A >> B): New.
26256 * generic-match-head.c: Include optabs-tree.h.
26257 * gimple-match-head.c: Likewise.
26258 * optabs-tree.h (target_supports_op_p): New.
26259 * optabs-tree.c (target_supports_op_p): New.
26261 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26263 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26264 $gcc_cv_ld --help output.
26265 (gcc_cv_ld_demangle): Likewise.
26266 (gcc_cv_ld_eh_frame_hdr): Likewise.
26267 (gcc_cv_ld_pie): Likewise.
26268 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26269 (gcc_cv_ld_buildid): Likewise.
26270 (gcc_cv_ld_sysroot): Likewise.
26271 (ld_bndplt_support): Likewise.
26272 (ld_pushpopstate_support): Likewise.
26273 * configure: Regenerate.
26274 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26276 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26279 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26280 operands indexes to avoid gap between operands and match_dups.
26282 2017-06-21 Andrew Pinski <apinski@cavium.com>
26284 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26285 Increment Arith_shift and Arith_shift_reg by 1.
26286 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26288 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26289 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26290 (aarch64_strip_extend): Add new argument and test for it.
26291 (aarch64_cheap_mult_shift_p): New function.
26292 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26293 add a cost if it is true.
26294 Update calls to aarch64_strip_extend.
26295 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26297 2017-06-21 Andrew Pinski <apinski@cavium.com>
26299 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26301 (thunderxt88): Likewise.
26302 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26303 (thunderx_prefetch_tune): New variable.
26304 (thunderx2t99_prefetch_tune): Update for the correct values.
26305 (thunderxt88_tunings): New variable.
26306 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26307 generic_prefetch_tune.
26308 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26310 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26312 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26313 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26314 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26315 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26316 (aarch64_atomic_cas<mode>, GPI): Likewise.
26318 2017-06-21 Martin Liska <mliska@suse.cz>
26320 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26321 statements on cold and hot labels.
26322 * predict.c (tree_estimate_probability_bb): Remove the
26323 prediction from this place.
26325 2017-06-21 Martin Liska <mliska@suse.cz>
26327 PR tree-optimization/79489
26328 * gimplify.c (maybe_add_early_return_predict_stmt): New
26330 (gimplify_return_expr): Call the function.
26331 * predict.c (tree_estimate_probability_bb): Remove handling
26333 * predict.def: Update comment about early return predictor.
26334 * gimple-predict.h (is_gimple_predict): New function.
26335 * predict.def: Change default value of early return to 66.
26336 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26338 * passes.def: Put pass_strip_predict_hints to the beginning of
26341 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26343 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26344 FUNCTION_DECL declarations.
26345 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26347 (dwaf2out_decl): Likewise.
26348 * godump.c (go_early_global_decl): Skip call to the real debug hook
26349 for FUNCTION_DECL declarations.
26350 * passes.c (rest_of_decl_compilation): Skip call to the
26351 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26352 -fdump-go-spec is passed.
26354 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26356 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26357 Reorder for compactness.
26358 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26359 (def_builtin_pure, def_builtin_pure2): New functions.
26360 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26362 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26364 * match.pd (nop_convert): New predicate.
26365 ((A +- CST1) +- CST2): Allow some NOP conversions.
26367 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26370 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26371 with ctors/dtors if GOVD_SHARED is set.
26373 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26375 * config/aarch64/aarch64.md (movti_aarch64):
26376 Emit mov rather than orr.
26377 (movtf_aarch64): Likewise.
26378 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26379 Emit mov rather than orr.
26381 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26383 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26384 Swap alternatives, make integer dup more expensive.
26386 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26388 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26389 Return true for non-tls symbols.
26391 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26393 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26394 (cortex-a75): Likewise.
26395 (cortex-a75.cortex-a55): Likewise.
26396 * config/aarch64/aarch64-tune.md: Regenerate.
26397 * doc/invoke.texi (-mtune): Document new values for -mtune.
26399 2017-06-21 Tom de Vries <tom@codesourcery.com>
26401 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26402 stack_size feature.
26403 (Effective-Target Keywords, Other attributes): Suggest using
26404 dg-add-options stack_size feature to get stack limit in stack_size
26405 effective target documentation.
26407 2017-06-21 Julian Brown <julian@codesourcery.com>
26408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26410 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26411 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26412 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26414 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26415 attribute type list for neon_multiply.
26416 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26417 attribute type list for neon_multiply.
26418 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26419 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26420 attribute type list for neon_multiply.
26421 * config/arm/types.md (crypto_pmull): Add.
26422 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26423 attribute type list.
26425 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26427 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26430 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26432 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26433 to make sure not to dereference a NULL cost_classes_ptr pointer.
26435 2017-06-20 Carl Love <cel@us.ibm.com>
26437 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26438 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26439 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26440 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26441 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26442 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26443 VMULOSW): New enum "unspec" values.
26444 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26445 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26446 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26447 altivec_vmulosw): New patterns.
26448 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26449 VMULOSW): Add definitions.
26451 2017-06-20 Julia Koval <julia.koval@intel.com>
26453 * config/i386/i386.c: Fix rounding expand for new pattern.
26454 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26456 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26458 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26459 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26461 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26463 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26466 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26468 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26469 architecture, then by implementer ID.
26470 * config/aarch64/aarch64-tune.md: Regenerate.
26472 2017-06-20 Richard Biener <rguenther@suse.de>
26474 PR middle-end/81097
26475 * fold-const.c (split_tree): Fold to type before negating.
26477 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26479 * diagnostic-show-locus.c
26480 (selftest::test_fixit_deletion_affecting_newline): New function.
26481 (selftest::diagnostic_show_locus_c_tests): Call it.
26483 2017-06-20 Andreas Schwab <schwab@suse.de>
26486 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26489 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26491 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26492 __ARM_FEATURE_COPROC according to support.
26494 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26496 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26497 Rewritten to avoid overflow for > 32-bit pointers.
26500 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26501 by removing enum keyword.
26502 (ubsan_type_descriptor): Likewise. Formatting fix.
26505 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26506 splitter): Require TARGET_SSE2 in the condition.
26508 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26511 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26512 for doing vector set of SFmode on ISA 3.0.
26513 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26514 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26516 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26517 SFmode value into a V4SF variable that was extracted from another
26518 V4SF variable without converting the element to double precision
26519 and back to single precision vector format.
26520 (vsx_insert_extract_v4sf_p9_2): Likewise.
26522 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26524 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26525 in UWHI to avoid undefined overflow.
26528 * ubsan.h (enum ubsan_encode_value_phase): New.
26529 (ubsan_encode_value): Change second argument to
26530 enum ubsan_encode_value_phase with default value of
26531 UBSAN_ENCODE_VALUE_GENERIC.
26532 * ubsan.c (ubsan_encode_value): Change second argument to
26533 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26534 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26535 create_tmp_var_raw instead of create_tmp_var and use a
26537 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26538 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26539 ubsan_encode_value callers.
26542 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26543 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26544 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26546 2017-06-19 Richard Biener <rguenther@suse.de>
26548 PR middle-end/81118
26549 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26550 estimates if we changed anything.
26552 2017-06-19 Richard Biener <rguenther@suse.de>
26554 PR tree-optimization/80887
26555 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26556 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26557 simplified lookups, then reset mprts_hook.
26558 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26560 (try_to_simplify): Likewise.
26562 2017-06-19 Martin Liska <mliska@suse.cz>
26565 * gimplify.c (gimplify_switch_expr):
26566 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26568 2017-06-19 Martin Liska <mliska@suse.cz>
26570 * doc/install.texi: Document that PGO runs in 4 stages.
26572 2017-06-19 Martin Liska <mliska@suse.cz>
26575 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26576 to dispatcher function name.
26577 * multiple_target.c (replace_function_decl): New function.
26578 (create_dispatcher_calls): Redirect both edges and references.
26580 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26582 * profile-count.c (profile_count::dump): Dump quality.
26583 (profile_count::differs_from_p): Update for unsigned val.
26584 * profile-count.h (profile_count_quality): New enum.
26585 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26587 2017-06-19 Richard Biener <rguenther@suse.de>
26589 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26590 struct function as arg.
26591 (estimate_numbers_of_iterations): Export overload with loop arg.
26592 (free_numbers_of_iterations_estimates_loop): Use an overload of
26593 free_numbers_of_iterations_estimates instead.
26594 * tree-cfg.c (remove_bb): Adjust.
26595 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26596 * tree-parloops.c (gen_parallel_loop): Likewise.
26597 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26599 (tree_unroll_loops_completely): Likewise.
26600 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26601 Use an overload instead and export.
26602 (estimated_loop_iterations): Adjust.
26603 (max_loop_iterations): Likewise.
26604 (likely_max_loop_iterations): Likewise.
26605 (estimate_numbers_of_iterations): Take struct function as arg
26607 (loop_exits_before_overflow): Adjust.
26608 (free_numbers_of_iterations_estimates_loop): Use an overload.
26609 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26610 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26612 2017-06-19 Richard Biener <rguenther@suse.de>
26615 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26616 RANGE_EXPR conservatively.
26618 2017-06-16 Carl Love <cel@us.ibm.com>
26620 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26621 definitions for vec_float, vec_float2, vec_floato,
26622 vec_floate built-ins.
26623 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26624 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26626 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26627 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26628 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26629 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26630 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26631 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26632 vec_floato): Add builtin defines.
26633 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26634 Update the built-in documentation file for the new built-in
26637 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26639 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26640 (mthumb): Mark as the negative of -marm.
26642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26644 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26646 (ARM Options, -mtune): Document that this accepts the same
26647 extension options as -mcpu.
26648 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26652 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26653 permitted extensions.
26655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26657 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26658 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26659 (armv8-m.main): Add option +nodsp.
26660 * config/arm/arm-cpu-cdata.h: Regenerated.
26662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26664 * config/arm/t-fuchsia: New file.
26665 * config.gcc (arm*-*-fuchsia*): Use it.
26667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26669 * config/arm/t-symbian: Rewrite for new option infrastructure.
26671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26673 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26674 (MULTILIB_REQUIRED): Likewise.
26676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26678 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26679 (MULTILIB_RESUE): Likewise.
26680 (MULTILIB_MATCHES): Likewise.
26681 (MULTLIB_REQUIRED): Likewise.
26683 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26685 * config/arm/t-rtems: Rewrite for new option framework.
26687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26689 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26690 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26691 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26692 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26693 * config/arm/t-multilib: ... here.
26694 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26695 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26696 armv7-a and armv8*-a when A-profile libraries have not been built.
26697 * config/arm/t-rmprofile: Rewrite.
26699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26701 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26702 with a backslash. Remove the backslash after substituting unescaped
26704 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26708 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26709 the driver to pass through the default setting of -mfloat-abi.
26710 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26712 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26713 (all_feat_combs): New rule.
26714 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26716 * config/arm/t-aprofile: Rewrite.
26718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26720 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26721 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26722 fpu is not specified by the user/command-line.
26723 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26724 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26725 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26726 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26727 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26728 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26732 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26733 * config/arm/t-arm-elf: Rewritten.
26735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26737 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26738 have some floating-point instructions.
26739 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26740 (TARGET_MAYBE_HARD_FLOAT): New macro.
26741 * config/arm/arm-builtins.c (arm_init_builtins): Use
26742 TARGET_MAYBE_HARD_FLOAT.
26743 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26747 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26748 (configargs.h): Include it.
26749 (arm_print_hint_for_fpu_option): New function.
26750 (arm_parse_fpu_option): New function.
26751 (candidate_extension): New class.
26752 (arm_canon_for_multilib): New function.
26753 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26754 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26755 (ARCH_CANONICAL_SPECS): New macro.
26756 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26760 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26761 are set after handling multilib fragments. Set target_cpu_default2
26764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26766 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26768 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26769 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26771 (check_arch): Likewise.
26772 * config/arm/arm.c (arm_configure_build_target): Handle
26773 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26774 options in the default.
26776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26778 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26779 when an option is an alias of another.
26780 * config/arm/parsecpu.awk (optalias): New parser token.
26781 (gen_comm_data): Mark non-alias options as such. Emit entries
26782 for extension aliases.
26783 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26784 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26785 (armv6kz, armv6zk, armv6t2): Likewise.
26786 (armv7): Make vfpv3-d16 an alias.
26787 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26789 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26790 Sort in canonical order.
26791 (armv8-a): Sort in canonical order.
26792 (armv8.1-a, armv8.2-a): Likewise.
26793 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26795 (cortex-a9): Sort in canonical order.
26796 * config/arm/arm.c (selftests.h): Include it.
26797 (arm_test_cpu_arch_data): New function.
26798 (arm_run_self_tests): New function.
26799 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26800 (targetm): Move declaration to the end of the file.
26801 * arm-cpu-cdata.h: Regenerated.
26803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26805 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26806 call to target_mode_check describing the type of option passed.
26807 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26808 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26809 arm_parse_cpu_option_name to match parameters against list of
26811 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26812 arm_arch_core_flags data structure.
26813 * config/arm/arm-cpu_cdata.h: Regenerated.
26815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26817 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26819 (arm_print_hint_for_cpu_option): Likewise.
26820 (arm_print_hint_for_arch_option): Likewise.
26821 (arm_parse_cpu_option_name): Likewise.
26822 (arm_parse_arch_option_name): Likewise.
26823 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26824 of entries in the all_fpus list.
26825 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26826 (arm_parse_cpu_option_name): Declare.
26827 (arm_parse_arch_option_name): Declare.
26828 (arm_parse_option_features): Declare.
26829 (arm_intialize_isa): Declare.
26830 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26832 (gen_comm_data): ... here. Make definitions non-static.
26833 * config/arm/arm-cpu-data.h: Regenerated.
26834 * config/arm/arm-cpu-cdata.h: Regenerated.
26836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26838 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26839 (cpu_arch_extension): New structure.
26840 (cpu_arch_option, arch_option, cpu_option): New structures.
26841 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26842 architecture types.
26843 (gen_data): Generate new format data tables.
26844 * config/arm/arm.c (cpu_tune): New structure.
26845 (cpu_option, processors): Delete.
26846 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26847 (arm_print_hint_for_cpu_option): ... this and ...
26848 (arm_print_hint_for_arch_option): ... this.
26849 (arm_parse_arch_cpu_name): Delete. Replace with ...
26850 (arm_parse_cpu_option_name): ... this and ...
26851 (arm_parse_arch_option_name): ... this.
26852 (arm_unrecognized_feature): Change type of target parameter to
26854 (arm_parse_arch_cpu_features): Delete. Replace with ...
26855 (arm_parse_option_features): ... this.
26856 (arm_configure_build_target): Rework to use new configuration data
26858 (arm_print_tune_info): Rework for new configuration data tables.
26859 * config/arm/arm-cpu-data.h: Regenerated.
26860 * config/arm/arm-cpu.h: Regenerated.
26862 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26864 * Makefile.in (OBJS): Move sbitmap.o from here ...
26865 (OBJS-libcommon): ... to here.
26867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26869 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26870 (ISA_ALL_CRYPTO): New macro.
26871 (ISA_ALL_SIMD): New macro
26872 (ISA_ALL_FP): New macro.
26873 * config/arm/arm.c (fpu_bitlist): Update initializer.
26874 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26876 (arm9e): Add fpu. Add option for nofp
26877 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26878 (arm926ej-s, arm1026ej-s): Likewise.
26879 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26880 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26881 neon-fp16, neon-vfpv4, nofp and nosimd.
26882 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26883 (cortex-a8): Add fpu. Add option for nofp.
26884 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26885 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26886 (cortex-r4f): Add fpu.
26887 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26888 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26890 (cortex-r8): Likewise.
26891 (cortex-m4): Add fpu. Add option for nofp.
26892 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26893 (cortex-a17.cortex-a7): Likewise.
26894 (cortex-a32): Add fpu. Add options for crypto and nofp.
26895 (cortex-a35, cortex-a53): Likewise.
26896 (cortex-a57): Add fpu. Add option for crypto.
26897 (cortex-a72, cortex-a73): Likewise.
26898 (exynos-m1): Likewise.
26899 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26900 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26901 (cortex-m33): Add fpu. Add option for nofp.
26902 * config/arm/arm-cpu-cdata.h: Regenerated
26903 * config/arm/arm-cpu-data.h: Regenerated.
26905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26907 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26908 (armv5te, armv5tej): Likewise.
26909 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26910 (armv7): Add options fp and vfpv3-d16.
26911 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26912 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26914 (armv7ve): Likewise.
26915 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26916 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26917 (armv8-a): Add nocrypto option.
26918 (armv8.1-a, armv8.2-a): Likewise.
26919 (armv8-m.main): add options fp, fp.dp and nofp.
26921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26923 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26925 (armv8-a+crc): Delete.
26926 (armv8.1-a): Add options simd, crypto and nofp.
26927 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26928 (armv8.2-a+fp16): Delete.
26929 (armv8-m.main): Add option dsp.
26930 (armv8-m.main+dsp): Delete.
26931 (cortex-a8): Add fpu. Add nofp option.
26932 (cortex-a9): Add fpu. Add nofp and nosimd options.
26933 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26934 link to main cpu and architecture data structures.
26935 (gen_comm_data): Only put isa attributes from the main architecture
26937 (option): New statement for architecture and CPU entries.
26938 * arm.c (struct cpu_option): New structure.
26939 (struct processors): Add entry for options.
26940 (arm_unrecognized_feature): New function.
26941 (arm_parse_arch_cpu_name): Ignore any characters after the first
26943 (arm_parse_arch_cpu_feature): New function.
26944 (arm_configure_build_target): Separate out any CPU and architecture
26945 features and parse separately. Don't error out if -mfpu=auto is
26946 used with only an architecture string.
26947 (arm_print_asm_arch_directives): New function.
26948 (arm_file_start): Call it.
26949 * config/arm/arm-cpu-cdata.h: Regenerated.
26950 * config/arm/arm-cpu-data.h: Likewise.
26951 * config/arm/arm-tables.opt: Likewise.
26953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26955 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26956 assembler when it is not -mfpu=auto.
26958 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26960 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26961 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26962 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26963 (ASM_CPU_SPEC): Rewrite.
26964 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26965 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26966 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26967 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26968 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26969 copied string is NUL-terminated. Also strip any characters prefixed
26971 (arm_rewrite_selected_arch): New function.
26972 (arm_rewrite_march): New function.
26974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26976 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26977 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26978 (march, mcpu, mtune): Convert to string-based options.
26979 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26980 (arm_parse_arch_cpu_name): New function.
26981 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26982 identify selected architecture or CPU.
26983 (arm_option_save): New function.
26984 (TARGET_OPTION_SAVE): Redefine.
26985 (arm_option_restore): Restore string options.
26986 (arm_option_print): Print string options.
26988 2017-06-16 Martin Sebor <msebor@redhat.com>
26990 PR tree-optimization/80933
26991 PR tree-optimization/80934
26992 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26993 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26994 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26995 (gimple_fold_builtin): Call them.
26997 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26999 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27000 as unlikely; update profile.
27002 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27004 * predict.c (force_edge_cold): Handle declaring edges impossible
27007 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27009 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27011 (try_unroll_loop_completely): Fix reporting.
27013 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27015 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27017 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27020 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27021 if given a non-constant argument for an intrinsic which requires a
27024 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27026 * profile.c (compare_freqs): New function.
27027 (branch_prob): Sort edge list.
27028 (find_spanning_tree): Assume that the list is priority sorted.
27030 2017-06-16 Richard Biener <rguenther@suse.de>
27032 PR tree-optimization/81090
27033 * passes.def (pass_record_bounds): Remove.
27034 * tree-pass.h (make_pass_record_bounds): Likewise.
27035 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27036 make_pass_record_bounds): Likewise.
27037 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27038 not free niter estimates at the beginning but at the end.
27039 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27041 2017-06-16 Richard Biener <rguenther@suse.de>
27043 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27044 initializer to workaround ICE in host GCC 4.8.
27046 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27048 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27050 (clone_inlined_nodes): Update.
27052 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27054 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27055 prefetch settings, and enable prefetching by default at -O3.
27057 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27059 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27060 Set flag_prefetch_loop_arrays according to tuning data.
27062 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27064 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27065 New tune structure.
27066 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27067 [Unrelated to main purpose of the patch] Place the pointer field last
27068 to enable type checking errors when tune structure are wrongly merged.
27069 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27070 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27071 (thunderx2t99_prefetch_tune): New tune constants.
27072 (tune_params *_tunings): Update all tunings (no functional change).
27073 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27074 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27075 from tunings structures.
27077 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27080 * ubsan.c (instrument_null): Add T argument, use it instead
27081 of computing it based on IS_LHS.
27082 (instrument_object_size): Likewise.
27083 (pass_ubsan::execute): Adjust instrument_null and
27084 instrument_object_size callers to pass gimple_get_lhs or
27085 gimple_assign_rhs1 result to it. Use instrument_null instead of
27086 calling get_base_address and instrument_mem_ref. Handle
27087 aggregate call arguments for object-size sanitization.
27089 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27091 PR tree-optimization/81089
27092 * tree-vrp.c (is_masked_range_test): Validate operands of
27095 2017-06-15 Martin Sebor <msebor@redhat.com>
27098 * dumpfile.c (dump_register): Avoid calling memset to initialize
27099 a class with a default ctor.
27100 * gcc.c (struct compiler): Remove const qualification.
27101 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27102 * hash-table.h: Ditto.
27103 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27105 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27106 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27108 * params.h (struct param_info): Make struct members non-const.
27109 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27110 with default initialization.
27111 * vec.h (vec_copy_construct, vec_default_construct): New helper
27113 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27114 with vec_copy_construct.
27115 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27116 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27117 * doc/invoke.texi (-Wclass-memaccess): Document.
27119 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27121 * emit-rtl.h (is_leaf): Update comment about local
27122 register allocator.
27124 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27127 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27128 for a variable to have a section before checking if the section has a
27130 Set section to.persistent if persistent attribute is set.
27131 Warn if .persistent attribute is used on an automatic variable.
27133 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27135 PR rtl-optimization/80474
27136 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27138 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27140 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27143 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27145 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27146 (casesi): Emit bounds checking as RTL.
27147 (casesi_internal_mips16_<mode>): Remove bounds checking.
27149 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27151 * config/xtensa/xtensa.c (xtensa_option_override): Append
27152 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27153 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27154 xtensa_doloop_hooks): Define unconditionally.
27155 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27157 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27158 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27159 for it in xtensa_option_override.
27160 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27161 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27163 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27165 * doc/cppopts.texi: Document '-' special value to -MF.
27167 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27169 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27170 (cortex_a53_fconst): Likewise.
27171 (cortex_a53_fpmul): Likewise.
27172 (cortex_a53_f_load_64): Likewise.
27173 (cortex_a53_f_load_many): Likewise.
27174 (cortex_a53_advsimd_alu): Likewise.
27175 (cortex_a53_advsimd_alu_q): Likewise.
27176 (cortex_a53_advsimd_mul): Likewise.
27177 (cortex_a53_advsimd_mul_q): Likewise.
27178 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27179 Add missing fmul, r2f_cvt and fconst cases.
27181 2017-06-14 Richard Biener <rguenther@suse.de>
27183 PR middle-end/81088
27184 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27186 (fold_binary_loc): When associating do not treat pre-existing
27187 TREE_OVERFLOW on literal constants as a reason to allow
27188 TREE_OVERFLOW on associated literal constants.
27190 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27192 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27193 (MASK_FEATURES): New macro.
27194 * config/sparc/sparc.c (sparc_option_override): Remove the special
27195 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27197 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27199 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27200 a division of 0 if non-call exceptions are enabled.
27202 2017-06-14 Andrew Pinski <apinski@cavium.com>
27203 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27206 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27207 Improve vector initialization code gen for only variable case.
27209 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27211 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27213 2017-06-14 Richard Biener <rguenther@suse.de>
27215 PR tree-optimization/81083
27216 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27219 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27221 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27222 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27223 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27224 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27225 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27226 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27228 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27230 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27231 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27233 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27235 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27237 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27239 * config/rs6000/t-rtems: Don't handle SPE.
27241 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27243 * config/rs6000/t-linux: Don't handle SPE.
27245 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27247 * config/rs6000/eabispe.h: Delete file.
27249 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27251 * config/rs6000/t-spe: Delete file.
27253 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27255 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27256 (rs6000_legitimate_offset_address_p): Return false for anything in
27257 V2SImode or V2SFmode.
27259 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27261 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27262 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27263 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27265 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27266 (rs6000_legitimate_offset_address_p): Ditto.
27267 (rs6000_emit_move): Ditto.
27268 (rs6000_init_builtins): Remove V4HI_type_node.
27270 2017-06-13 Martin Liska <mliska@suse.cz>
27273 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27274 (gate_asan): Likewise.
27275 * asan.h (asan_no_sanitize_address_p): Remove the function.
27276 (sanitize_flags_p): New function.
27277 * builtins.def: Fix coding style.
27278 * common.opt: Use renamed enum value.
27279 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27280 * doc/extend.texi: Document no_sanitize attribute.
27281 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27282 to SANITIZE_UNDEFINED_NONDEFAULT.
27283 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27284 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27285 Use sanitize_flags_p.
27286 * gimplify.c (gimplify_function_tree): Likewise.
27287 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27288 * opts.c (parse_no_sanitize_attribute): New function.
27289 (common_handle_option): Use renamed enum value.
27290 * opts.h (parse_no_sanitize_attribute): Declare.
27291 * tree.c (sanitize_flags_p): New function.
27292 * tree.h: Declared here.
27293 * tsan.c: Use sanitize_flags_p.
27294 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27295 (instrument_mem_ref): Likewise.
27296 (instrument_bool_enum_load): Likewise.
27297 (do_ubsan_in_current_function): Remove the function.
27298 (pass_ubsan::execute): Use sanitize_flags_p.
27299 * ubsan.h: Remove do_ubsan_in_current_function
27300 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27301 (dump_function_to_file): Use it here.
27303 2017-06-13 Martin Jambor <mjambor@suse.cz>
27305 PR tree-optimization/80803
27306 PR tree-optimization/81063
27307 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27308 (propagate_subaccesses_across_link): Enqueue subtree whenever
27309 necessary instead of relying on the caller.
27311 2017-06-13 Martin Jambor <mjambor@suse.cz>
27313 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27314 that have a first_link.
27315 (sort_and_splice_var_accesses): Do not check first_link before
27317 (subtree_mark_written_and_enqueue): Likewise.
27318 (propagate_all_subaccesses): Likewise and do not stop at first
27319 parent with a first_link.
27321 2017-06-13 Martin Jambor <mjambor@suse.cz>
27323 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27326 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27328 * match.pd: New pattern.
27330 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27332 * tree-vrp.c (is_masked_range_test): New function.
27333 (register_edge_assert_for): Determine ranges for
27336 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27338 PR tree-optimization/67328
27339 * fold-const.c (maskable_range_p): New function.
27340 (build_range_check): Generate bittests if possible.
27342 2017-06-13 Martin Liska <mliska@suse.cz>
27344 * gimple-pretty-print.c (dump_probability): Add new argument.
27345 (dump_edge_probability): Dump both probability and count.
27346 (dump_gimple_label): Likewise.
27347 (dump_gimple_bb_header): Likewise.
27349 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27352 * config/avr/avr-devices.c: Fix indentation.
27353 * config/avr/gen-avr-mmcu-specs.c: Dito.
27355 2017-06-13 Richard Biener <rguenther@suse.de>
27357 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27358 instead get vector type from stmt_info.
27359 (vectorizable_reduction): Adjust. Remove dead code.
27361 2017-06-13 Richard Biener <rguenther@suse.de>
27363 PR middle-end/81065
27364 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27365 case of C * (x * C2 + C3).
27366 (fold_addr_of_array_ref_difference): Properly fold index difference.
27368 2017-06-12 David S. Miller <davem@davemloft.net>
27371 * config/sparc/sparc.md (return expander): Emit frame blockage if
27372 function uses alloca.
27374 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27376 * combine.c (make_field_assignment): Check len rather than the mode
27377 precision when calling force_to_mode.
27379 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27381 Support multilibs and devices that see flash in RAM address range.
27384 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27385 (avr_mcu_t) <flash_pm_offset>: New field.
27386 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27387 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27388 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27389 (AVR_TINY_PM_OFFSET): Remove macro.
27390 * config/avr/avr.opt (-mshort-calls): New option.
27391 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27392 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27393 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27394 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27395 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27396 instead of avr_arch->have_jmp_call.
27397 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27398 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27399 avr_arch->flash_pm_offset to define.
27400 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27401 new field flash_pm_offset. Add entry for avrxmega3.
27402 (avr_texinfo): Add entry for avrxmega3.
27403 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27404 attiny212, attiny214,
27405 attiny412, attiny414, attiny416, attiny417,
27406 attiny814, attiny816, attiny817,
27407 attiny1614, attiny1616, attiny1617,
27408 attiny3214, attiny3216, attiny3217.
27409 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27410 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27411 (avr_print_operand_address) [AVR_TINY]: Same.
27412 (avr_asm_init_sections) <readonly_data_section>: Only patch
27413 callback if avr_arch->flash_pm_offset = 0.
27414 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27415 for rodata if avr_arch->flash_pm_offset != 0.
27416 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27417 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27418 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27419 (m_options): Append opt_rcall.
27420 (m_dirnames): Append dir_rcall.
27421 * config/avr/t-multilib: Regenerate.
27423 * configure.ac [target=avr]: Check whether avrxmega3 default
27424 linker description file works as needed.
27425 * configure: Regenerate.
27426 * doc/avr-mmcu.texi: Regenerate.
27427 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27428 <__AVR_ARCH__>: Document avrxmega3 and 103.
27429 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27430 <__AVR_SHORT_CALLS__>: Document it.
27431 <__AVR_PM_BASE_ADDRESS__>: Document it.
27432 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27433 (AVR Variable Attributes) <progmem>: Document this is
27434 not needed for avrxmega3.
27435 (AVR Named Address Spaces) <__flash>: Dito.
27437 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27439 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27441 2017-06-12 Doug Rupp <rupp@adacore.com>
27443 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27444 Append vxworks-stdint.h to the tm_file list.
27445 * config/vxworks-stdint.h: New file.
27447 2017-06-12 Martin Liska <mliska@suse.cz>
27449 PR tree-optimization/81041
27450 * tree-profile.c (gimple_gen_ic_func_profiler):
27451 Create an extra BB in profile-generate
27452 (gimple_gen_time_profiler): Likewise.
27454 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27456 PR tree-optimization/81003
27457 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27458 (update_range_test): Use it instead of force_gimple_operand_gsi.
27460 2017-06-12 Richard Biener <rguenther@suse.de>
27462 PR tree-optimization/81053
27463 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27464 with backedge value not defined in loop. Simplify def stmt
27467 2017-06-11 Tom de Vries <tom@codesourcery.com>
27470 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27472 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27473 nvptx_cannot_force_const_mem.
27475 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27477 * opts.c (finish_options): Move test for flag_split_stack after
27478 it has been initialized.
27480 2017-06-11 Jason Merrill <jason@redhat.com>
27482 * tree.h (id_equal): New.
27483 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27484 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27485 instead of strcmp of IDENTIFIER_POINTER.
27487 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27489 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27490 (mark_all_inlined_calls_cdtor): Fix formating.
27491 (inline_transform): Rescale profile before inlining.
27493 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27495 * cgraph.h (cgraph_edge::clone): Update prototype.
27496 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27497 (cgraph_node::create_clone): Update.
27498 (cgraph_node::create_version_clone): Update.
27499 * tree-inline.c (copy_bb): Update.
27500 (expand_call_inline): Update.
27502 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27504 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27505 factored out from ...
27506 (rs6000_emit_prologue): ... here.
27508 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27510 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27511 factored out from ...
27512 (rs6000_emit_prologue): ... here.
27514 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27516 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27518 (handle_missing_profiles): Fix computation of tp_first_run.
27519 (counts_to_freqs): Do not touch freqs when count is 0.
27521 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27523 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27526 2017-06-10 Tom de Vries <tom@codesourcery.com>
27528 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27529 attributes): Document signal effective target.
27531 2017-06-10 Tom de Vries <tom@codesourcery.com>
27533 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27534 Document effective target stack_size.
27536 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27538 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27539 to the edit_context if they can be auto-applied.
27541 2017-06-9 Ian Lance Taylor <iant@golang.org>
27543 * opts.c (finish_options): If -fsplit-stack, disable implicit
27544 -forder-blocks-and-partition.
27545 * doc/invoke.texi (Optimize Options): Document that when using
27546 -fsplit-stack -forder-blocks-and-partition is not implicitly
27549 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27551 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27552 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27553 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27554 * builtins.def (abort, trap, unreachable): Declare cold.
27555 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27556 * tree-core.h (ECF_COLD): New.
27557 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27558 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27560 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27562 * predict.c (unlikely_executed_stmt_p): Cleanup.
27564 2017-06-09 Richard Biener <rguenther@suse.de>
27566 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27567 model if the ref is always written to.
27569 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27571 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27573 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27575 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27578 2017-06-09 Tom de Vries <tom@codesourcery.com>
27581 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27582 "target cannot support label values" when encountering LABEL_REF.
27584 2017-06-09 Martin Liska <mliska@suse.cz>
27586 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27587 (gimple_gen_ic_func_profiler): Emit direct comparison
27588 of __gcov_indirect_call_callee with NULL.
27589 (gimple_gen_time_profiler): Change probability from
27590 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27592 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27594 * profile.c (edge_gcov_counts): Turn to pointer.
27595 (compute_branch_probabilities, compute_branch_probabilities): Update.
27596 (branch_prob): Do not clear edge_gcov_count.
27597 * profile.h (edge_gcov_counts): Turn to pointer.
27598 (edge_gcov_count): Update.
27600 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27602 * gimple.h (gimple_check_failed): Mark cold.
27604 2017-06-09 Richard Biener <rguenther@suse.de>
27606 PR tree-optimization/66623
27607 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27608 refactor check_reduction into two parts, properly computing
27609 whether we have to check reduction validity for outer loop
27612 2017-06-09 Richard Biener <rguenther@suse.de>
27614 PR tree-optimization/79483
27615 * graphite-scop-detection.c (order): New global.
27616 (get_order): Compute bb to order mapping that satisfies code
27617 generation constraints.
27618 (cmp_pbbs): New helper.
27619 (build_scops): Start domwalk at entry block, sort generated
27622 2017-06-09 Richard Biener <rguenther@suse.de>
27624 PR middle-end/81007
27625 * ipa-polymorphic-call.c
27626 (ipa_polymorphic_call_context::restrict_to_inner_class):
27627 Skip FIELD_DECLs with error_mark_node type.
27628 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27631 2017-06-09 Martin Liska <mliska@suse.cz>
27633 * predict.c (struct branch_predictor): New struct.
27634 (test_prediction_value_range): New test.
27635 (predict_c_tests): New function.
27636 * selftest-run-tests.c (selftest::run_tests): Run the function.
27637 * selftest.h: Declare new tests.
27639 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27642 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27643 gen_add3_insn did not fail.
27644 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27645 r0, construct that number in a temporary reg and add that reg to r0.
27646 If asked to put the result in r0 as well, fail.
27648 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27650 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27651 for early expansion of vec_eqv.
27653 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27655 PR middle-end/81005
27656 * ubsan.c (instrument_null): Avoid pointless code temporary.
27657 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27659 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27663 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27666 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27669 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27671 * predict.c (unlikely_executed_edge_p): Move ahead.
27672 (probably_never_executed_edge_p): Use it.
27674 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27676 PR middle-end/79988
27677 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27678 gimple_call_builtin_p call.
27680 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27682 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27683 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27684 rtl_check_failed_type2, rtl_check_failed_code1,
27685 rtl_check_failed_code2, rtl_check_failed_code_mode,
27686 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27687 rtvec_check_failed_bounds, rtl_check_failed_flag,
27688 _fatal_insn_not_found, _fatal_insn): Likewise.
27689 * tree.h (tree_contains_struct_check_failed,
27690 tree_check_failed, tree_not_check_failed,
27691 tree_class_check_failed, tree_range_check_failed,
27692 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27693 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27694 tree_operand_check_failed, omp_clause_check_failed,
27695 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27698 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27700 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27701 flag_branch_probabilities.
27702 * ipa-inline.c (edge_badness): Likewise.
27703 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27704 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27705 * predict.c (maybe_hot_frequency_p): Likewise.
27706 (probably_never_executed): Likewise.
27707 * sched-ebb.c (schedule_ebbs): Likewise.
27708 * sched-rgn.c (find_single_block_region): Likewise.
27709 * tracer.c (tail_duplicate): Likewise.
27711 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27713 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27714 longer requires x_flag_profile_use.
27716 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27718 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27719 instead of flag_reorder_blocks_and_partition.
27720 * dbxout.c (dbxout_function_end): Likewise.
27721 * dwarf2out.c (gen_subprogram_die): Likewise.
27722 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27723 * hw-doloop.c (reorg_loops): Likewise.
27724 * varasm.c (assemble_start_function,
27725 assemble_end_function): Likewise.
27726 (decide_function_section): Do not check for
27727 flag_reorder_blocks_and_partition.
27729 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27731 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27733 (chkp_get_hard_register_fake_addr_expr): Ditto.
27734 (chkp_build_addr_expr): Add check for hard reg case.
27735 (chkp_parse_array_and_component_ref): Ditto.
27736 (chkp_find_bounds_1): Ditto.
27737 (chkp_process_stmt): Don't generate bounds store for
27740 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27742 * predict.c (maybe_hot_bb_p): Do not check profile status.
27743 (maybe_hot_edge_p): Likewise.
27744 (probably_never_executed): Check for zero counts even if profile
27746 (unlikely_executed_edge_p): New function.
27747 (unlikely_executed_stmt_p): New function.
27748 (unlikely_executed_bb_p): New function.
27749 (set_even_probabilities): Use unlikely predicates.
27750 (combine_predictions_for_bb): Likewise.
27751 (predict_paths_for_bb): Likewise.
27752 (predict_paths_leading_to_edge): Likewise.
27753 (determine_unlikely_bbs): New function.
27754 (estimate_bb_frequencies): Use it.
27755 (compute_function_frequency): Use zero counts even if profile is
27757 * profile-count.h: Fix typo.
27759 2017-08-08 Julia Koval <julia.koval@intel.com>
27761 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27762 _mm512_mask_cvtsepi16_storeu_epi8,
27763 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27764 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27765 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27766 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27767 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27768 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27769 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27770 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27771 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27772 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27773 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27774 __builtin_ia32_pmovuswb256mem_mask,
27775 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27776 __builtin_ia32_pmovwb512mem_mask): New builtins.
27778 2017-08-08 Julia Koval <julia.koval@intel.com>
27780 PR target/73350,80862
27781 * config/i386/subst.md (round): Fix round pattern.
27782 * config/i386/i386.c (ix86_erase_embedded_rounding):
27783 Fix erasing rounding for the fixed pattern.
27785 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27787 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27789 2017-06-08 Martin Liska <mliska@suse.cz>
27791 PR gcov-profile/80911
27792 * gcov.c (block_info::block_info): New constructor.
27794 2017-06-07 Carl Love <cel@us.ibm.com>
27796 * config/rs6000/rs6000-c: The return type of the following
27797 built-in functions was implemented as int not long long. Fix sign
27798 of return value for the unsigned version of vec_mulo and vec_mule.
27799 vector unsigned long long vec_bperm (vector unsigned long long,
27800 vector unsigned char)
27801 vector signed long long vec_mule (vector signed int,
27803 vector unsigned long long vec_mule (vector unsigned int,
27804 vector unsigned int)
27805 vector signed long long vec_mulo (vector signed int,
27807 vector unsigned long long vec_mulo (vector unsigned int,
27808 vector unsigned int)
27809 * doc/extend.texi: Fix the documentation for the built-in
27812 2017-06-07 Carl Love <cel@us.ibm.com>
27815 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27818 2017-06-07 Carl Love <cel@us.ibm.com>
27820 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27821 support, Generate doublehv for signed int/float for BE case only.
27823 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27825 * doc/invoke.texi (mcx16): Rewrite.
27827 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27829 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27830 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27831 *mov<mode>_softfloat, and an anonymous splitter): Use
27832 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27834 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27836 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27838 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27839 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27840 (rs6000_debug_reg_global): Adjust.
27841 (rs6000_init_hard_regno_mode_ok): Adjust.
27842 (rs6000_dbx_register_number): Adjust.
27843 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27844 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27845 Remove SPE_ACC and SPEFSCR.
27846 (REG_ALLOC_ORDER): Ditto.
27847 (FRAME_POINTER_REGNUM): Change to 111.
27848 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27849 (REG_CLASS_NAMES): Ditto.
27850 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27851 (REGISTER_NAMES): Ditto.
27852 (ADDITIONAL_REG_NAMES): Ditto.
27853 (rs6000_reg_names): Ditto.
27854 * config/rs6000/rs6000.md: Renumber some register number
27857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27859 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27861 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27862 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27864 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27865 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27866 Delete the SPE high registers.
27867 (REG_ALLOC_ORDER): Ditto.
27868 (enum reg_class): Remove SPE_HIGH_REGS.
27869 (REG_CLASS_NAMES): Ditto.
27870 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27871 (REGISTER_NAMES): Ditto.
27872 (rs6000_reg_names): Ditto.
27873 * doc/tm.texi.in: Remove SPE as example.
27874 * doc/tm.texi: Regenerate.
27876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27878 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27879 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27880 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27881 * config/rs6000/rs6000.md (type): Remove "brinc".
27883 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27885 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27886 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27887 * config/rs6000/linuxspe.h: Delete file.
27888 * config/rs6000/rs6000.md: Don't include spe.md.
27889 * config/rs6000/spe.h: Delete file.
27890 * config/rs6000/spe.md: Delete file.
27891 * config/rs6000/t-rs6000: Remove spe.md.
27893 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27895 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27896 (reg_or_none500mem_operand): Delete.
27897 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27898 instead of reg_or_none500mem_operand.
27900 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27902 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27903 handling of SPE flags.
27904 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27906 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27908 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27910 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27911 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27912 paired_divv2sf3): Similar.
27913 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27914 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27915 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27917 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27918 Rename the paired_* instruction patterns.
27919 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27921 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27922 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27923 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27924 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27925 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27926 PAIRED_VECTOR_MODE.
27927 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27928 (spe_func_has_64bit_regs_p): Delete.
27929 (spe_expand_predicate_builtin): Delete.
27930 (spe_expand_evsel_builtin): Delete.
27931 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27932 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27933 (invalid_e500_subreg): Delete.
27934 (rs6000_legitimize_address): Always force_reg op2 as well, for
27935 paired single memory accesses.
27936 (rs6000_member_type_forces_blk): Delete.
27937 (rs6000_spe_function_arg): Delete.
27938 (rs6000_expand_unop_builtin): Delete SPE handling.
27939 (rs6000_expand_binop_builtin): Ditto.
27940 (spe_expand_stv_builtin): Delete.
27941 (bdesc_2arg_spe): Delete.
27942 (spe_expand_builtin): Delete.
27943 (spe_expand_predicate_builtin): Delete.
27944 (spe_expand_evsel_builtin): Delete.
27945 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27946 (spe_init_builtins): Delete.
27947 (spe_func_has_64bit_regs_p): Delete.
27948 (savres_routine_name): Delete "info" parameter. Adjust callers.
27949 (rs6000_emit_stack_reset): Ditto.
27950 (rs6000_dwarf_register_span): Delete.
27951 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27952 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27953 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27955 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27957 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27958 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27959 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27960 mulv2sf3, divv2sf3): Delete expanders.
27962 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27964 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27968 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27969 * config/rs6000/rs6000.c: Ditto.
27971 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27973 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27974 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27975 comparison_operator.
27977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27979 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27980 * config/rs6000/rs6000.opt: Ditto.
27981 * config/rs6000/t-rtems: Ditto.
27983 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27985 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27986 TARGET_E500_SINGLE by 0, simplify.
27987 * config/rs6000/rs6000.c: Ditto.
27988 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27989 (spe_build_register_parallel): Delete.
27990 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27991 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27992 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27993 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27994 (E500_CONVERT): Delete.
27995 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27997 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27999 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28000 * config/rs6000/dfp.md: Ditto.
28001 (negdd2, *negdd2_fpr): Merge.
28002 (absdd2, *absdd2_fpr): Merge.
28003 (negtd2, *negtd2_fpr): Merge.
28004 (abstd2, *abstd2_fpr): Merge.
28005 * config/rs6000/e500.h: Delete file.
28006 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28007 TARGET_FPRS by 1 and simplify.
28008 * config/rs6000/rs6000-c.c: Ditto.
28009 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28010 TARGET_DF_SPE by 0.
28011 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28013 * config/rs6000/rs6000.md: Ditto.
28014 (floatdidf2, *floatdidf2_fpr): Merge.
28015 (move_from_CR_gt_bit): Delete.
28016 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28017 (E500_CR_IOR_COMPARE): Delete.
28018 (All patterns that require !TARGET_FPRS): Delete.
28019 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28021 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28023 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28025 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28027 * graphds.c (add_edge): Intitialize edge's attached data.
28028 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28029 pointer parameter. Call pointed function on each edge during
28030 graph traversing. Skip traversing the edge when the function
28032 (graphds_dfs, graphds_scc): Ditto.
28033 (for_each_edge): New parameter. Pass the new parameter to callback
28035 * graphds.h (skip_edge_callback): New function pointer type.
28036 (graphds_dfs, graphds_scc): New function pointer parameter.
28037 (graphds_edge_callback, for_each_edge): New parameter.
28039 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28041 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28042 out code checking if runtime alias check is possible to below ...
28043 Call the new function.
28044 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28045 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28047 2017-06-07 Marek Polacek <polacek@redhat.com>
28050 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28051 TYPE_OVERFLOW_WRAPS check.
28053 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28055 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28056 if versioning is required.
28057 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28058 peeling with the check for versioning.
28060 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28062 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28063 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28064 Set true to new parameter if new ssa variable is defined.
28065 (vect_gen_vector_loop_niters): Refactor. Set range information
28066 for the new vector loop bound variable.
28067 (vect_do_peeling): Ditto.
28069 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28071 * tree-affine.c (ssa.h): Include header file.
28072 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28073 has wrapping overflow behavior.
28075 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28077 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28079 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28081 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28082 (tree_to_aff_combination): ... here.
28084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28086 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28087 reg_pressure model function.
28088 (ivopts_global_cost_for_size): Delete.
28089 (determine_set_costs, iv_ca_recount_cost): Call new model function
28090 ivopts_estimate_reg_pressure.
28092 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28094 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28095 expensive than udiv. Remove floating point cases from mod.
28097 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28099 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28100 Increase idiv cost.
28102 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28104 * config/aarch64/aarch64.md
28105 (copysignsf3): Fix mask generation.
28107 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28109 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28111 (class dump_manager): Add register_dumps method.
28112 * dumpfile.c: Include langhooks.h.
28113 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28114 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28115 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28116 (dump_manager::dump_register): Start with 512 entries instead of 32.
28117 (dump_manager::register_dumps): New method.
28118 * toplev.c (general_init): Instead of invoking register_dumps
28119 langhook, invoke register_dumps method on the dump manager.
28120 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28123 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28125 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28126 Say that # requires an associated define_split to exist, and that
28127 the define_split must be suitable for use after register allocation.
28129 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28131 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28132 (compute_outgoing_frequencies): Also initialize zero counts.
28133 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28134 around loops; preserve more of profile when nothing changes.
28136 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28138 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28140 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28141 * config/arm/arm-cpu-cdata.h: Regenerate.
28142 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28143 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28144 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28145 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28146 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28148 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28150 * config/arm/t-rmprofile: Likewise.
28151 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28153 2017-06-06 David S. Miller <davem@davemloft.net>
28156 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28157 blockage if function uses alloca.
28159 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28161 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28162 New "uid" fields to hold pretty-print IDs of group and ref.
28163 Memory references are now identified as <group_id>:<ref_id>
28164 instead of using [random] addresses.
28165 (dump_mem_details): Simplify, no functional change.
28166 (dump_mem_ref): Simplify and make output more concise.
28167 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28168 (find_or_create_group): Initialize group uid.
28169 (record_ref): Initialize ref uid. Improve debug output.
28170 (prune_group_by_reuse, should_issue_prefetch_p,)
28171 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28172 (mark_nontemporal_store, determine_loop_nest_reuse):
28173 Improve debug output.
28175 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28177 * dbgcnt.def (prefetch): New debug counter.
28178 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28179 (schedule_prefetches): Stop issueing prefetches if debug counter
28182 2017-06-06 Tom de Vries <tom@codesourcery.com>
28184 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28185 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28187 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28189 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28190 Use aarch64_reg_or_zero predicate for operand 4.
28191 (aarch64_compare_and_swap<mode> define_insn_and_split):
28192 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28193 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28195 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28197 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28198 (arm_compute_save_core_reg_mask): This.
28199 (thumb1_compute_save_reg_mask): Rename into ...
28200 (thumb1_compute_save_core_reg_mask): This.
28201 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28202 (arm_compute_frame_layout): Likewise.
28204 2017-06-06 Richard Biener <rguenther@suse.de>
28206 PR tree-optimization/80974
28207 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28208 keep or clear leaders SSA info.
28210 2017-06-06 Tom de Vries <tom@codesourcery.com>
28212 * config/nvptx/nvptx.c (split_mode_p): New function.
28213 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28215 2017-06-06 Tom de Vries <tom@codesourcery.com>
28217 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28219 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28222 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28225 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28227 * shrink-wrap.c (handle_simple_exit): Update profile.
28228 (try_shrink_wrapping): Upate profile.
28230 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28232 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28233 (tree_guess_outgoing_edge_probabilities): New.
28234 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28235 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28237 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28239 * ipa-split.c (split_function): Initialize return bb profile.
28241 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28243 * profile.c (compute_branch_probabilities): Also initialize
28244 EXIT_BLOCK profile.
28246 2017-06-06 Richard Biener <rguenther@suse.de>
28248 PR tree-optimization/80928
28249 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28250 (vect_analyze_loop_operations): Properly guard analysis for
28252 (vect_transform_loop): Likewise.
28253 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28254 (vect_model_induction_cost): Do not cost for pure SLP.
28255 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28256 of induction in inner loop vectorization.
28257 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28258 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28259 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28261 (vect_analyze_slp_cost_1): Cost induction.
28262 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28263 (vect_get_slp_vect_defs): Likewise.
28264 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28265 (vect_transform_stmt): Handle SLP reductions.
28266 * tree-vectorizer.h (vectorizable_induction): Adjust.
28268 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28270 * config/rs6000/rs6000.c (make_resolver_func): Update
28271 init_lowered_empty_function call.
28273 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28275 * doc/invoke.texi: Document the -fprofile-abs-path option.
28276 * common.opt (fprofile-abs-path): New option.
28277 * gcov-io.h (gcov_write_filename): Declare.
28278 * gcov-io.c (gcov_write_filename): New function.
28279 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28280 * profile.c (output_location): Likewise.
28282 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28284 * shring-wrap.c: Revert accidental commit.
28286 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28288 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28290 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28292 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28294 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28295 profile in callgraph edge.
28296 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28297 (apply_scale): Likewise.
28298 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28299 Also scale profile when inlining function with zero profile.
28300 (initialize_cfun): Update exit block profile even when it is zero.
28301 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28302 when profile is read.
28304 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28306 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28307 (CLONE_*): New constants to define the processors we can generate
28308 code for with the target_clone attribute.
28309 (rs6000_clone_map): New array to identify which clone processors
28310 the current program is running on.
28311 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28312 target_clone attribute.
28313 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28314 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28315 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28316 (cpu_expand_builtin): Add support for target_clone attribute.
28317 (rs6000_valid_attribute_p): Allow "default" attribute.
28318 (get_decl_name): New debug function to simplify printing the
28319 current function name in debugging statements.
28320 (rs6000_clone_priority): New functions to support the target_clone
28321 attribute, and be able to generate code to switch between ISA 2.05
28322 through ISA 3.0 (power6 through power9).
28323 (rs6000_compare_version_priority): Likewise.
28324 (rs6000_get_function_versions_dispatcher): Likewise.
28325 (make_resolver_func): Likewise.
28326 (add_condition_to_bb): Likewise.
28327 (dispatch_function_versions): Likewise.
28328 (rs6000_generate_version_dispatcher_body): Likewise.
28329 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28330 (fusion_gpr_load_p): Fix a spacing issue.
28331 * doc/extend.texi (Common Function Attributes): Document that the
28332 PowerPC supports the target_clone attribute.
28334 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28336 * config/arm/arm.h: explain F symbol found in description of ARM
28337 register allocation in its legend.
28339 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28341 * config/mips/frame-header-opt.c: Include profile-count.h.
28342 * config/riscv/riscv.c: Include profile-count.h
28344 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28346 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28348 (sm_set_flag_if_changed): Add bbs field.
28349 (execute_sm_if_changed_flag_set): Pass BBS.
28350 (execute_sm): Update.
28352 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28354 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28357 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28359 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28360 (peephole2): New peephole2 to emit the above.
28361 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28363 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28365 * config/aarch64/aarch64.c (define_peephole2 above
28366 *sub_<shift>_<mode>): New peephole.
28368 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28370 * config/i386/i386.c (make_resolver_func): Update.
28371 * Makefile.in: Add profile-count.h and profile-count.o
28372 * auto-profile.c (afdo_indirect_call): Update to new API.
28373 (afdo_set_bb_count): Update.
28374 (afdo_propagate_edge): Update.
28375 (afdo_propagate_circuit): Update.
28376 (afdo_calculate_branch_prob): Update.
28377 (afdo_annotate_cfg): Update.
28378 * basic-block.h: Include profile-count.h
28379 (struct edge_def): Turn count to profile_count.
28380 (struct basic_block_def): Likewie.
28381 (REG_BR_PROB_BASE): Move to profile-count.h
28382 (RDIV): Move to profile-count.h
28383 * bb-reorder.c (max_entry_count): Turn to profile_count.
28384 (find_traces): Update.
28385 (rotate_loop):Update.
28386 (connect_traces):Update.
28387 (sanitize_hot_paths):Update.
28388 * bt-load.c (migrate_btr_defs): Update.
28389 * cfg.c (RDIV): Remove.
28390 (init_flow): Use alloc_block.
28391 (alloc_block): Uninitialize count.
28392 (unchecked_make_edge): Uninitialize count.
28393 (check_bb_profile): Update.
28394 (dump_edge_info): Update.
28395 (dump_bb_info): Update.
28396 (update_bb_profile_for_threading): Update.
28397 (scale_bbs_frequencies_int): Update.
28398 (scale_bbs_frequencies_gcov_type): Update.
28399 (scale_bbs_frequencies_profile_count): New.
28400 * cfg.h (update_bb_profile_for_threading): Update.
28401 (scale_bbs_frequencies_profile_count): Declare.
28402 * cfgbuild.c (compute_outgoing_frequencies): Update.
28403 (find_many_sub_basic_blocks): Update.
28404 * cfgcleanup.c (try_forward_edges): Update.
28405 (try_crossjump_to_edge): Update.
28406 * cfgexpand.c (expand_gimple_tailcall): Update.
28407 (construct_exit_block): Update.
28408 * cfghooks.c (verify_flow_info): Update.
28409 (dump_bb_for_graph): Update.
28410 (split_edge): Update.
28411 (make_forwarder_block): Update.
28412 (duplicate_block): Update.
28413 (account_profile_record): Update.
28414 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28415 (get_estimated_loop_iterations): Update.
28416 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28417 (single_likely_exit): Update.
28418 * cfgloopmanip.c (scale_loop_profile): Update.
28420 (set_zero_probability): Update.
28421 (lv_adjust_loop_entry_edge): Update.
28422 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28423 (purge_dead_edges): Update.
28424 (rtl_account_profile_record): Update.
28425 * cgraph.c (cgraph_node::create): Uninitialize count.
28426 (symbol_table::create_edge): Uninitialize count.
28427 (cgraph_update_edges_for_call_stmt_node): Update.
28428 (cgraph_edge::dump_edge_flags): Update.
28429 (cgraph_node::dump): Update.
28430 (cgraph_edge::maybe_hot_p): Update.
28431 * cgraph.h: Include profile-count.h
28432 (create_clone), create_edge, create_indirect_edge): Update.
28433 (cgraph_node): Turn count to profile_count.
28434 (cgraph_edge0: Likewise.
28435 (make_speculative, clone): Update.
28436 (create_edge): Update.
28437 (init_lowered_empty_function): Update.
28438 * cgraphclones.c (cgraph_edge::clone): Update.
28439 (duplicate_thunk_for_node): Update.
28440 (cgraph_node::create_clone): Update.
28441 * cgraphunit.c (cgraph_node::analyze): Update.
28442 (cgraph_node::expand_thunk): Update.
28443 * final.c (dump_basic_block_info): Update.
28444 * gimple-streamer-in.c (input_bb): Update.
28445 * gimple-streamer-out.c (output_bb): Update.
28446 * graphite.c (print_global_statistics): Update.
28447 (print_graphite_scop_statistics): Update.
28448 * hsa-brig.c: Include basic-block.h.
28449 * hsa-dump.c: Include basic-block.h.
28450 * hsa-gen.c (T sum_slice): Update.
28451 (convert_switch_statements):Update.
28452 * hsa-regalloc.c: Include basic-block.h.
28453 * ipa-chkp.c (chkp_produce_thunks): Update.
28454 * ipa-cp.c (struct caller_statistics): Update.
28455 (init_caller_stats): Update.
28456 (gather_caller_stats): Update.
28457 (ipcp_cloning_candidate_p): Update.
28458 (good_cloning_opportunity_p): Update.
28459 (get_info_about_necessary_edges): Update.
28460 (dump_profile_updates): Update.
28461 (update_profiling_info): Update.
28462 (update_specialized_profile): Update.
28463 (perhaps_add_new_callers): Update.
28464 (decide_about_value): Update.
28465 (ipa_cp_c_finalize): Update.
28466 * ipa-devirt.c (struct odr_type_warn_count): Update.
28467 (struct decl_warn_count): Update.
28468 (struct final_warning_record): Update.
28469 (possible_polymorphic_call_targets): Update.
28470 (ipa_devirt): Update.
28471 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28472 * ipa-icf.c (sem_function::merge): Update.
28473 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28474 * ipa-inline.c (compute_uninlined_call_time): Update.
28475 (compute_inlined_call_time): Update.
28476 (want_inline_small_function_p): Update.
28477 (want_inline_self_recursive_call_p): Update.
28478 (edge_badness): Update.
28479 (lookup_recursive_calls): Update.
28480 (recursive_inlining): Update.
28481 (inline_small_functions): Update.
28482 (dump_overall_stats): Update.
28483 (dump_inline_stats): Update.
28484 * ipa-profile.c (ipa_profile_generate_summary): Update.
28485 (ipa_propagate_frequency): Update.
28486 (ipa_profile): Update.
28487 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28488 * ipa-utils.c (ipa_merge_profiles): Update.
28489 * loop-doloop.c (doloop_modify): Update.
28490 * loop-unroll.c (report_unroll): Update.
28491 (unroll_loop_runtime_iterations): Update.
28492 * lto-cgraph.c (lto_output_edge): Update.
28493 (lto_output_node): Update.
28494 (input_node): Update.
28495 (input_edge): Update.
28496 (merge_profile_summaries): Update.
28497 * lto-streamer-in.c (input_cfg): Update.
28498 * lto-streamer-out.c (output_cfg): Update.
28499 * mcf.c (create_fixup_graph): Update.
28500 (adjust_cfg_counts): Update.
28501 (sum_edge_counts): Update.
28502 * modulo-sched.c (sms_schedule): Update.
28503 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28504 * predict.c (maybe_hot_count_p): Update.
28505 (probably_never_executed): Update.
28506 (dump_prediction): Update.
28507 (combine_predictions_for_bb): Update.
28508 (propagate_freq): Update.
28509 (handle_missing_profiles): Update.
28510 (counts_to_freqs): Update.
28511 (rebuild_frequencies): Update.
28512 (force_edge_cold): Update.
28513 * predict.h: Include profile-count.h
28514 (maybe_hot_count_p, counts_to_freqs): UPdate.
28515 * print-rtl-function.c: Do not include cfg.h
28516 * print-rtl.c: Include basic-block.h
28517 * profile-count.c: New file.
28518 * profile-count.h: New file.
28519 * profile.c (is_edge_inconsistent): Update.
28520 (correct_negative_edge_counts): Update.
28521 (is_inconsistent): Update.
28522 (set_bb_counts): Update.
28523 (read_profile_edge_counts): Update.
28524 (compute_frequency_overlap): Update.
28525 (compute_branch_probabilities): Update; Initialize and deinitialize
28527 (branch_prob): Update.
28528 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28529 (edge_gcov_count): New.
28530 (bb_gcov_count): New.
28531 * shrink-wrap.c (try_shrink_wrapping): Update.
28532 * tracer.c (better_p): Update.
28533 * trans-mem.c (expand_transaction): Update.
28534 (ipa_tm_insert_irr_call): Update.
28535 (ipa_tm_insert_gettmclone_call): Update.
28536 * tree-call-cdce.c: Update.
28537 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28538 (gimple_duplicate_sese_tail): Update.
28539 (gimple_account_profile_record): Update.
28540 (execute_fixup_cfg): Update.
28541 * tree-inline.c (copy_bb): Update.
28542 (copy_edges_for_bb): Update.
28543 (initialize_cfun): Update.
28544 (freqs_to_counts): Update.
28545 (copy_cfg_body): Update.
28546 (expand_call_inline): Update.
28547 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28548 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28549 (try_unroll_loop_completely): Update.
28550 (try_peel_loop): Update.
28551 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28552 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28553 * tree-ssa-loop-split.c (connect_loops): Update.
28554 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28555 * tree-ssa-reassoc.c (branch_fixup): Update.
28556 * tree-ssa-tail-merge.c (replace_block_by): Update.
28557 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28558 (compute_path_counts): Update.
28559 (update_profile): Update.
28560 (recompute_probabilities): Update.
28561 (update_joiner_offpath_counts): Update.
28562 (estimated_freqs_path): Update.
28563 (freqs_to_counts_path): Update.
28564 (clear_counts_path): Update.
28565 (ssa_fix_duplicate_block_edges): Update.
28566 (duplicate_thread_path): Update.
28567 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28568 (struct switch_conv_info): Update.
28569 * tree-tailcall.c (decrease_profile): Update.
28570 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28571 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28572 * value-prof.c (check_counter): Update.
28573 (gimple_divmod_fixed_value): Update.
28574 (gimple_mod_pow2): Update.
28575 (gimple_mod_subtract): Update.
28576 (gimple_ic_transform): Update.
28577 (gimple_stringop_fixed_value): Update.
28578 * value-prof.h (gimple_ic): Update.
28580 2017-06-02 Carl Love <cel@us.ibm.com>
28582 * config/rs6000/rs6000-c: Add support for built-in functions
28583 vector double vec_doublee (vector signed int);
28584 vector double vec_doublee (vector unsigned int);
28585 vector double vec_doublee (vector float);
28586 vector double vec_doubleh (vector signed int);
28587 vector double vec_doubleh (vector unsigned int);
28588 vector double vec_doubleh (vector float);
28589 vector double vec_doublel (vector signed int);
28590 vector double vec_doublel (vector unsigned int);
28591 vector double vec_doublel (vector float);
28592 vector double vec_doubleo (vector signed int);
28593 vector double vec_doubleo (vector unsigned int);
28594 vector double vec_doubleo (vector float);.
28595 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28596 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28598 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28599 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28600 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28602 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28603 vec_doublel, vec_doubleh.
28604 * doc/extend.texi: Update the built-in documentation file for the
28605 new built-in functions.
28607 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28610 * ipa-inline-analysis.c (free_growth_caches): Set
28611 edge_removal_hook_holder to NULL after removing it.
28613 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28615 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28616 comparision with zero.
28618 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28619 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28620 for early expansion of vec_min and vec_max builtins.
28621 (builtin_function_type): Add min/max unsigned variants to those
28622 identified as having unsigned arguments.
28624 2017-06-02 Olivier Hainque <hainque@adacore.com>
28626 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28628 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28630 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28631 Use VALL_F16 iterator rather than VALL.
28633 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28635 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28636 Emit CBNZ inside loop when doing a strong exchange and comparing
28637 against zero. Generate the CC flags after the loop.
28639 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28641 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28642 (dl_section_ref): New.
28643 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28644 On AIX, append an expression to subtract the size of the
28645 section length to dl_section_ref.
28647 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28649 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28650 for early expansion of vector absolute builtins.
28652 2017-06-02 Richard Biener <rguenther@suse.de>
28654 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28655 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28657 2017-06-02 Richard Biener <rguenther@suse.de>
28659 PR tree-optimization/80948
28660 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28661 stmt order as well.
28663 2017-06-02 Richard Biener <rguenther@suse.de>
28665 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28667 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28668 for inductions as relevant.
28670 2017-06-02 Richard Biener <rguenther@suse.de>
28672 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28673 (vectorizable_induction): ... this. Remove dead code.
28675 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28677 * builtins. (expand_builtin_alloca): Remove second parameter and
28678 infer its value from the first parameter instead.
28679 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28681 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28683 PR rtl-optimization/80903
28684 * loop-doloop.c (add_test): Unshare sequence.
28686 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28688 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28690 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28692 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28694 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28695 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28696 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28697 (xlogue_layout::get_stub_rtx): Make static.
28698 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28699 (xlogue_layout::compute_stub_managed_regs): Rename to...
28700 (xlogue_layout::count_stub_managed_regs): ...this.
28701 (xlogue_layout::is_stub_managed_reg): New function.
28702 (xlogue_layout::m_stub_names): Rename to...
28703 (xlogue_layout::s_stub_names): ...this, make static.
28704 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28705 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28706 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28707 xlogue_layout::s_stub_names): Instantiate statics.
28708 (stub_managed_regs): Remove.
28709 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28710 (disable_call_ms2sysv_xlogues): Rename to...
28711 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28712 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28714 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28715 change after reload_completed.
28716 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28718 (ix86_expand_prologue): Likewise.
28719 (ix86_expand_epilogue): Likewise.
28720 (ix86_expand_split_stack_prologue): Likewise.
28721 (ix86_compute_frame_layout): Remove frame parameter ...
28722 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28723 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28725 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28726 (ix86_frame): Move from here ...
28727 * config/i386/i386.h (ix86_frame): ... to here.
28728 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28729 complete ix86_frame data structure instead. Remove some_ld_name.
28731 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28733 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28734 symbols that hold a DECL_VALUE_EXPR.
28736 2017-06-01 Martin Jambor <mjambor@suse.cz>
28738 PR tree-optimization/80898
28739 * tree-sra.c (process_subtree_disqualification): Removed.
28740 (disqualify_candidate): Do not acll
28741 process_subtree_disqualification.
28742 (subtree_mark_written_and_enqueue): New function.
28743 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28744 RHS has been disqualified and re-queue LHS if necessary. Apart
28745 from that, ignore disqualified RHS.
28747 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28749 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28750 address fetch for z10 or later.
28752 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28754 * config/arc/arc.md (tst_movb): Add guard when splitting.
28756 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28758 * config/arc/arc.c (arc_can_eliminate): Test against
28759 arc_frame_pointer_needed.
28761 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28763 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28764 to prevent store reordering.
28765 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28766 (type): Add block type.
28767 (stack_tie): Define special instruction to be used in
28770 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28772 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28773 constraint. It is not valid for the pattern.
28774 (noncommutative_binary_comparison): Likewise.
28776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28778 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28781 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28783 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28784 be used by the reg-alloc.
28786 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28788 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28789 reg-alloc when having mul64 or mul32x16 instructions.
28790 (mulsidi3): Likewise.
28791 (umulsidi3): Likewise.
28792 (mulsi32x16): New pattern.
28793 (mulsi64): Likewise.
28794 (mulsidi64): Likewise.
28795 (umulsidi64): Likewise.
28796 (MUL32x16_REG): Define.
28797 (mul64_600): Use MUL32x16_REG.
28798 (mac64_600): Likewise.
28799 (umul64_600): Likewise.
28800 (umac64_600): Likewise.
28802 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28804 * config/arc/arc.md (mulsi3_700): Make it commutative.
28806 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28808 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28810 (*sign_extendsidi2_insn): Likewise for movstosw.
28812 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28814 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28815 the type of the input discriminant value. Convert the
28816 discriminant value of signedness vary.
28818 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28820 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28821 Add to -Wall section.
28823 2017-06-01 Richard Biener <rguenther@suse.de>
28825 PR middle-end/66313
28826 * fold-const.c (fold_plusminus_mult_expr): If the factored
28827 factor may be zero use a wrapping type for the inner operation.
28828 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28829 and handle moved defs.
28830 (process_assignment): Properly guard the unary op case. Return a
28831 tri-state indicating that moving the stmt before the call may allow
28832 to continue. Pass through to_move.
28833 (find_tail_calls): Handle moving unrelated defs before
28836 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28839 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28840 splitter result in the canonical way.
28842 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28844 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28845 also for 32bit target. Update insn attributes.
28846 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28848 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28850 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28851 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28852 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28853 (_mm_maskz_min_ss): New intrinsics.
28855 2017-05-31 Martin Liska <mliska@suse.cz>
28857 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28858 Change comment style to one we normally use.
28859 (vectorizable_reduction): Likewise.
28860 (vectorizable_induction): Likewise.
28861 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28862 (vectorizable_call): Likewise.
28863 (vectorizable_simd_clone_call): Likewise.
28864 (vectorizable_conversion): Likewise.
28865 (vectorizable_assignment): Likewise.
28866 (vectorizable_shift): Likewise.
28867 (vectorizable_operation): Likewise.
28868 (vectorizable_store): Likewise.
28869 (vectorizable_load): Likewise.
28870 * tree-vectorizer.h: Likewise.
28872 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28874 * passes.c (emergency_dump_function): New.
28875 * tree-pass.h (emergency_dump_function): Declare.
28876 * plugin.c (plugins_internal_error_function): Remove.
28877 * plugin.h (plugins_internal_error_function): Remove declaration.
28878 * toplev.c (internal_error_function): New static function. Use it...
28879 (general_init): ...here.
28881 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28883 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28884 (arc_rtx_costs): Add costs for new patterns.
28885 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28886 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28888 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28890 * tree-ssa-strlen.c (get_next_strinfo): New function.
28891 (get_stridx_plus_constant): Use it.
28892 (zero_length_string): Likewise.
28893 (adjust_related_strinfos): Likewise.
28894 (adjust_last_stmt): Likewise.
28896 2017-05-31 Richard Biener <rguenther@suse.de>
28899 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28900 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28902 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28904 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28905 loop_vinfo argument and use of dependence distance vectors.
28906 Check instead whether the two references differ only in their
28907 initial value and assume that they have the same alignment if the
28908 difference is a multiple of the vector alignment.
28909 (vect_analyze_data_refs_alignment): Update call accordingly.
28911 2017-05-31 Martin Liska <mliska@suse.cz>
28914 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28916 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28918 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28919 (create_intersect_range_checks): Move from ...
28920 * tree-data-ref.c (create_intersect_range_checks_index)
28921 (create_intersect_range_checks): ... to here.
28922 (create_runtime_alias_checks): New function factored from ...
28923 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28924 here. Call above function.
28925 * tree-data-ref.h (create_runtime_alias_checks): New function.
28927 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28929 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28930 segment length for dr_b and compute it in wide_int.
28932 2017-05-31 Richard Biener <rguenther@suse.de>
28934 PR tree-optimization/80906
28935 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28936 and pass through iv_map.
28937 (copy_bb_and_scalar_dependences): Adjust.
28938 (translate_pending_phi_nodes): Likewise.
28939 (copy_loop_close_phi_args): Handle code-generating IVs instead
28942 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28944 * diagnostic-color.c (color_dict): Add "type-diff".
28945 (parse_gcc_colors): Update comment.
28946 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28947 -fdiagnostics-show-template-tree and -fno-elide-type.
28948 (GCC_COLORS): Add type-diff to example.
28950 (-fdiagnostics-show-template-tree): New.
28951 (-fno-elide-type): New.
28952 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28953 the pp_format_decoder callback. Call any m_format_postprocessor's
28955 (pretty_printer::pretty_printer): Initialize
28956 m_format_postprocessor.
28957 (pretty_printer::~pretty_printer): Delete any
28958 m_format_postprocessor.
28959 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28960 (class format_postprocessor): New class.
28961 (struct pretty_printer::format_decoder): Document the new parameters.
28962 (struct pretty_printer::m_format_postprocessor): New field.
28963 * tree-diagnostic.c (default_tree_printer): Update for new
28964 bool and const char ** params.
28965 * tree-diagnostic.h (default_tree_printer): Likewise.
28967 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28969 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28970 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28972 rs6000_gen_cell_microcode code.
28973 (rs6000_final_prescan_insn): Delete.
28974 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28975 "warn-cell-microcode" entries.
28976 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28977 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28978 throughout. Change cc_reg_not_micro_cr0_operand to
28979 cc_reg_not_cr0_operand throughout.
28980 (*extendhi<mode>2_noload): Delete.
28981 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28982 (mwarn-cell-microcode): Delete.
28983 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28984 -mgen-cell-microcode and -mwarn-cell-microcode.
28986 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28989 * config/i386/constraints.md (Yd): New constraint.
28991 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28992 and (?Yd, r) alternatives. Update insn attributes.
28993 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28994 and (?*Yd, r) alternatives. Update insn attributes.
28995 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28997 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28999 * gimplify.c (gimplify_modify_expr): Don't create a
29000 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29003 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29005 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29007 2017-05-30 Richard Biener <rguenther@suse.de>
29009 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29010 and reduc_def fields.
29011 (STMT_VINFO_REDUC_TYPE): New define.
29012 (STMT_VINFO_REDUC_DEF): Likewise.
29013 (vect_force_simple_reduction): Adjust prototype.
29014 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29015 (vect_is_simple_reduction): Remove check_reduction argument.
29016 (vect_force_simple_reduction): Adjust and set
29017 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29018 (vectorizable_reduction): Do not re-do reduction analysis
29019 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29020 * tree-parloops.c (gather_scalar_reductions): Adjust.
29022 2017-05-30 Richard Biener <rguenther@suse.de>
29024 PR middle-end/80901
29025 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29028 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29030 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29031 Introduce unknown_misalignment parameter and remove vf.
29032 (vect_peeling_hash_get_lowest_cost):
29033 Pass unknown_misalignment parameter.
29034 (vect_enhance_data_refs_alignment):
29035 Fix unsupportable data ref treatment.
29037 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29039 * tree-vect-data-refs.c (vect_get_data_access_cost):
29040 Workaround for SLP handling.
29041 (vect_enhance_data_refs_alignment):
29042 Compute costs for doing no peeling at all, compare to the best
29043 peeling costs so far and avoid peeling if cheaper.
29045 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29047 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29048 Return peeling info and set costs to zero for unlimited cost
29050 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29051 with unknown misalignment. Compute and costs for unknown
29052 misalignment, compare them to the costs for known misalignment
29053 and choose the cheapest for peeling.
29055 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29057 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29058 (vect_get_peeling_costs_all_drs): Create function.
29059 (vect_peeling_hash_get_lowest_cost):
29060 Use vect_get_peeling_costs_all_drs.
29061 (vect_peeling_supportable): Create function.
29062 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29064 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29066 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29067 DR_HAS_NEGATIVE_STEP.
29068 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29069 (vect_enhance_data_refs_alignment): Use.
29070 (vect_duplicate_ssa_name_ptr_info): Use.
29071 * tree-vectorizer.h (dr_misalignment): Use.
29072 (known_alignment_for_access_p): Use.
29074 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29077 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29079 (has_section_name): New function.
29081 2017-05-30 Martin Liska <mliska@suse.cz>
29084 * auto-profile.c (get_function_decl_from_block): Fix
29087 2017-05-30 Richard Biener <rguenther@suse.de>
29089 PR middle-end/80876
29090 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29092 2017-05-30 Martin Liska <mliska@suse.cz>
29094 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29095 * dumpfile.h (struct dump_file_info): Remove ctors.
29097 2017-05-30 Martin Liska <mliska@suse.cz>
29099 * predict.def: Fix GNU coding style.
29101 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29103 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29104 Mark 'to' argument with ATTRIBUTE_UNUSED.
29106 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29108 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29109 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29110 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29113 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29115 * doc/install.texi (Options specification): Restore entry of
29116 --enable-sjlj-exceptions.
29118 2017-05-27 Michael Eager <eager@eagercon.com>
29121 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29123 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29125 * config/microblaze/microblaze.h
29126 (FIXED_REGISTERS): Update in macro.
29127 (CALL_USED_REGISTERS): Update in macro.
29129 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29131 * doc/install.texi: Add links to macOS binary distributions.
29133 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29137 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29139 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29141 2017-05-26 Martin Liska <mliska@suse.cz>
29143 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29145 2017-05-26 Martin Liska <mliska@suse.cz>
29147 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29148 always leading ';; '.
29149 (dump_bb_info): Likewise.
29150 (brief_dump_cfg): Likewise.
29151 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29152 * dumpfile.c: Remove usage of TDF_VERBOSE.
29153 * dumpfile.h (enum dump_kind): Likewise.
29154 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29155 * print-tree.c (debug_verbose): Remove.
29156 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29157 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29158 * tree-diagnostic.c (default_tree_printer): Replace
29159 TDF_DIAGNOSTIC with TDF_SLIM.
29161 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29163 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29164 in parameter loop, rather than loop_vinfo.
29165 (create_intersect_range_checks): Ditto.
29166 (vect_create_cond_for_alias_checks): Update call to above functions.
29168 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29170 PR tree-optimization/80815
29171 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29172 for merging runtime alias checks. Handle negative DR_STEPs.
29174 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29176 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29178 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29179 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29180 out code pruning runtime alias checks.
29181 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29182 factored out from above.
29183 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29185 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29187 (prune_runtime_alias_test_list): New decalaration.
29189 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29191 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29192 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29193 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29194 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29195 (operator==, comp_dr_with_seg_len_pair): Ditto.
29196 (vect_prune_runtime_alias_test_list): Ditto.
29198 2017-05-26 Martin Liska <mliska@suse.cz>
29201 * params.def: Bound partial-inlining-entry-probability param.
29203 2017-05-26 Marek Polacek <polacek@redhat.com>
29206 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29209 2017-05-26 Richard Biener <rguenther@suse.de>
29211 PR tree-optimization/80842
29212 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29215 2017-05-26 Richard Biener <rguenther@suse.de>
29217 PR tree-optimization/80844
29218 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29220 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29222 * doc/md.texi (Machine Constraints): Update x86 family
29223 machine constraints section to match 'config/i386/constraints.md'.
29225 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29227 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29229 2017-05-25 Nathan Sidwell <nathan@acm.org>
29231 * doc/invoke.texi (--enable-languages): Update documentation.
29233 2017-05-25 Martin Liska <mliska@suse.cz>
29235 * dumpfile.c: Add TDF_FOLDING.
29236 * dumpfile.h (enum dump_kind): Likewise.
29237 * genmatch.c (dt_simplify::gen_1): Use it.
29239 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29241 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29243 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29245 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29246 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29248 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29250 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29251 * match.pd (X == C): Rewrite it here.
29252 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29253 with_certain_nonzero_bits2): New predicates.
29254 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29256 2017-05-24 Nathan Sidwell <nathan@acm.org>
29258 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29261 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29264 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29266 * config/powerpcspe: New port. Files are copied from the rs6000
29267 port, with "rs6000" in filenames replaced by "powerpcspe".
29269 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29271 PR rtl-optimization/80754
29272 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29274 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29276 * config/sparc/sparc.md (length): Return the correct value for -mflat
29277 sibcalls to match output_sibcall.
29279 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29283 * config/rs6000/rs6000.c (struct machine_function): Add new field
29285 (rs6000_get_separate_components): Init that field, use it.
29286 (rs6000_components_for_bb): Use the field.
29288 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29290 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29292 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29294 PR middle-end/80823
29295 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29297 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29300 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29301 address against address_operand predicate.
29302 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29304 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29306 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29307 parameters passed indirectly.
29309 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29311 * config/i386/i386.md (*movdi_internal): Remove SSE4
29312 alternative 18 (?r, *v). Update insn attributes.
29313 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29314 Update insn attributes.
29315 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29316 Update insn attributes.
29317 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29318 alternative 1 (r, v). Remove isa attribute.
29319 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29320 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29321 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29323 2017-05-23 Tom de Vries <tom@codesourcery.com>
29325 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29328 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29330 * cgraphunit.c (symbol_table::process_new_functions): Update.
29331 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29332 (inline_generate_summary): Rename to ...
29333 (ipa_fn_summary_generate): ... this one.
29334 (inline_read_summary): Rename to ...
29335 (ipa_fn_summary_read): ... this one.
29336 (inline_write_summary): Rename to ...
29337 (ipa_fn_summary_write): ... this one.
29338 (inline_free_summary): Rename to ...
29339 (ipa_free_fn_summary): ... this one.
29340 (pass_data_local_fn_summary, pass_local_fn_summary,
29341 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29342 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29343 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29344 make_pass_ipa_fn_summary): New.
29345 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29346 inline_write_summary, inline_free_summary): Remove.
29347 (ipa_free_fn_summary) : New.
29348 * ipa-inline.c (ipa_inline): Update.
29349 (pass_ipa_inline): Do not generate summaries.
29350 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29352 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29353 and add pass_ipa_fn_summary.
29354 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29356 (make_pass_inline_parameters): Remove.
29358 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29360 * omp-low.c (struct omp_context): Remove "default_kind" member.
29363 * omp-offload.c (execute_oacc_device_lower): Remove the
29364 parallelism dimensions function attributes for unparallelized
29365 OpenACC kernels constructs.
29367 2017-05-23 Martin Liska <mliska@suse.cz>
29369 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29371 (cgraph_edge::make_speculative): Likewise.
29372 (cgraph_edge::resolve_speculation): Likewise.
29373 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29374 (cgraph_node::dump): Likewise.
29375 * cgraph.h: Likewise.
29376 * cgraphunit.c (analyze_functions): Likewise.
29377 (symbol_table::compile): Likewise.
29378 * ipa-cp.c (print_all_lattices): Likewise.
29379 (determine_versionability): Likewise.
29380 (initialize_node_lattices): Likewise.
29381 (ipcp_verify_propagated_values): Likewise.
29382 (estimate_local_effects): Likewise.
29383 (update_profiling_info): Likewise.
29384 (create_specialized_node): Likewise.
29385 (perhaps_add_new_callers): Likewise.
29386 (decide_about_value): Likewise.
29387 (decide_whether_version_node): Likewise.
29388 (identify_dead_nodes): Likewise.
29389 (ipcp_store_bits_results): Likewise.
29390 * ipa-devirt.c (dump_targets): Likewise.
29391 (ipa_devirt): Likewise.
29392 * ipa-icf.c (sem_item::dump): Likewise.
29393 (sem_function::equals): Likewise.
29394 (sem_variable::equals): Likewise.
29395 (sem_item_optimizer::read_section): Likewise.
29396 (sem_item_optimizer::execute): Likewise.
29397 (congruence_class::dump): Likewise.
29398 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29399 (dump_inline_summary): Likewise.
29400 (estimate_node_size_and_time): Likewise.
29401 (inline_analyze_function): Likewise.
29402 * ipa-inline-transform.c (inline_call): Likewise.
29403 * ipa-inline.c (report_inline_failed_reason): Likewise.
29404 (want_early_inline_function_p): Likewise.
29405 (edge_badness): Likewise.
29406 (update_edge_key): Likewise.
29407 (inline_small_functions): Likewise.
29408 * ipa-profile.c (ipa_profile): Likewise.
29409 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29410 (ipa_make_edge_direct_to_target): Likewise.
29411 (remove_described_reference): Likewise.
29412 (ipa_impossible_devirt_target): Likewise.
29413 (propagate_controlled_uses): Likewise.
29414 (ipa_print_node_params): Likewise.
29415 (ipcp_transform_function): Likewise.
29416 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29417 (propagate_pure_const): Likewise.
29418 * ipa-reference.c (generate_summary): Likewise.
29419 (read_write_all_from_decl): Likewise.
29420 (propagate): Likewise.
29421 (ipa_reference_read_optimization_summary): Likewise.
29422 * ipa-utils.c (ipa_merge_profiles): Likewise.
29423 * ipa.c (walk_polymorphic_call_targets): Likewise.
29424 (symbol_table::remove_unreachable_nodes): Likewise.
29425 (ipa_single_use): Likewise.
29426 * passes.c (execute_todo): Likewise.
29427 * predict.c (drop_profile): Likewise.
29428 * symtab.c (symtab_node::get_dump_name): New function.
29429 (symtab_node::dump_name): Likewise.
29430 (symtab_node::dump_asm_name): Likewise.
29431 (symtab_node::dump_references): Likewise.
29432 (symtab_node::dump_referring): Likewise.
29433 (symtab_node::dump_base): Likewise.
29434 (symtab_node::debug_symtab): Likewise.
29435 * tree-sra.c (convert_callers_for_node): Likewise.
29436 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29437 * value-prof.c (init_node_map): Likewise.
29439 2017-05-23 Martin Liska <mliska@suse.cz>
29441 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29442 and symtab_node::debug_symtab to symbol_table::debug.
29443 * cgraphunit.c (analyze_functions): Use the renamed function.
29444 (symbol_table::compile): Likewise.
29445 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29446 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29447 * passes.c (execute_todo): Likewise.
29448 * symtab.c (symbol_table::dump): New function.
29449 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29451 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29453 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29454 that nonconst implies exec.
29456 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29458 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29459 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29460 (inline_edge_summary_vec): Turn into ...
29461 (ipa_call_summaries): ... this one.
29462 (redirect_to_unreachable, edge_set_predicate,
29463 evaluate_properties_for_edge, inline_summary_alloc,
29464 reset_ipa_call_summary, reset_inline_summary,
29465 inline_summary_t::duplicate): Update.
29466 (inline_edge_duplication_hook): Turn to ...
29467 (ipa_call_summary_t::duplicate): ... this one.
29468 (inline_edge_removal_hook): Turn to ...
29469 (ipa_call_summary_t::remove): ... this one.
29470 (dump_inline_edge_summary): Turn to ...
29471 (dump_ipa_call_summary): ... this one.
29472 (estimate_function_body_sizes): Update.
29473 (inline_update_callee_summaries): Update.
29474 (remap_edge_change_prob): Update.
29475 (remap_edge_summaries): Update.
29476 (inline_merge_summary): Update.
29477 (do_estimate_edge_time): Update.
29478 (inline_generate_summary): Update.
29479 (inline_read_section): Update.
29480 (inline_read_summary): Update.
29481 (inline_free_summary): Update.
29482 * ipa-inline.c (can_inline_edge_p): Update.
29483 (compute_inlined_call_time): Update.
29484 (want_inline_small_function_p): Update.
29485 (edge_badness): Update.
29486 (early_inliner): Update.
29487 * ipa-inline.h (inline_edge_summary): Turn to ...
29488 (ipa_call_summary): ... this one.
29489 (ipa_call_summary_t): New class.
29490 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29491 (ipa_call_summaries): New.
29492 (inline_edge_summary): Remove.
29493 (estimate_edge_growth): Update.
29494 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29495 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29496 * ipa-split.c (execute_split_functions): Update.
29497 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29499 2017-05-23 Tom de Vries <tom@codesourcery.com>
29501 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29502 attributes): Document rdrand effective target.
29504 2017-05-23 Tom de Vries <tom@codesourcery.com>
29506 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29507 attributes): Sort alphabetically.
29509 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29511 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29513 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29516 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29517 V2DF/V2DI splat into two separate patterns, one that handles
29518 registers, and the other that only handles memory. Drop support
29519 for splatting from a GPR on ISA 2.07 and then splitting the
29520 splat into direct move and splat.
29521 (vsx_splat_<mode>_reg): Likewise.
29522 (vsx_splat_<mode>_mem): Likewise.
29524 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29526 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29528 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29530 PR middle-end/80809
29531 * omp-low.c (finish_taskreg_remap): New function.
29532 (finish_taskreg_scan): If unit size of ctx->record_type
29533 is non-constant, unshare the size expression and replace
29534 decls in it with possible outer var refs.
29536 PR middle-end/80809
29537 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29538 GOVD_SHARED rather than GOVD_PRIVATE with it.
29539 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29540 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29542 PR middle-end/80853
29543 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29544 as last argument to build_outer_var_ref for pointer bases of array
29545 section reductions.
29547 2017-05-19 Martin Sebor <msebor@redhat.com>
29549 * print-tree.c (print_node): Print DECL_READ_P flag.
29551 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29553 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29554 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29555 * cgraph.c: Likewise.
29556 * cgraphunit.c: Likewise.
29557 * gengtype.c: Likewise.
29558 * ipa-cp.c: Likewise.
29559 * ipa-devirt.c: Likewise.
29560 * ipa-icf.c: Likewise.
29561 * ipa-predicate.c: Likewise.
29562 * ipa-profile.c: Likewise.
29563 * ipa-prop.c: Likewise.
29564 * ipa-split.c: Likewise.
29566 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29567 edge_predicate_pool, dump_inline_hints,
29568 inline_summary::account_size_time, redirect_to_unreachable,
29569 edge_set_predicate, set_hint_predicate,
29570 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29571 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29572 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29573 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29574 ipa_call_summary_t::remove, initialize_growth_caches,
29575 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29576 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29577 mark_modified, unmodified_parm_1, unmodified_parm,
29578 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29579 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29580 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29581 will_be_nonconstant_predicate, record_modified_bb_info,
29582 get_minimal_bb, record_modified, param_change_prob,
29583 phi_result_unknown_predicate, predicate_for_phi_result,
29584 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29585 estimate_function_body_sizes, compute_inline_parameters,
29586 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29587 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29588 inline_update_callee_summaries, remap_edge_change_prob,
29589 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29590 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29591 inline_analyze_function, inline_summary_t::insert,
29592 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29593 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29594 inline_free_summary): Move to ipa-fnsummary.h
29595 (predicate_t): Remove.
29596 * ipa-fnsummary.c: New file.
29597 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29598 (enum inline_hints_vals, inline_hints, agg_position_info,
29599 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29600 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29601 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29602 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29603 inline_read_summary, inline_write_summary, inline_free_summary,
29604 inline_analyze_function, initialize_inline_failed,
29605 inline_merge_summary, inline_update_overall_summary,
29606 compute_inline_parameters): Move to ipa-fnsummary.h
29607 * ipa-fnsummary.h: New file.
29608 * ipa-inline-transform.h: Include ipa-inline.h.
29609 * ipa-inline.c: LIkewise.
29611 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29613 * ipa-inline.c (edge_badness): Use inlined_time instead of
29614 inline_summaries->get.
29616 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29618 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29620 2017-05-22 Nathan Sidwell <nathan@acm.org>
29622 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29623 (fdump-lang): Document 'raw' option.
29624 * dumpfile.h (TDI_tu): Delete.
29625 * dumpfile.c (dump_files): Remove translation-unit.
29626 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29628 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29630 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29631 command option from $(AWK) call.
29632 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29633 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29634 [FORMAT]: Remove handling of variable.
29635 * config/avr/t-multilib: Regenerate.
29637 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29639 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29641 (dump_inline_summary): Do not print self_time.
29642 (estimate_function_body_sizes): Do not set self_time.
29643 (compute_inline_parameters): Likewise.
29644 (inline_read_section, inline_write_summary): Do not stream self_time.
29645 * ipa-inline.h (inline_summary): Drop self_time.
29647 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29649 * ipa-inline-analysis.c (account_size_time): Rename to ...
29650 (inline_summary::account_size_time): ... this one.
29651 (reset_ipa_call_summary): Turn to ...
29652 (ipa_call_summary::reset): ... this one.
29653 (reset_inline_summary): Turn to ...
29654 (inline_summary::reset): ... this one.
29655 (inline_summary_t::remove): Update.
29656 (inline_summary_t::duplicate): Update.
29657 (ipa_call_summary_t::remove): Update.
29658 (dump_inline_summary): Update.
29659 (estimate_function_body_sizes): Update.
29660 (compute_inline_parameters): Update.
29661 (estimate_node_size_and_time): Update.
29662 (inline_merge_summary): Update.
29663 (inline_update_overall_summary): Update.
29664 (inline_read_section): Update.
29665 (inline_write_summary): Update.
29666 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29667 add account_size_time and reset member functions.
29668 (ipa_call_summary): Add reset function.
29669 * ipa-predicate.h (predicate::operator &): Constify.
29671 2017-05-22 Richard Biener <rguenther@suse.de>
29673 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29675 2017-05-19 Jason Merrill <jason@redhat.com>
29677 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29679 2017-05-19 Marek Polacek <polacek@redhat.com>
29682 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29683 TYPE_OVERFLOW_WRAPS checks.
29685 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29687 * tree-core.h (enum omp_clause_default_kind): Add
29688 "OMP_CLAUSE_DEFAULT_PRESENT".
29689 * tree-pretty-print.c (dump_omp_clause): Handle it.
29690 * gimplify.c (enum gimplify_omp_var_data): Add
29691 "GOVD_MAP_FORCE_PRESENT".
29692 (gimplify_adjust_omp_clauses_1): Map it to
29693 "GOMP_MAP_FORCE_PRESENT".
29694 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29696 * gimplify.c (oacc_default_clause): Clarify.
29698 2017-05-19 Nathan Sidwell <nathan@acm.org>
29700 LANG_HOOK_REGISTER_DUMPS
29701 * toplev.c (general_init): Call register dump lang hook.
29702 * doc/invoke.texi: Document -fdump-lang option family.
29703 * dumpfile.c (dump_files): Remove class dump here.
29704 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29705 * dumpfile.h (tree_dump_index): Remove TDI_class.
29706 * langhooks-def.h (lhd_register_dumps): Declare.
29707 (LANG_HOOKS_REGISTER_DUMPS): Define.
29708 (LANG_HOOKS_INITIALIZER): Add it.
29709 * langhooks.c (lhd_register_dumps): Define.
29710 * langhooks.h (struct lang_hooks): Add register_dumps.
29712 2017-05-19 Nathan Sidwell <nathan@acm.org>
29714 * context.h (context::set_passes): New.
29715 * context.c (context::context): Do not create pass manager.
29716 * toplev.c (general_init): Create pass manager here.
29718 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29720 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29721 use this splitter if two add or or instructions would also work for
29722 the constant we want to generate.
29724 2017-05-19 Richard Biener <rguenther@suse.de>
29727 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29728 predicate evaluation.
29730 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29732 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29734 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29737 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29739 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29740 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29741 (inline_edge_summary_vec): Turn into ...
29742 (ipa_call_summaries): ... this one.
29743 (redirect_to_unreachable, edge_set_predicate,
29744 evaluate_properties_for_edge, inline_summary_alloc,
29745 reset_ipa_call_summary, reset_inline_summary,
29746 inline_summary_t::duplicate): Update.
29747 (inline_edge_duplication_hook): Turn to ...
29748 (ipa_call_summary_t::duplicate): ... this one.
29749 (inline_edge_removal_hook): Turn to ...
29750 (ipa_call_summary_t::remove): ... this one.
29751 (dump_inline_edge_summary): Turn to ...
29752 (dump_ipa_call_summary): ... this one.
29753 (estimate_function_body_sizes): Update.
29754 (inline_update_callee_summaries): Update.
29755 (remap_edge_change_prob): Update.
29756 (remap_edge_summaries): Update.
29757 (inline_merge_summary): Update.
29758 (do_estimate_edge_time): Update.
29759 (inline_generate_summary): Update.
29760 (inline_read_section): Update.
29761 (inline_read_summary): Update.
29762 (inline_free_summary): Update.
29763 * ipa-inline.c (can_inline_edge_p): Update.
29764 (compute_inlined_call_time): Update.
29765 (want_inline_small_function_p): Update.
29766 (edge_badness): Update.
29767 (early_inliner): Update.
29768 * ipa-inline.h (inline_edge_summary): Turn to ...
29769 (ipa_call_summary): ... this one.
29770 (ipa_call_summary_t): New class.
29771 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29772 (ipa_call_summaries): New.
29773 (inline_edge_summary): Remove.
29774 (estimate_edge_growth): Update.
29775 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29776 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29777 * ipa-split.c (execute_split_functions): Update.
29778 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29780 2017-05-19 Richard Biener <rguenther@suse.de>
29782 PR middle-end/80764
29783 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29785 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29787 * config/rs6000/rs6000.c (struct machine_function): Add field
29788 fpr_is_wrapped_separately.
29789 (rs6000_get_separate_components): Use 64 components. Handle the
29790 new FPR components.
29791 (rs6000_components_for_bb): Handle the FPR components.
29792 (rs6000_emit_prologue_components): Handle the FPR components.
29793 (rs6000_emit_epilogue_components): Handle the FPR components.
29794 (rs6000_set_handled_components): Handle the FPR components.
29795 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29796 that are already separately shrink-wrapped.
29797 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29798 that are already separately shrink-wrapped.
29800 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29803 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29806 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29807 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29808 cases where the register allocator uses a move and an offsettable
29809 memory operation to/from a FPR register on ISA 2.06/2.07.
29810 (define_peephole2 for Altivec d-form store): Likewise.
29812 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29815 * config/i386/mmx.md (*mov<mode>_internal): Enable
29816 alternatives 11, 12, 13 and 14 also for 32bit targets.
29817 Remove alternatives 15, 16, 17 and 18.
29818 * config/i386/sse.md (vec_concatv2di): Change
29819 alternative (!x, *y) to (x, ?!*Yn).
29821 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29823 * dumpfile.h (enum dump_kind): Remove stray comma.
29825 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29827 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29828 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29829 predicate::num_conditions
29830 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29831 (CHANGED): turn into predicate::changed.
29832 (agg_position_info): Move to ipa-predicate.h
29833 (add_condition, predicate::add_clause, predicate::operator &=,
29834 predicate::or_with, predicate::evaluate, predicate::probability,
29835 dump_condition, dump_clause, predicate::dump,
29836 predicate::remap_after_duplication, predicate::remap_after_inlining,
29837 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29838 (evaluate_conditions_for_known_args): Update.
29839 (set_cond_stmt_execution_predicate): Update.
29840 * ipa-inline.h: Include ipa-predicate.h
29841 (condition, inline_param_summary, conditions, agg_position_info,
29842 predicate): Move to ipa-predicate.h
29843 * ipa-predicate.c: New file.
29844 * ipa-predicate.h: New file.
29846 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29848 * final.c (leaf_function_p): Check we are not in a sequence.
29850 2017-05-18 Martin Liska <mliska@suse.cz>
29852 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29853 * dumpfile.c (dump_register): Use new enum dump_kind.
29854 (get_dump_file_name): Likewise.
29855 (dump_enable_all): Likewise.
29856 (dump_switch_p_1): Likewise.
29857 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29858 * dumpfile.h (enum dump_kind): New enum type.
29859 (struct dump_file_info): Create constructor and
29860 format fields and comments.
29861 * passes.c (pass_manager::register_one_dump_file):
29863 * statistics.c (statistics_early_init): Likewise.
29864 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29865 TDF_TREE with TDF_SLIM.
29866 (gather_memory_references_ref): Likewise.
29868 2017-05-18 Martin Liska <mliska@suse.cz>
29870 * vec.h (struct vnull): Use it.
29872 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29874 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29875 (true_predicate, false_predicate, true_predicate_p,
29876 false_predicate_p): Remove.
29877 (single_cond_predicate, not_inlined_predicate): Turn to member function
29879 (add_condition): Update.
29880 (add_clause): Turn to...
29881 (predicate::add_clause): ... this one; update; allow passing NULL
29883 (and_predicates): Turn to ...
29884 (predicate::operator &=): ... this one.
29885 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29886 (or_predicates): Turn to ...
29887 (predicate::or_with): ... this one.
29888 (evaluate_predicate): Turn to ...
29889 (predicate::evaluate): ... this one.
29890 (predicate_probability): Turn to ...
29891 (predicate::probability): ... this one.
29892 (dump_condition): Update.
29893 (dump_predicate): Turn to ...
29894 (predicate::dump): ... this one.
29895 (account_size_time): Update.
29896 (edge_set_predicate): Update.
29897 (set_hint_predicate): UPdate.
29898 (evaluate_conditions_for_known_args): Update.
29899 (evaluate_properties_for_edge): Update.
29900 (remap_predicate_after_duplication): Turn to...
29901 (predicate::remap_after_duplication): ... this one.
29902 (remap_hint_predicate_after_duplication): Update.
29903 (inline_summary_t::duplicate): UPdate.
29904 (dump_inline_edge_summary): Update.
29905 (dump_inline_summary): Update.
29906 (set_cond_stmt_execution_predicate): Update.
29907 (set_switch_stmt_execution_predicate): Update.
29908 (compute_bb_predicates): Update.
29909 (will_be_nonconstant_expr_predicate): Update.
29910 (will_be_nonconstant_predicate): Update.
29911 (phi_result_unknown_predicate): Update.
29912 (predicate_for_phi_result): Update.
29913 (array_index_predicate): Update.
29914 (estimate_function_body_sizes): Update.
29915 (estimate_node_size_and_time): Update.
29916 (estimate_ipcp_clone_size_and_time): Update.
29917 (remap_predicate): Rename to ...
29918 (predicate::remap_after_inlining): ... this one.
29919 (remap_hint_predicate): Update.
29920 (inline_merge_summary): Update.
29921 (inline_update_overall_summary): Update.
29922 (estimate_size_after_inlining): Update.
29923 (read_predicate): Rename to ...
29924 (predicate::stream_in): ... this one.
29925 (read_inline_edge_summary): Update.
29926 (write_predicate): Rename to ...
29927 (predicate::stream_out): ... this one.
29928 (write_inline_edge_summary): Update.
29929 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29930 (clause_t): Turn to uint32_t
29931 (predicate): Turn to class; implement constructor and operators
29933 (size_time_entry): Update.
29934 (inline_summary): Update.
29935 (inline_edge_summary): Update.
29937 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29939 * fold-const.c (fold_binary_loc): Move transformation...
29940 * match.pd (C - X CMP X): ... here.
29942 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29944 * config/sparc/sparc.c (sparc_option_override): Set function
29945 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29946 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29948 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29950 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29952 2017-05-18 Marek Polacek <polacek@redhat.com>
29955 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29956 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29957 gimple_assign_load_p.
29959 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29961 PR middle-end/80692
29962 * real.c (do_compare): Give decimal_do_compare preference over
29963 comparing just the signs.
29965 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29967 * doc/md.texi (Canonicalization of Instructions): Describe the
29968 canonical form of instructions that inherently set a condition
29971 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29973 PR middle-end/80775
29974 * tree-cfg.c: Move deletion of unreachable case statements to after
29975 the merging of consecutive case labels.
29977 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29979 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29980 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29981 restoring of callee-saved registers.
29983 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29985 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29986 * config/visium/visium.c (single_set_and_flags): Likewise.
29987 * config/visium/visium.md (Substitutions): Likewise.
29989 2017-05-17 Martin Liska <mliska@suse.cz>
29991 * cfg.c: Introduce dump_flags_t type and
29992 use it instead of int type.
29994 * cfghooks.c: Likewise.
29995 * cfghooks.h (struct cfg_hooks): Likewise.
29996 * cfgrtl.c: Likewise.
29997 * cfgrtl.h: Likewise.
29998 * cgraph.c (cgraph_node::get_body): Likewise.
29999 * coretypes.h: Likewise.
30000 * domwalk.c: Likewise.
30001 * domwalk.h: Likewise.
30002 * dumpfile.c (struct dump_option_value_info): Likewise.
30003 (dump_enable_all): Likewise.
30004 (dump_switch_p_1): Likewise.
30005 (opt_info_switch_p): Likewise.
30006 * dumpfile.h (enum tree_dump_index): Likewise.
30007 (struct dump_file_info): Likewise.
30008 * genemit.c: Likewise.
30009 * generic-match-head.c: Likewise.
30010 * gengtype.c (open_base_files): Likewise.
30011 * gimple-pretty-print.c: Likewise.
30012 * gimple-pretty-print.h: Likewise.
30013 * graph.c (print_graph_cfg): Likewise.
30014 * graphite-scop-detection.c (dot_all_sese): Likewise.
30015 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30016 * loop-unroll.c (report_unroll): Likewise.
30017 * passes.c (pass_manager::register_one_dump_file): Likewise.
30018 * print-tree.c: Likewise.
30019 * statistics.c: Likewise.
30020 * tree-cfg.c: Likewise.
30021 * tree-cfg.h: Likewise.
30022 * tree-dfa.c: Likewise.
30023 * tree-dfa.h: Likewise.
30024 * tree-dump.c (dump_function): Likewise.
30025 * tree-dump.h (struct dump_info): Likewise.
30026 * tree-pretty-print.c: Likewise.
30027 * tree-pretty-print.h: Likewise.
30028 * tree-ssa-live.c: Likewise.
30029 * tree-ssa-live.h: Likewise.
30030 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30031 * tree-vect-loop.c: Likewise.
30032 * tree-vect-slp.c: Likewise.
30034 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30035 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30037 PR tree-optimization/80457
30038 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30039 of all arguments to a statement as scalar_to_vec operations.
30040 (vectorizable_call): Adjust call to vect_model_simple_cost for
30042 (vectorizable_conversion): Likewise.
30043 (vectorizable_assignment): Likewise.
30044 (vectorizable_shift): Likewise.
30045 (vectorizable_operation): Likewise.
30046 (vectorizable_comparison): Likewise.
30047 (vect_is_simple_cond): Record the def types for operands.
30048 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30049 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30050 for statement argument count.
30052 2017-05-16 Carl Love <cel@us.ibm.com>
30054 * config/rs6000/rs6000-c: Add support for built-in functions
30055 vector unsigned long long vec_bperm (vector unsigned long long,
30056 vector unsigned char)
30057 vector signed long long vec_mule (vector signed int,
30059 vector unsigned long long vec_mule (vector unsigned int,
30060 vector unsigned int)
30061 vector signed long long vec_mulo (vector signed int,
30063 vector unsigned long long vec_mulo (vector unsigned int,
30064 vector unsigned int)
30065 vector signed char vec_sldw (vector signed char,
30066 vector signed char,
30068 vector unsigned char vec_sldw (vector unsigned char,
30069 vector unsigned char,
30071 vector signed short vec_sldw (vector signed short,
30072 vector signed short,
30074 vector unsigned short vec_sldw (vector unsigned short,
30075 vector unsigned short,
30077 vector signed int vec_sldw (vector signed int,
30080 vector unsigned int vec_sldw (vector unsigned int,
30081 vector unsigned int,
30083 vector signed long long vec_sldw (vector signed long long,
30084 vector signed long long,
30086 vector unsigned long long vec_sldw (vector unsigned long long,
30087 vector unsigned long long,
30089 * config/rs6000/rs6000-c: Add support for built-in functions
30090 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30091 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30092 * doc/extend.texi: Update the built-in documentation for the
30093 new built-in functions.
30095 2017-05-16 Marek Polacek <polacek@redhat.com>
30099 * tree.c (save_expr): Don't fold the expression.
30101 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30103 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30104 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30105 and (?*y,m). Update insn attributes.
30107 2017-05-16 Martin Liska <mliska@suse.cz>
30109 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30110 flags argument of print_gimple_stmt, print_gimple_expr,
30111 print_generic_stmt and print_generic_expr.
30112 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30113 * coretypes.h: Likewise.
30114 * except.c (dump_eh_tree): Likewise.
30115 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30116 * gimple-pretty-print.h: Likewise.
30117 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30118 (backprop::push_to_worklist): Likewise.
30119 (backprop::pop_from_worklist): Likewise.
30120 (backprop::process_use): Likewise.
30121 (backprop::intersect_uses): Likewise.
30122 (note_replacement): Likewise.
30123 * gimple-ssa-store-merging.c
30124 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30125 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30126 (pass_store_merging::execute): Likewise.
30127 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30128 (ssa_base_cand_dump_callback): Likewise.
30129 (dump_incr_vec): Likewise.
30130 (replace_refs): Likewise.
30131 (replace_mult_candidate): Likewise.
30132 (create_add_on_incoming_edge): Likewise.
30133 (create_phi_basis): Likewise.
30134 (insert_initializers): Likewise.
30135 (all_phi_incrs_profitable): Likewise.
30136 (introduce_cast_before_cand): Likewise.
30137 (replace_one_candidate): Likewise.
30138 * gimplify.c (gimplify_expr): Likewise.
30139 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30140 (set_rename): Likewise.
30141 (rename_uses): Likewise.
30142 (copy_loop_phi_nodes): Likewise.
30143 (add_close_phis_to_merge_points): Likewise.
30144 (copy_loop_close_phi_args): Likewise.
30145 (copy_cond_phi_args): Likewise.
30146 (graphite_copy_stmts_from_block): Likewise.
30147 (translate_pending_phi_nodes): Likewise.
30148 * graphite-poly.c (print_pdr): Likewise.
30149 (dump_gbb_cases): Likewise.
30150 (dump_gbb_conditions): Likewise.
30151 (print_scop_params): Likewise.
30152 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30153 (build_cross_bb_scalars_use): Likewise.
30154 (gather_bbs::before_dom_children): Likewise.
30155 * hsa-dump.c (dump_hsa_immed): Likewise.
30156 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30157 (get_replacement_map): Likewise.
30158 * ipa-inline-analysis.c (dump_condition): Likewise.
30159 (estimate_function_body_sizes): Likewise.
30160 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30161 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30162 * ipa-prop.c (ipa_dump_param): Likewise.
30163 (ipa_print_node_jump_functions_for_edge): Likewise.
30164 (ipa_modify_call_arguments): Likewise.
30165 (ipa_modify_expr): Likewise.
30166 (ipa_dump_param_adjustments): Likewise.
30167 (ipa_dump_agg_replacement_values): Likewise.
30168 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30169 * ipa-pure-const.c (check_stmt): Likewise.
30170 (pass_nothrow::execute): Likewise.
30171 * ipa-split.c (execute_split_functions): Likewise.
30172 * omp-offload.c (dump_oacc_loop_part): Likewise.
30173 (dump_oacc_loop): Likewise.
30174 * trans-mem.c (tm_log_emit): Likewise.
30175 (tm_memopt_accumulate_memops): Likewise.
30176 (dump_tm_memopt_set): Likewise.
30177 (dump_tm_memopt_transform): Likewise.
30178 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30179 (print_loop): Likewise.
30180 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30181 (chkp_gather_checks_info): Likewise.
30182 (chkp_get_check_result): Likewise.
30183 (chkp_remove_check_if_pass): Likewise.
30184 (chkp_use_outer_bounds_if_possible): Likewise.
30185 (chkp_reduce_bounds_lifetime): Likewise.
30186 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30187 (chkp_mark_completed_bounds): Likewise.
30188 (chkp_register_incomplete_bounds): Likewise.
30189 (chkp_mark_invalid_bounds): Likewise.
30190 (chkp_maybe_copy_and_register_bounds): Likewise.
30191 (chkp_build_returned_bound): Likewise.
30192 (chkp_get_bound_for_parm): Likewise.
30193 (chkp_build_bndldx): Likewise.
30194 (chkp_get_bounds_by_definition): Likewise.
30195 (chkp_generate_extern_var_bounds): Likewise.
30196 (chkp_get_bounds_for_decl_addr): Likewise.
30197 * tree-chrec.c (chrec_apply): Likewise.
30198 * tree-data-ref.c (dump_data_reference): Likewise.
30199 (dump_subscript): Likewise.
30200 (dump_data_dependence_relation): Likewise.
30201 (analyze_overlapping_iterations): Likewise.
30202 * tree-inline.c (expand_call_inline): Likewise.
30203 (tree_function_versioning): Likewise.
30204 * tree-into-ssa.c (dump_defs_stack): Likewise.
30205 (dump_currdefs): Likewise.
30206 (dump_names_replaced_by): Likewise.
30207 (dump_update_ssa): Likewise.
30208 (update_ssa): Likewise.
30209 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30210 * tree-parloops.c (build_new_reduction): Likewise.
30211 (try_create_reduction_list): Likewise.
30212 (ref_conflicts_with_region): Likewise.
30213 (oacc_entry_exit_ok_1): Likewise.
30214 (oacc_entry_exit_single_gang): Likewise.
30215 * tree-pretty-print.h: Likewise.
30216 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30217 (get_scalar_evolution): Likewise.
30218 (add_to_evolution): Likewise.
30219 (get_loop_exit_condition): Likewise.
30220 (analyze_evolution_in_loop): Likewise.
30221 (analyze_initial_condition): Likewise.
30222 (analyze_scalar_evolution): Likewise.
30223 (instantiate_scev): Likewise.
30224 (number_of_latch_executions): Likewise.
30225 (gather_chrec_stats): Likewise.
30226 (final_value_replacement_loop): Likewise.
30227 (scev_const_prop): Likewise.
30228 * tree-sra.c (dump_access): Likewise.
30229 (disqualify_candidate): Likewise.
30230 (create_access): Likewise.
30231 (reject): Likewise.
30232 (maybe_add_sra_candidate): Likewise.
30233 (create_access_replacement): Likewise.
30234 (analyze_access_subtree): Likewise.
30235 (analyze_all_variable_accesses): Likewise.
30236 (sra_modify_assign): Likewise.
30237 (initialize_constant_pool_replacements): Likewise.
30238 (find_param_candidates): Likewise.
30239 (decide_one_param_reduction): Likewise.
30240 (replace_removed_params_ssa_names): Likewise.
30241 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30242 * tree-ssa-copy.c (dump_copy_of): Likewise.
30243 (copy_prop_visit_cond_stmt): Likewise.
30244 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30245 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30246 (record_equivalences_from_stmt): Likewise.
30247 * tree-ssa-dse.c (compute_trims): Likewise.
30248 (delete_dead_call): Likewise.
30249 (delete_dead_assignment): Likewise.
30250 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30251 (forward_propagate_into_cond): Likewise.
30252 (pass_forwprop::execute): Likewise.
30253 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30254 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30256 (move_computations_worker): Likewise.
30257 (execute_sm): Likewise.
30258 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30259 (remove_exits_and_undefined_stmts): Likewise.
30260 (remove_redundant_iv_tests): Likewise.
30261 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30262 (adjust_iv_update_pos): Likewise.
30263 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30264 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30265 (value_replacement): Likewise.
30266 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30267 * tree-ssa-pre.c (print_pre_expr): Likewise.
30268 (get_representative_for): Likewise.
30269 (create_expression_by_pieces): Likewise.
30270 (insert_into_preds_of_block): Likewise.
30271 (eliminate_insert): Likewise.
30272 (eliminate_dom_walker::before_dom_children): Likewise.
30273 (eliminate): Likewise.
30274 (remove_dead_inserted_code): Likewise.
30275 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30276 * tree-ssa-reassoc.c (get_rank): Likewise.
30277 (eliminate_duplicate_pair): Likewise.
30278 (eliminate_plus_minus_pair): Likewise.
30279 (eliminate_not_pairs): Likewise.
30280 (undistribute_ops_list): Likewise.
30281 (eliminate_redundant_comparison): Likewise.
30282 (update_range_test): Likewise.
30283 (optimize_range_tests_var_bound): Likewise.
30284 (optimize_vec_cond_expr): Likewise.
30285 (rewrite_expr_tree): Likewise.
30286 (rewrite_expr_tree_parallel): Likewise.
30287 (linearize_expr): Likewise.
30288 (break_up_subtract): Likewise.
30289 (linearize_expr_tree): Likewise.
30290 (attempt_builtin_powi): Likewise.
30291 (attempt_builtin_copysign): Likewise.
30292 (transform_stmt_to_copy): Likewise.
30293 (transform_stmt_to_multiply): Likewise.
30294 (dump_ops_vector): Likewise.
30295 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30296 (print_scc): Likewise.
30297 (set_ssa_val_to): Likewise.
30298 (visit_reference_op_store): Likewise.
30299 (visit_use): Likewise.
30300 (sccvn_dom_walker::before_dom_children): Likewise.
30301 (run_scc_vn): Likewise.
30302 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30304 (expr_hash_elt::print): Likewise.
30305 (const_and_copies::pop_to_marker): Likewise.
30306 (const_and_copies::record_const_or_copy_raw): Likewise.
30307 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30308 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30309 (dump_predicates): Likewise.
30310 (find_uninit_use): Likewise.
30311 (warn_uninitialized_phi): Likewise.
30312 (pass_late_warn_uninitialized::execute): Likewise.
30313 * tree-ssa.c (verify_vssa): Likewise.
30314 (verify_ssa): Likewise.
30315 (maybe_optimize_var): Likewise.
30316 * tree-vrp.c (dump_value_range): Likewise.
30317 (dump_all_value_ranges): Likewise.
30318 (dump_asserts_for): Likewise.
30319 (register_edge_assert_for_2): Likewise.
30320 (vrp_visit_cond_stmt): Likewise.
30321 (vrp_visit_switch_stmt): Likewise.
30322 (vrp_visit_stmt): Likewise.
30323 (vrp_visit_phi_node): Likewise.
30324 (simplify_cond_using_ranges_1): Likewise.
30325 (fold_predicate_in): Likewise.
30326 (evrp_dom_walker::before_dom_children): Likewise.
30327 (evrp_dom_walker::push_value_range): Likewise.
30328 (evrp_dom_walker::pop_value_range): Likewise.
30329 (execute_early_vrp): Likewise.
30331 2017-05-16 Richard Biener <rguenther@suse.de>
30333 * dwarf2out.c (loc_list_from_tree_1): Do not create
30334 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30336 2017-05-16 Richard Biener <rguenther@suse.de>
30338 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30340 (note_variable_value_in_expr): If we resolved the decl ref
30341 do not push to the stack.
30343 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30345 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30346 operations in fast-math mode.
30347 (vaddq_f16): Likewise.
30348 (vmul_f16): Likewise.
30349 (vmulq_f16): Likewise.
30350 (vsub_f16): Likewise.
30351 (vsubq_f16): Likewise.
30352 * config/arm/neon.md (add<mode>3): New.
30354 (fma:<VH:mode>3): New. Also remove outdated comment.
30357 2017-05-16 Martin Liska <mliska@suse.cz>
30361 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30362 (odr_types_equivalent_p): Likewise.
30364 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30366 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30368 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30371 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30372 non-interunit SSE move alternatives with '?'.
30373 (zero-extendsidi peephole2): New peephole to skip intermediate
30374 general register in SSE zero-extend sequence.
30376 2017-05-15 Jeff Law <law@redhat.com>
30378 * reorg.c (relax_delay_slots): Create a new variable to hold
30379 the temporary target rather than clobbering TARGET_LABEL.
30381 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30382 missing argument to extract_bit_field call.
30383 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30385 2017-05-15 Martin Liska <mliska@suse.cz>
30388 * gcc.c (process_command): Do not allow empty argument of -o option.
30390 2017-05-15 Renlin Li <renlin.li@arm.com>
30392 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30393 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30394 * config/aarch64/constraints.md (Usf): Add long call check.
30395 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30396 (call_value): Likewise.
30397 (sibcall): Likewise.
30398 (sibcall_value): Likewise.
30400 (call_value_insn): New.
30401 (sibcall_insn): Update rtx pattern.
30402 (sibcall_value_insn): Likewise.
30403 (call_internal): Remove.
30404 (call_value_internal): Likewise.
30405 (sibcall_internal): Likewise.
30406 (sibcall_value_internal): Likewise.
30407 (call_reg): Likewise.
30408 (call_symbol): Likewise.
30409 (call_value_reg): Likewise.
30410 (call_value_symbol): Likewise.
30412 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30415 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30417 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30419 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30420 compatible with CCGOCmode and with CCZmode.
30422 2017-05-14 Martin Sebor <msebor@redhat.com>
30424 PR middle-end/77671
30425 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30426 (gimple_fold_builtin_snprintf): Same.
30427 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30428 (gimple_fold_builtin_snprintf): Same.
30429 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30430 of character types.
30431 (is_call_safe): New function.
30432 (try_substitute_return_value): Call it.
30433 (try_simplify_call): New function.
30434 (pass_sprintf_length::handle_gimple_call): Call it.
30436 2017-05-14 Martin Sebor <msebor@redhat.com>
30438 PR middle-end/80669
30439 * builtins.c (expand_builtin_stpncpy): Simplify.
30441 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30443 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30444 * config/i386/i386.h
30445 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30446 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30447 (struct machine_function): Add new members call_ms2sysv,
30448 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30449 (struct machine_frame_state): New fields sp_realigned and
30450 sp_realigned_offset.
30451 * config/i386/i386.c
30452 (enum xlogue_stub): New enum.
30453 (enum xlogue_stub_sets): New enum.
30454 (class xlogue_layout): New class.
30455 (struct ix86_frame): New fields stack_realign_allocate_offset,
30456 stack_realign_offset and outlined_save_offset. Modify comments to
30457 detail stack layout when using out-of-line stubs.
30458 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30459 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30460 -mcall-ms2sysv-xlogues.
30461 (stub_managed_regs): New static variable.
30462 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30463 registers managed by out-of-line stub.
30464 (disable_call_ms2sysv_xlogues): New function.
30465 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30466 m->call_ms2sysv when appropriate and compute frame layout for
30468 (sp_valid_at, fp_valid_at): New inline functions.
30469 (choose_basereg): New function.
30470 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30472 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30473 Use align parameter of choose_baseaddr to generated aligned SSE movs
30475 (pro_epilogue_adjust_stack): Modify to track
30476 machine_frame_state::sp_realigned.
30477 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30478 (ix86_nsaved_sseregs): Likewise.
30479 (ix86_emit_save_regs): Likewise.
30480 (ix86_emit_save_regs_using_mov): Likewise.
30481 (ix86_emit_save_sse_regs_using_mov): Likewise.
30482 (get_scratch_register_on_entry): Likewise.
30483 (gen_frame_set): New function.
30484 (gen_frame_load): Likewise.
30485 (gen_frame_store): Likewise.
30486 (emit_outlined_ms2sysv_save): Likewise.
30487 (emit_outlined_ms2sysv_restore): Likewise.
30488 (ix86_expand_prologue): Modify stack re-alignment code and call
30489 emit_outlined_ms2sysv_save when appropriate.
30490 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30491 parameter rtx_insn *insn, which allows the function to be used to only
30492 generate the notes.
30493 (ix86_expand_epilogue): Modify validity checks of frame and stack
30494 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30495 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30496 * config/i386/predicates.md
30497 (save_multiple): New predicate.
30498 (restore_multiple): Likewise.
30499 * config/i386/sse.md
30500 (save_multiple<mode>): New pattern.
30501 (save_multiple_realign<mode>): Likewise.
30502 (restore_multiple<mode>): Likewise.
30503 (restore_multiple_and_return<mode>): Likewise.
30504 (restore_multiple_leave_return<mode>): Likewise.
30505 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30507 2017-05-14 Julia Koval <julia.koval@intel.com>
30509 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30510 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30511 (__builtin_ia32_xsetbv): New builtins.
30512 * config/i386/i386.c (ix86_expand_special_args_builtin):
30514 (ix86_expand_builtin): Special expand for new intrinsics.
30515 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30516 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30517 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30519 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30521 * cfganal.c (inverted_post_order_compute): Change argument type
30523 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30524 * df-core.c (rest_of_handle_df_initialize): Adjust.
30525 (rest_of_handle_df_finish): Likewise.
30526 (df_analyze_1): Likewise.
30527 (df_analyze): Likewise.
30528 (loop_inverted_post_order_compute): Change argument to be a vec *.
30529 (df_analyze_loop): Adjust.
30530 (df_get_n_blocks): Likewise.
30531 (df_get_postorder): Likewise.
30532 * df.h (struct df_d): Change field to be a vec.
30533 * lcm.c (compute_laterin): Adjust.
30534 (compute_available): Likewise.
30535 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30536 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30537 * tree-ssa-pre.c (compute_antic): Likewise.
30539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30541 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30542 (depth_first_search::depth_first_search): Change structure init
30543 function to this constructor.
30544 (depth_first_search::add_bb): Rename function to this member.
30545 (depth_first_search::execute): Likewise.
30546 (flow_dfs_compute_reverse_finish): Adjust.
30548 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30550 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30551 (longest_simple_path): Likewise.
30552 * shrink-wrap.c (spread_components): Likewise.
30553 (disqualify_problematic_components): Likewise.
30554 (emit_common_heads_for_components): Likewise.
30555 (emit_common_tails_for_components): Likewise.
30556 (insert_prologue_epilogue_for_components): Likewise.
30558 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30560 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30565 * df-core.c (df_set_blocks): Start using auto_bitmap.
30566 (df_compact_blocks): Likewise.
30567 * df-problems.c (df_rd_confluence_n): Likewise.
30568 * df-scan.c (df_insn_rescan_all): Likewise.
30569 (df_process_deferred_rescans): Likewise.
30570 (df_update_entry_block_defs): Likewise.
30571 (df_update_exit_block_uses): Likewise.
30572 (df_entry_block_bitmap_verify): Likewise.
30573 (df_exit_block_bitmap_verify): Likewise.
30574 (df_scan_verify): Likewise.
30575 * lra-constraints.c (lra_constraints): Likewise.
30576 (undo_optional_reloads): Likewise.
30577 (lra_undo_inheritance): Likewise.
30578 * lra-remat.c (calculate_gen_cands): Likewise.
30579 (do_remat): Likewise.
30580 * lra-spills.c (assign_spill_hard_regs): Likewise.
30581 (spill_pseudos): Likewise.
30582 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30583 (bitmap_set_subtract_values): Likewise.
30585 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30587 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30588 management with auto_bitmap.
30589 (fix_inter_tick): Likewise.
30590 (fix_recovery_deps): Likewise.
30591 * ira.c (add_store_equivs): Likewise.
30592 (find_moveable_pseudos): Likewise.
30593 (split_live_ranges_for_shrink_wrap): Likewise.
30594 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30595 (rtx_reuse_manager::seen_def_p): Likewise.
30596 (rtx_reuse_manager::set_seen_def): Likewise.
30597 * print-rtl.h (class rtx_reuse_manager): Likewise.
30599 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30601 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30603 (migrate_btr_def): Likewise.
30604 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30605 * df-core.c (loop_post_order_compute): Likewise.
30606 (loop_inverted_post_order_compute): Likewise.
30607 * hsa-common.h: Likewise.
30608 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30609 * init-regs.c (initialize_uninitialized_regs): Likewise.
30610 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30611 (inline_small_functions): Likewise.
30612 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30613 * ira.c (combine_and_move_insns): Likewise.
30614 (build_insn_chain): Likewise.
30615 * loop-invariant.c (find_invariants): Likewise.
30616 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30617 * predict.c (tree_predict_by_opcode): Likewise.
30618 (predict_paths_leading_to): Likewise.
30619 (predict_paths_leading_to_edge): Likewise.
30620 (estimate_loops_at_level): Likewise.
30621 (estimate_loops): Likewise.
30622 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30623 (spread_components): Likewise.
30624 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30625 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30626 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30627 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30628 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30629 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30630 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30631 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30632 (mark_threaded_blocks): Likewise.
30633 (thread_through_all_blocks): Likewise.
30634 * tree-ssa.c (verify_ssa): Likewise.
30635 (execute_update_addresses_taken): Likewise.
30636 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30640 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30642 (post_order_compute): Likewise.
30643 (inverted_post_order_compute): Likewise.
30644 (pre_and_rev_post_order_compute_fn): Likewise.
30646 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30648 * genrecog.c (int_set::int_set): Explicitly construct our
30649 auto_vec base class.
30650 * vec.h (auto_vec::auto_vec): New constructor.
30652 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30654 * bitmap.h (class auto_bitmap): New constructor taking
30655 bitmap_obstack * argument.
30657 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30659 * bitmap.h (class auto_bitmap): Change type of m_bits to
30660 bitmap_head, and adjust ctor / dtor and member operators.
30662 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30664 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30665 when returned register mode doesn't match original mode.
30667 2017-05-12 Jeff Law <law@redhat.com>
30668 Jakub Jelinek <jakub@redhat.com>
30670 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30671 we look for cc setter after the compare-elim changes.
30672 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30673 within the vector to match what compare-elim now expects.
30674 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30675 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30677 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30678 after the compare-elim changes.
30679 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30680 the vector to match what compare-elim now expects.
30681 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30682 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30683 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30684 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30685 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30687 * config/visium/visium.c (single_set_and_flags): Fix where
30688 we look for cc setter after the compare-elim changes.
30689 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30690 with the vector to match what compare-elim now expects.
30691 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30692 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30693 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30694 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30695 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30696 (neg<mode>2_insn_set_overflow): Likewise.
30698 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30700 PR middle-end/79794
30701 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30702 maybe_expand_insn call, set ops[0].target. If still set after call,
30703 set alt_rtl. Add extra arg to recursive calls.
30704 (extract_bit_field): Add alt_rtl argument. Pass to
30706 * expmed.h (extract_bit_field): Fix prototype.
30707 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30708 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30709 to extract_bit_field_calls.
30710 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30711 Pass alt_rtl to extract_bit_field calls.
30712 * calls.c (store_unaligned_arguments_into_psuedos)
30713 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30714 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30716 * optabs.h (struct expand_operand): Add target bitfield.
30718 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30720 * compare-elim.c (try_eliminate_compare): Canonicalize
30721 operation with embedded compare to
30722 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30723 (set (reg) (operation)].
30725 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30727 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30730 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30731 cost of adding a carry flag for ADC instruction.
30732 [case MINUS]: Ignore the cost of subtracting a carry flag
30733 for SBB instruction.
30735 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30737 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30739 * config/rs6000/bmiintrin.h: New file.
30740 * config/rs6000/bmi2intrin.h: New file.
30741 * config/rs6000/x86intrin.h: New file.
30743 2017-05-12 Jeff Law <law@redhat.com>
30745 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30748 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30750 PR middle-end/80707
30751 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30753 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30756 early expansion of vector divide builtins.
30757 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30758 builtins identified as having unsigned arguments.
30760 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30762 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30763 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30764 expansion of vector logical operations (and, andc, or, xor,
30767 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30769 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30770 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30772 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30775 early GIMPLE expansion of vector multiplies.
30777 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30779 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30780 TARGET_HAVE_MOVT conditional.
30781 (movt splitter): Likewise.
30783 2017-05-12 Richard Biener <rguenther@suse.de>
30785 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30786 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30787 Fold all stmts not inplace.
30789 2017-05-12 Richard Biener <rguenther@suse.de>
30791 PR tree-optimization/80713
30792 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30793 inserted_exprs bit for not removed stmts.
30795 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30797 PR middle-end/69921
30798 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30799 parallelized" attribute for parallelized OpenACC kernels.
30800 * omp-offload.c (execute_oacc_device_lower): Use it.
30802 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30803 Set "oacc kernels" attribute.
30804 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30805 parameter. Adjust all users.
30806 (oacc_fn_attrib_kernels_p): Remove function.
30807 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30808 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30809 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30810 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30811 assert "oacc kernels" attribute is set.
30813 2017-05-11 Carl Love <cel@us.ibm.com>
30815 * config/rs6000/rs6000-c: Add support for built-in functions
30816 vector unsigned char vec_popcnt (vector signed char)
30817 vector unsigned char vec_popcnt (vector unsigned char)
30818 vector unsigned short vec_popcnt (vector signed short)
30819 vector unsigned short vec_popcnt (vector unsigned short)
30820 vector unsigned int vec_popcnt (vector signed int)
30821 vector unsigned int vec_popcnt (vector unsigned int)
30822 vector unsigned long long vec_popcnt (vector signed long long)
30823 vector unsigned long long vec_popcnt (vector unsigned long long)
30824 vector signed long long vec_slo (vector signed long long,
30825 vector signed char)
30826 vector signed long long vec_slo (vector signed long long,
30827 vector unsigned char)
30828 vector unsigned long long vec_slo (vector unsigned long long,
30829 vector signed char)
30830 vector unsigned long long vec_slo (vector unsigned long long,
30831 vector unsigned char)
30832 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30833 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30834 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30835 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30836 * doc/extend.texi: Update the built-in documentation file for the
30837 new built-in functions.
30839 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30841 * attribs.h (sorted_attr_string): Move machine independent
30842 functions for target clone support from the i386 port to common
30843 code. Rename ix86_function_versions to common_function_versions.
30844 Rename make_name to make_unique_name.
30845 (common_function_versions): Likewise.
30846 (make_unique_name): Likewise.
30847 (make_dispatcher_decl): Likewise.
30848 (is_function_default_version): Likewise.
30849 * attribs.c (attr_strcmp): Likewise.
30850 (sorted_attr_string): Likewise.
30851 (common_function_versions): Likewise.
30852 (make_unique_name): Likewise.
30853 (make_dispatcher_decl): Likewise.
30854 (is_function_default_version): Likewise.
30855 * config/i386/i386.c (attr_strcmp): Likewise.
30856 (sorted_attr_string): Likewise.
30857 (ix86_function_versions): Likewise.
30858 (make_name): Likewise.
30859 (make_dispatcher_decl): Likewise.
30860 (is_function_default_version): Likewise.
30861 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30863 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30866 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30867 Account for direct move costs for vec_construct of integer
30870 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30873 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30874 (UNSPEC_STX_ATOMIC): Ditto.
30875 (loaddi_via_sse): New insn.
30876 (storedi_via_sse): Ditto.
30877 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30878 Update corresponding peephole2 patterns.
30879 (atomic_storedi_fpu): Ditto.
30881 2017-05-11 Julia Koval <julia.koval@intel.com>
30883 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30884 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30886 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30887 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30888 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30890 2017-05-11 Nathan Sidwell <nathan@acm.org>
30892 * graphite-poly.c: Include dumpfile.h.
30894 * dumpfle.h (dump_function): Declare here ...
30895 * tree-dump.h (dump_function): ... not here.
30896 * dumpfile.c: #include tree-cfg.h.
30897 (dump_function): Move here from ...
30898 * tree-dump.c (dump_function): ... here.
30899 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30900 * graphite-poly.c: Don't include tree-dump.h.
30901 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30902 * print-tree.c: Likewise.
30903 * stor-layout.c: Likewise.
30904 * tree-nested.c: Likewise.
30906 * dumpfile.c (dump_start): Use TDF_FLAGS.
30907 (dump_enable_all): Fix TDF_KIND check thinko.
30909 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30911 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30912 array entries to represent two legal parameterizations of the
30913 overloaded __builtin_cmpb function, as represented by the
30914 P6_OV_BUILTIN_CMPB constant.
30915 (altivec_resolve_overloaded_builtin): Add special case handling
30916 for the __builtin_cmpb function, as represented by the
30917 P6_OV_BUILTIN_CMPB constant.
30918 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30919 (BU_P6_64BIT_2): New macro.
30920 (BU_P6_OVERLOAD_2): New macro
30921 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30922 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30923 (CMPB): Add overload support to represent both 32-bit and 64-bit
30924 compare-bytes function.
30925 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30926 support for TARGET_CMPB.
30927 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30928 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30929 documentation of the __builtin_cmpb overloaded built-in function.
30931 2017-05-11 Richard Biener <rguenther@suse.de>
30933 PR tree-optimization/80705
30934 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30935 bases are not vectorizable.
30937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30939 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30940 when counting register pressure.
30942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30944 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30945 (iv_ca_more_deps): Renamed to ...
30946 (iv_ca_compare_deps): ... this.
30947 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30949 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30951 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30953 (determine_group_iv_costs): ... here.
30954 (find_inv_vars_cb): Record inv var if it's not recorded before.
30956 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30958 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30959 (get_shiftadd_cost): Ditto.
30961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30963 * tree-ssa-address.c: Include header file.
30964 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30966 (add_to_parts): Refactor.
30967 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30968 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30973 PR tree-optimization/53090
30974 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30976 (extract_cond_operands): Detect condition with IV on both sides
30977 and return COMP_IV_EXPR_2.
30978 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30979 (rewrite_use_compare): Simplify by removing call to function
30980 extract_cond_operands.
30982 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30984 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30985 (extract_cond_operands): Detect condition comparing against non-
30986 invariant bound and return appropriate enum value.
30987 (find_interesting_uses_cond): Update use of extract_cond_operands.
30988 Handle its return value accordingly.
30989 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30991 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30993 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30994 nonlinear iv_use computation in loop invariant sensitive way.
30996 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30998 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30999 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31001 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31003 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31004 (dump_cand): Support iv_cand.inv_exprs.
31005 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31007 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31010 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31012 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31014 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31015 as local function. Include necessary header files.
31016 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31018 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31020 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31024 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31025 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31026 RSHIFT_EXPR and BIT_NOT_EXPR.
31028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31030 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31031 (adjust_setup_cost): New parameter supporting round up adjustment.
31032 (struct address_cost_data): Delete.
31033 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31034 (split_address_cost, ptr_difference_cost): Delete.
31035 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31036 (struct ainc_cost_data): New struct.
31037 (get_address_cost_ainc): New function.
31038 (get_address_cost, get_computation_cost): Reimplement.
31039 (determine_group_iv_cost_address): Record inv_expr for all uses of
31041 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31042 (iv_ca_has_deps): Reimplemented to ...
31043 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31045 (iv_ca_extend): Call iv_ca_more_deps.
31047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31049 * tree-ssa-address.c (struct mem_address): Move to header file.
31050 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31051 * tree-ssa-address.h (struct mem_address): Move from C file.
31052 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31056 * tree-affine.h (aff_combination_type): New interface.
31057 (aff_combination_zero_p): Remove static.
31058 (aff_combination_const_p): New interface.
31059 (aff_combination_singleton_var_p): New interfaces.
31061 2017-05-11 Richard Biener <rguenther@suse.de>
31063 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31064 Skip unreachable blocks and destinations.
31065 (eliminate): Move stmt removal and fixup ...
31066 (fini_eliminate): ... here. Skip inserted exprs.
31067 (pass_pre::execute): Move fini_pre after fini_eliminate.
31068 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31069 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31070 PRE to get rid of dead code that has invalid SSA form and
31071 split critical edges again.
31073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31075 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31077 2017-05-11 Richard Biener <rguenther@suse.de>
31079 * passes.c (execute_function_todo): Verify loops if they are
31080 said to be up-to-date.
31081 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31082 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31084 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31087 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31088 handle calling assemble_external ourself.
31091 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31092 modes with zero size. Enhance comment.
31094 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31096 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31097 built-ins for vec_xl and vec_xst with short and char pointer
31100 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31102 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31103 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31104 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31105 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31106 (_mm_maskz_min_round_ss): New intrinsics.
31107 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31108 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31109 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31110 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31111 (__builtin_ia32_minss_mask_round): New builtins.
31112 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31113 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31114 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31116 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31117 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31119 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31122 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31124 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31125 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31126 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31127 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31128 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31129 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31130 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31131 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31132 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31133 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31134 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31135 (__builtin_ia32_mulss_mask_round): New builtins.
31136 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31138 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31140 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31141 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31143 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31146 2017-05-10 Julia Koval <julia.koval@intel.com>
31148 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31149 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31150 (_mm256_setr_m128i): New intrinsics.
31152 2017-05-10 Julia Koval <julia.koval@intel.com>
31154 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31155 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31156 (_mm_maskz_rcp14_ss): New intrinsics.
31157 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31158 (__builtin_ia32_rcp14ss_mask): New builtins.
31159 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31161 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31163 PR tree-optimization/51513
31164 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31165 (assert_unreachable_fallthru_edge_p): Use it.
31166 (group_case_labels_stmt): Likewise.
31167 * tree-cfg.h: Prototype it.
31168 * stmt.c: Include cfghooks.h and tree-cfg.h.
31169 (emit_case_dispatch_table) <gap_label>: New local variable.
31170 Use it to fill dispatch table gaps.
31171 Test for default_label before updating probabilities.
31172 (expand_case) <default_label>: Remove unneeded initialization.
31173 Test for unreachable default case statement and remove its edge.
31174 Set default_label accordingly.
31175 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31177 2017-05-10 Carl Love <cel@us.ibm.com>
31179 * config/rs6000/rs6000-c: Add support for built-in functions
31180 vector signed char vec_neg (vector signed char)
31181 vector signed short int vec_neg (vector short int)
31182 vector signed int vec_neg (vector signed int)
31183 vector signed long long vec_neg (vector signed long long)
31184 vector float vec_neg (vector float)
31185 vector double vec_neg (vector double)
31186 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31188 * config/rs6000/altivec.h: Add define for vec_neg
31189 * doc/extend.texi: Update the built-in documentation for the
31190 new built-in functions.
31192 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31194 PR tree-optimization/77644
31195 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31197 2017-05-10 Nathan Sidwell <nathan@acm.org>
31199 * dumpfile.h (TDI_lang_all): New.
31200 (TDF_KIND): New. Renumber others
31201 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31203 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31205 (get_dump_file_name): Adjust suffix generation.
31206 (dump_enable_all): Use TDF_KIND.
31207 * doc/invoke.texi (-fdump-lang-all): Document.
31209 * dumpfile.h: Tabify.
31211 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31214 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31215 Move member access before delete.
31217 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31219 * tree-inline.c (expand_call_inline): Split block at stmt
31222 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31225 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31226 are now unused after splitting mov{sf,sd}_hardfloat.
31227 (f32_lr2): Likewise.
31228 (f32_lm): Likewise.
31229 (f32_lm2): Likewise.
31230 (f32_li): Likewise.
31231 (f32_li2): Likewise.
31232 (f32_lv): Likewise.
31233 (f32_sr): Likewise.
31234 (f32_sr2): Likewise.
31235 (f32_sm): Likewise.
31236 (f32_sm2): Likewise.
31237 (f32_si): Likewise.
31238 (f32_si2): Likewise.
31239 (f32_sv): Likewise.
31240 (f32_dm): Likewise.
31241 (f32_vsx): Likewise.
31242 (f32_av): Likewise.
31243 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31244 For movsf, order stores so the VSX stores occur before the GPR
31245 store which encourages the register allocator to use a traditional
31246 FPR instead of a GPR. For movsd, order the stores so that the GPR
31247 store comes before the VSX stores to allow the power6 to work.
31248 This is due to the power6 not having a 32-bit integer store
31249 instruction from a FPR.
31250 (movsf_hardfloat): Likewise.
31251 (movsd_hardfloat): Likewise.
31253 2017-05-09 Martin Sebor <msebor@redhat.com>
31255 PR translation/80280
31256 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31259 PR translation/80280
31260 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31261 data member added in r247778.
31262 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31264 2017-05-09 Nathan Sidwell <nathan@acm.org>
31266 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31268 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31271 2017-05-09 Marek Polacek <polacek@redhat.com>
31273 * doc/invoke.texi: Fix typo.
31275 2017-05-09 Richard Biener <rguenther@suse.de>
31277 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31278 (vrp_val_is_min): Likewise.
31279 (set_value_range_to_value): Likewise.
31280 (set_value_range_to_nonnegative): Likewise.
31281 (gimple_assign_nonzero_p): Likewise.
31282 (gimple_stmt_nonzero_p): Likewise.
31283 (vrp_int_const_binop): Likewise. Remove unreachable case.
31284 (adjust_range_with_scev): Adjust comments.
31285 (compare_range_with_value): Likewise.
31286 (extract_range_from_phi_node): Likewise.
31287 (test_for_singularity): Likewise.
31289 2017-05-09 Richard Biener <rguenther@suse.de>
31291 * tree-vrp.c (get_single_symbol): Add assert that we don't
31292 get overflowed constants as invariant part.
31293 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31294 checks. Use wi::cmp instead of recursing for integer constants.
31295 (compare_values): Just ignore whether we assumed undefined
31296 overflow instead of failing the compare.
31297 (extract_range_for_var_from_comparison_expr): Add comment before the
31298 TREE_NO_WARNING sets.
31299 (test_for_singularity): Likewise.
31300 (extract_range_from_comparison): Do not disable optimization
31301 when we assumed undefined overflow.
31302 (extract_range_basic): Remove init of unused var.
31304 2017-05-09 Richard Biener <rguenther@suse.de>
31306 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31307 (extract_range_from_multiplicative_op_1): Adjust.
31308 (extract_range_from_binary_expr_1): Use int_const_binop.
31310 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31313 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31314 rs6000_store_data_bypass_p in seven define_bypass directives and
31315 in several comments.
31316 * config/rs6000/rs6000-protos.h: Add prototype for
31317 rs6000_store_data_bypass_p function.
31318 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31319 function implements slightly different (rs6000-specific) semantics
31320 than store_data_bypass_p, returning false rather than aborting
31321 with assertion error when arguments do not satisfy the
31322 requirements of store data bypass.
31323 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31324 rs6000_store_data_bypass_p.
31326 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31328 * config/xtensa/xtensa-protos.h
31329 (xtensa_initial_elimination_offset): New declaration.
31330 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31331 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31332 macro definition, add case for FRAME_POINTER_REGNUM when
31333 FRAME_GROWS_DOWNWARD.
31334 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31335 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31336 xtensa_initial_elimination_offset.
31338 2017-05-08 Nathan Sidwell <nathan@acm.org>
31340 * doc/invoke.texi: Alphabetize -fdump options.
31342 2017-05-08 Martin Sebor <msebor@redhat.com>
31344 PR translation/80280
31345 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31347 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31349 * target.def (compute_frame_layout): New optional target hook.
31350 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31351 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31352 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31354 * reload1.c (verify_initial_elim_offsets): Likewise.
31355 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31356 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31357 (arm_get_frame_offsets): Split up into this ...
31358 (arm_compute_frame_layout): ... and this function.
31360 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31362 * config/aarch64/constraints.md (Usa): New constraint.
31363 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31365 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31367 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31368 with_multilib_list after it has been checked.
31370 2017-05-08 Richard Biener <rguenther@suse.de>
31372 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31373 (bitmap_set_subtract_values): Likewise.
31375 2017-05-08 Richard Biener <rguenther@suse.de>
31377 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31378 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31380 (gimple_stmt_nonzero_warnv_p): Rename to ...
31381 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31383 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31384 (extract_range_basic): Adjust, do not disable propagation on
31385 strict overflow sensitive simplification.
31386 (vrp_visit_cond_stmt): Likewise.
31388 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31390 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31391 body size unconditionally.
31393 2017-05-07 Jeff Law <law@redhat.com>
31396 2017-05-06 Jeff Law <law@redhat.com>
31397 PR tree-optimization/78496
31398 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31401 PR tree-optimization/78496
31402 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31403 (simplify_stmt_using_ranges): Call it.
31404 (vrp_dom_walker::before_dom_children): Extract equivalences
31405 from an ASSERT_EXPR with an equality comparison against a
31408 2017-05-06 Jeff Law <law@redhat.com>
31410 PR tree-optimization/78496
31411 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31414 PR tree-optimization/78496
31415 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31416 (simplify_stmt_using_ranges): Call it.
31417 (vrp_dom_walker::before_dom_children): Extract equivalences
31418 from an ASSERT_EXPR with an equality comparison against a
31421 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31423 * lra-constraints.c (lra_copy_reg_equiv): New function.
31424 (split_reg): Use it to copy equivalence information from the
31425 original register to the spill register.
31427 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31429 PR rtl-optimization/75964
31430 * simplify-rtx.c (simplify_const_relational_operation): Remove
31431 invalid handling of comparisons of integer ABS.
31433 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31435 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31436 initialize to zero.
31437 (init_regs): Remove declaration.
31438 (function_arg_advance_32): Initialize error_p as boolean variable.
31440 2017-05-05 Nathan Sidwell <nathan@acm.org>
31442 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31443 lines. Use for (;;).
31445 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31447 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31448 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31449 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31450 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31451 VF=2 that require versioning.
31453 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31455 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31458 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31460 * diagnostic.h (diagnostic_override_option_index): Convert from
31461 macro to inline function.
31463 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31465 * diagnostic.c (last_module_changed_p): New function.
31466 (set_last_module): New function.
31467 (diagnostic_report_current_module): Convert macro usage to
31468 the above functions.
31469 * diagnostic.h (diagnostic_context::last_module): Strengthen
31470 from const line_map * to const line_map_ordinary *.
31471 (diagnostic_last_module_changed): Delete macro.
31472 (diagnostic_set_last_module): Delete macro.
31474 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31476 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31477 with diagnostic_report_diagnostic.
31478 (diagnostic_n_impl_richloc): Likewise.
31479 * diagnostic.h (report_diagnostic): Delete macro.
31480 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31481 with diagnostic_report_diagnostic.
31482 * substring-locations.c (format_warning_va): Likewise.
31484 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31486 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31487 save/restor of format_spec. Move option-printing code to...
31488 (print_option_information): ...this new function, and
31489 reimplement by simply printing to the pretty_printer,
31490 rather than appending to the format string.
31492 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31494 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31495 handling logic into...
31496 (update_effective_level_from_pragmas): ...this new function.
31498 2017-05-04 Andrew Waterman <andrew@sifive.com>
31500 * config/riscv/riscv.opt (mstrict-align): New option.
31501 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31502 (SLOW_UNALIGNED_ACCESS): Define.
31503 (riscv_slow_unaligned_access): Declare.
31504 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31506 (riscv_slow_unaligned_access): New variable.
31507 (rocket_tune_info): Set slow_unaligned_access to true.
31508 (optimize_size_tune_info): Set slow_unaligned_access to false.
31509 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31510 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31511 (riscv_option_override): Set riscv_slow_unaligned_access.
31512 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31514 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31516 * config/riscv/riscv.md: Unify indentation.
31518 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31523 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31525 (extendsi<mode>2): Add support for doing sign extension via
31526 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31527 don't have ISA 3.0 instructions.
31528 (extendsi<mode>2 splitter): Likewise.
31529 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31530 generate the normal insns since SImode can now go in vector
31531 registers. Disallow the special UNSPECs needed for previous
31532 machines to hide SImode being used. Add new insns
31533 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31534 (fix_trunc<mode>si2_stfiwx): Likewise.
31535 (fix_trunc<mode>si2_internal): Likewise.
31536 (fixuns_trunc<mode>si2): Likewise.
31537 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31538 (fctiw<u>z_<mode>_smallint): Likewise.
31539 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31540 of floating point to 32-bit integer from doing a direct move to
31541 the GPR registers to do a store.
31542 (fctiwz_<mode>): Break long line.
31544 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31546 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31547 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31548 (addr_list, addr_offset_valid_p): New.
31549 (split_address_groups): Check offset validity with above function.
31550 (gt-tree-ssa-loop-ivopts.h): Include header file.
31552 2017-05-05 Nathan Sidwell <nathan@acm.org>
31554 * config.gcc (arm*-*-*): Add missing 'fi'.
31556 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31558 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31559 included in -fopt-info does not matter.
31560 * doc/optinfo.texi (-fopt-info): Fix description of default
31561 behavour. Explicitly say order of options included in -fopt-info
31564 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31566 * config.gcc: Allow combinations of aprofile and rmprofile values for
31567 --with-multilib-list.
31568 * config/arm/t-multilib: New file.
31569 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31570 variables. Remove setting of ISA and floating-point ABI in
31571 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31572 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31573 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31574 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31576 * config/arm/t-rmprofile: Likewise except for the matches changes.
31577 * doc/install.texi (--with-multilib-list): Document the combination of
31578 aprofile and rmprofile values and warn about pitfalls in doing that.
31580 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31582 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31583 (movdi_aarch64): Likewise.
31585 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31587 PR tree-optimization/80632
31588 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31590 (build_arrays): Initialize it for virtual phis.
31591 (fix_phi_nodes): Use it for virtual phis.
31593 PR tree-optimization/80558
31594 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31595 [x, y] op z into [x op, y op z] for op & or | if conditions
31598 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31599 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31602 * config/arm/arm.md (use_literal_pool): Remove.
31603 (64-bit immediate split): No longer takes cost into consideration
31604 if arm_disable_literal_pool is enabled.
31605 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31606 used when arm_disable_literal_pool is enabled.
31607 (arm_max_const_double_inline_cost): Remove use of
31608 arm_disable_literal_pool.
31609 (push_minipool_fix): Add assert.
31610 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31611 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31612 (no_literal_pool_sf_immediate): New.
31614 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31616 PR tree-optimization/80613
31617 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31618 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31620 2017-05-05 Richard Biener <rguenther@suse.de>
31622 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31624 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31626 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31627 of this flag from insn conditions due to removal from r247495.
31629 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31631 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31633 (arm_early_store_addr_dep_ptr): Likewise.
31634 * config/arm/aarch-common-protos.h
31635 (arm_early_load_addr_dep_ptr): Add prototype.
31636 (arm_early_store_addr_dep_ptr): Likewise.
31637 * config/arm/cortex-a53.md: Add new bypasses.
31639 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31641 * tree.c (next_type_uid): Change type to unsigned.
31642 (type_hash_canon): Decrement back next_type_uid if
31643 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31644 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31647 2017-05-04 Martin Sebor <msebor@redhat.com>
31649 * builtins.c: Fix a trivial typo in a comment.
31651 PR middle-end/79234
31652 * builtins.c (check_sizes): Adjust to handle reading past the end.
31653 Avoid printing excessive upper bound of ranges. Use %E to print
31654 tree nodes instead of converting them to %wu.
31655 (expand_builtin_memchr): New function.
31656 (compute_dest_size): Rename...
31657 (compute_objsize): ...to this.
31658 (expand_builtin_memcpy): Adjust.
31659 (expand_builtin_mempcpy): Adjust.
31660 (expand_builtin_strcat): Adjust.
31661 (expand_builtin_strcpy): Adjust.
31662 (check_strncat_sizes): Adjust.
31663 (expand_builtin_strncat): Adjust.
31664 (expand_builtin_strncpy): Adjust and simplify.
31665 (expand_builtin_memset): Adjust.
31666 (expand_builtin_bzero): Adjust.
31667 (expand_builtin_memcmp): Adjust.
31668 (expand_builtin): Handle memcmp.
31669 (maybe_emit_chk_warning): Check strncat just once.
31671 2017-05-04 Martin Sebor <msebor@redhat.com>
31673 PR preprocessor/79214
31674 PR middle-end/79222
31675 PR middle-end/79223
31676 * builtins.c (check_sizes): Add inlining context and issue
31677 warnings even when -Wno-system-headers is set.
31678 (check_strncat_sizes): Same.
31679 (expand_builtin_strncat): Same.
31680 (expand_builtin_memmove): New function.
31681 (expand_builtin_stpncpy): Same.
31682 (expand_builtin): Handle memmove and stpncpy.
31684 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31686 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31687 which is not used any more.
31689 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31691 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31693 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31695 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31696 (cortexa53_tunings): Likewise.
31697 (cortexa57_tunings): Likewise.
31698 (cortexa72_tunings): Likewise.
31699 (cortexa73_tunings): Likewise.
31701 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31703 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31704 Set loop alignment to 8.
31706 2017-05-04 Martin Sebor <msebor@redhat.com>
31708 PR translation/80280
31709 * builtins.c (expand_builtin_object_size): Add missing quoting to
31710 %D and like directives.
31711 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31712 (hsa_type_for_tree_type): Same.
31713 (verify_function_arguments): Same.
31714 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31715 * varasm.c (get_section): Same.
31718 2017-05-04 Martin Sebor <msebor@redhat.com>
31720 PR translation/80280
31721 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31723 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31725 * config/aarch64/aarch64.c (generic_addrcost_table):
31726 Change HI/TI mode setting.
31728 2017-05-04 Martin Jambor <mjambor@suse.cz>
31730 PR tree-optimization/80622
31731 * tree-sra.c (comes_initialized_p): New function.
31732 (build_accesses_from_assign): Only set write lazily when
31733 comes_initialized_p is false.
31734 (analyze_access_subtree): Use comes_initialized_p.
31735 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31736 instead of testing for PARM_DECL.
31738 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31740 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31741 constraint on operand 0 to allow more general addressing modes.
31742 Adjust output template.
31743 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31745 * config/aarch64/aarch64-protos.h
31746 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31747 * config/aarch64/constraints.md (Dp): New address constraint.
31748 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31751 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31753 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31754 update use of estimate_ipcp_clone_size_and_time.
31755 (estimate_local_effects): Update use of
31756 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31757 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31758 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31759 Return nonspecialized time.
31761 2017-05-04 Richard Biener <rguenther@suse.de>
31763 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31764 for the last VUSE which def dominates the PHI. Directly call
31766 (get_continuation_for_phi_1): Remove.
31768 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31770 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31771 to explain the use of truncating division. Cap the number of
31772 iterations to the maximum given by nb_iterations_upper_bound,
31775 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31777 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31778 * configure: Regenerate.
31779 * config.in: Regenerate.
31780 * config/i386/driver-mingw32.c: new file.
31781 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31782 * config.host: Link driver-mingw32.o on MinGW host.
31783 * doc/install.texi: Document new --enable-mingw-wildcard configure
31786 2017-05-04 Marek Polacek <polacek@redhat.com>
31788 PR tree-optimization/80612
31789 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31791 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31792 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31794 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31795 (movt splitter): Likewise.
31796 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31797 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31798 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31799 block for Thumb-1 with MOVT.
31800 (thumb2_legitimate_address_p): Move code block ...
31801 (can_avoid_literal_pool_for_label_p): ... into this new function.
31802 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31804 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31805 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31806 "M-profile targets with the MOVT instruction".
31808 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31810 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31811 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31812 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31814 2017-05-04 Martin Liska <mliska@suse.cz>
31816 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31817 variable cond_code.
31819 2017-05-04 Richard Biener <rguenther@suse.de>
31821 * tree.c (array_at_struct_end_p): Handle arrays at struct
31822 end with flexarrays more conservatively. Refactor and treat
31823 arrays of arrays or aggregates more strict. Fix
31824 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31825 * tree.c (array_at_struct_end_p): Adjust prototype.
31826 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31827 * gimple-fold.c (get_range_strlen): Likewise.
31828 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31830 2017-05-04 Richard Biener <rguenther@suse.de>
31832 PR tree-optimization/31130
31833 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31835 (supports_overflow_infinity): Likewise.
31836 (is_negative_overflow_infinity): Likewise.
31837 (is_positive_overflow_infinity): Likewise.
31838 (is_overflow_infinity): Likewise.
31839 (stmt_overflow_infinity): Likewise.
31840 (overflow_infinity_range_p): Likewise.
31841 (usable_range_p): Remove as always returning true.
31842 (make_overflow_infinity): Remove.
31843 (negative_overflow_infinity): Likewise.
31844 (positive_overflow_infinity): Likewise.
31845 (avoid_overflow_infinity): Likewise.
31846 (set_value_range): Adjust accordingly.
31847 (set_value_range_to_nonnegative): Likewise, remove now unused
31848 overflow_infinity arg.
31849 (vrp_operand_equal_p): Adjust.
31850 (update_value_range): Likewise.
31851 (range_int_cst_singleton_p): Likewise.
31852 (operand_less_p): Likewise.
31853 (compare_values_warnv): Likewise.
31854 (extract_range_for_var_from_comparison_expr): Likewise.
31855 (vrp_int_const_binop): Likewise.
31856 (zero_nonzero_bits_from_vr): Likewise.
31857 (extract_range_from_multiplicative_op_1): Likewise.
31858 (extract_range_from_binary_expr_1): Likewise.
31859 (extract_range_from_unary_expr): Likewise.
31860 (extract_range_from_comparison): Likewise.
31861 (extract_range_basic): Likewise.
31862 (adjust_range_with_scev): Likewise.
31863 (compare_ranges): Likewise.
31864 (compare_range_with_value): Likewise.
31865 (dump_value_range): Likewise.
31866 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31868 (simplify_cond_using_ranges): Adjust.
31870 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31872 * brig-builtins.def: Added a builtin for class_f64.
31873 * builtin-types.def: Added a builtin type needed by class_f64.
31875 2017-05-03 Jason Merrill <jason@redhat.com>
31877 * timevar.def: Add TV_CONSTEXPR.
31879 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31881 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31883 2017-05-03 Martin Jambor <mjambor@suse.cz>
31885 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31886 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31887 * ipa-cp.c (ipcp_propagate_stage): Do not call
31888 ipa_update_after_lto_read.
31889 * ipa-inline.c (ipa_inline): Likewise.
31891 2017-05-03 Martin Jambor <mjambor@suse.cz>
31893 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31894 tag. Added a default constructor and a destructor.
31895 (ipa_edge_args_sum_t): New class;
31896 (ipa_edge_args_sum): Declare.
31897 (ipa_edge_args_vector): Remove declaration.
31898 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31899 (ipa_free_edge_args_substructures): Remove declaration.
31900 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31901 (ipa_edge_args_info_available_for_edge_p): Likewise.
31902 * ipa-prop.c (ipa_edge_args_vector): Removed.
31903 (edge_removal_hook_holder): Likewise.
31904 (edge_duplication_hook_holder): Likewise.
31905 (ipa_edge_args_sum): New variable.
31906 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31907 ipa_edge_args_vector.
31908 (ipa_free_edge_args_substructures): Likewise.
31909 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31910 ipa_edge_args_vector.
31911 (ipa_edge_removal_hook): Turned into method
31912 ipa_edge_args_sum_t::remove.
31913 (ipa_edge_duplication_hook): Turned into method
31914 ipa_edge_args_sum_t::duplicate.
31915 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31916 registering edge hooks.
31917 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31918 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31919 ipa_edge_args_sum instead of ipa_edge_args_vector.
31920 * ipa-profile.c (ipa_profile): Likewise.
31922 2017-05-03 Martin Jambor <mjambor@suse.cz>
31924 * symbol-summary.h (function_summary): New method exists.
31925 (function_summary::symtab_removal): Deallocate through release.
31926 (call_summary): New class.
31927 (gt_ggc_mx): New overload.
31928 (gt_pch_nx): Likewise.
31929 (gt_pch_nx): Likewise.
31931 2017-05-03 Jeff Law <law@redhat.com>
31933 PR tree-optimization/78496
31934 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31935 from simplify_cond_using_ranges. Split off code to walk
31936 backwards through casts into ...
31937 (simplify_cond_using_ranges_2): New function.
31938 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31939 (execute_vrp): After identifying jump threads, call
31940 simplify_cond_using_ranges_2.
31942 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31945 * ipa-inline.h (inline_summary): Add ctor.
31946 (create_ggc): Do not use ggc_cleared_alloc.
31948 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31949 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31951 * gcc.c (handle_braces): Support escaping in switch matching
31953 * doc/invoke.texi (Spec Files): Document it.
31954 Remove superfluous @code markup in items.
31956 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31958 * diagnostic-show-locus.c (struct column_range): New struct.
31959 (get_affected_columns): New function.
31960 (get_printed_columns): New function.
31961 (struct correction): New struct.
31962 (correction::ensure_capacity): New function.
31963 (correction::ensure_terminated): New function.
31964 (struct line_corrections): New struct.
31965 (line_corrections::~line_corrections): New dtor.
31966 (line_corrections::add_hint): New function.
31967 (layout::print_trailing_fixits): Reimplement in terms of the new
31969 (selftest::test_overlapped_fixit_printing): New function.
31970 (selftest::diagnostic_show_locus_c_tests): Call it.
31972 2017-05-03 Nathan Sidwell <nathan@acm.org>
31974 Canonicalize canonical type hashing
31975 * tree.h (type_hash_canon_hash): Declare.
31976 * tree.c (type_hash_list, attribute_hash_list): Move into
31977 type_hash_canon_hash.
31978 (build_type_attribute_qual_variant): Break out hash code calc into
31979 type_hash_canon_hash.
31980 (type_hash_canon_hash): New. Generic type hash computation.
31981 (build_range_type_1, build_array_type_1, build_function_type,
31982 build_method_type_directly, build_offset_type, build_complex_type,
31983 make_vector_type): Call it.
31985 2017-05-03 Richard Biener <rguenther@suse.de>
31987 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31988 When all DRs have unknown misaligned do not always peel
31989 when there is a store but apply the same costing model as if
31990 there were only loads.
31992 2017-05-03 Richard Biener <rguenther@suse.de>
31995 PR tree-optimization/80492
31996 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31997 compare_base_decls returning dont-know properly.
31999 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32001 * config/arm/iterators.md (CCSI): New mode iterator.
32002 (arch): New mode attribute.
32003 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32004 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32005 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32006 code iterator for success result mode.
32007 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32008 the corresponding new insn generators.
32010 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32013 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32014 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32016 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32018 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32019 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32020 (DDR_A): Wrap DDR argument in brackets.
32021 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32022 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32023 (DDR_REVERSED_P): Likewise.
32025 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32027 PR tree-optimization/79472
32028 * tree-switch-conversion.c (struct switch_conv_info): Add
32029 contiguous_range and default_case_nonstandard fields.
32030 (collect_switch_conv_info): Compute contiguous_range and
32031 default_case_nonstandard fields, don't clear final_bb if
32032 contiguous_range and only the default case doesn't have the required
32034 (check_all_empty_except_final): Set default_case_nonstandard instead
32035 of failing if contiguous_range and the default case doesn't have empty
32037 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32038 and only the default case doesn't have the required constants. Skip
32040 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32041 if default_case_nonstandard.
32042 (build_constructors): Build constant 1 just once. Assert that default
32043 values aren't inserted in between cases if contiguous_range. Skip
32045 (build_arrays): Skip virtual phis.
32046 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32047 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32048 Handle virtual phis.
32049 (gen_inbound_check): Handle default_case_nonstandard case.
32050 (process_switch): Adjust check_final_bb caller. Call
32051 gather_default_values with the first non-default case instead of
32052 default case if default_case_nonstandard.
32054 2017-05-02 Nathan Sidwell <nathan@acm.org>
32056 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32057 check. Fix formatting.
32059 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32061 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32062 errors when comparing specialized and unspecialized times.
32064 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32066 * diagnostic-show-locus.c
32067 (layout::should_print_annotation_line_p): Make private.
32068 (layout::print_annotation_line): Make private.
32069 (layout::annotation_line_showed_range_p): Make private.
32070 (layout::show_ruler): Make private.
32071 (layout::print_source_line): Make private. Pass in line and
32072 line_width, rather than calling location_get_source_line. Drop
32074 (layout::print_leading_fixits): New method.
32075 (layout::print_any_fixits): Rename to...
32076 (layout::print_trailing_fixits): ...this, and make private.
32077 Don't print newline fixits.
32078 (diagnostic_show_locus): Move logic for printing one row into...
32079 (layout::print_line): ...this new function. Move the
32080 location_get_source_line call and error-handling from
32081 print_source_line to here. Call print_leading_fixits, and rename
32082 print_any_fixits to print_trailing_fixits.
32083 (selftest::test_fixit_insert_containing_newline): Update now that
32084 newlines are partially supported.
32085 (selftest::test_fixit_insert_containing_newline_2): New test.
32086 (selftest::test_fixit_replace_containing_newline): Update comments.
32087 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32088 * edit-context.c (class added_line): New class.
32089 (class edited_line): Describe newline handling in comment.
32090 (edited_line::actually_edited_p): New method.
32091 (edited_line::print_content): Delete redundant decl.
32092 (edited_line::m_predecessors): New field.
32093 (edited_file::print_content): Call edited_line::print_content.
32094 (edited_file::print_diff): Update to support newlines.
32095 (edited_file::print_diff_hunk): Likewise.
32096 (edited_file::print_run_of_changed_lines): New function.
32097 (edited_file::print_diff_line): Convert to...
32098 (print_diff_line): ...this.
32099 (edited_file::get_effective_line_count): New function.
32100 (edited_line::edited_line): Initialize new field m_predecessors.
32101 (edited_line::~edited_line): Clean up m_predecessors.
32102 (edited_line::apply_fixit): Handle newlines.
32103 (edited_line::get_effective_line_count): New function.
32104 (edited_line::print_content): New function.
32105 (edited_line::print_diff_lines): New function.
32106 (selftest::test_applying_fixits_insert_containing_newline): New
32108 (selftest::test_applying_fixits_replace_containing_newline): New
32110 (selftest::insert_line): New function.
32111 (selftest::test_applying_fixits_multiple_lines): Add example of
32113 (selftest::edit_context_c_tests): Call the new tests.
32115 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32117 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32118 parameter cand. Update dump information.
32119 (get_computation_cost): Update uses.
32121 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32123 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32124 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32125 (get_computation_at, rewrite_use_address): Update use of
32126 get_computation_aff.
32128 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32130 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32131 (get_computation): Delete.
32132 (get_computation_cost): Implement like get_computation_cost_at.
32133 Use get_computation_at.
32134 (get_computation_cost_at): Delete.
32135 (rewrite_use_nonlinear_expr): Use get_computation_at.
32136 (rewrite_use_compare, remove_unused_ivs): Ditto.
32138 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32140 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32142 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32144 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32145 (ivopts_global_cost_for_size): Rename parameter and update uses.
32146 (iv_ca_recount_cost): Update uses.
32147 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32148 candidates seperately in n_invs and n_cands.
32149 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32151 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32153 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32154 (find_inv_vars_cb): New.
32155 (find_depends): Renamed to ...
32156 (find_inv_vars): ... this.
32157 (add_candidate_1, force_var_cost): Call find_inv_vars.
32158 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32160 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32162 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32163 inv_vars. Add inv_exprs.
32164 (struct iv_cand): Rename depends_on to inv_vars.
32165 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32166 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32167 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32168 (dump_cand): Dump inv_vars.
32169 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32170 (record_invariant, find_depends, add_candidate_1): Ditto.
32171 (set_group_iv_cost, force_var_cost): Ditto.
32172 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32173 (get_computation_cost_at, get_computation_cost): Ditto.
32174 (determine_group_iv_cost_generic): Ditto.
32175 (determine_group_iv_cost_address): Ditto.
32176 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32177 (determine_group_iv_costs): Ditto.
32178 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32179 (iv_ca_set_remove_invariants): Renamed to ...
32180 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32181 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32182 (iv_ca_set_add_invariants): Renamed to ...
32183 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32184 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32185 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32186 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32187 (create_new_ivs): Remove useless dump.
32189 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32191 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32193 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32194 (iv_ca_set_no_cp, create_new_iv): Ditto.
32196 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32198 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32200 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32202 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32203 function tree_check2.
32205 2017-05-02 Martin Liska <mliska@suse.cz>
32207 * doc/gcov.texi: Add missing preposition.
32208 * gcov.c (function_info::function_info): Properly fill up
32209 all member variables.
32211 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32213 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32215 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32217 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32219 2017-05-02 Martin Liska <mliska@suse.cz>
32222 * lto-streamer-in.c (lto_read_tree_1): Remove
32223 LTO_STREAMER_DEBUG.
32224 * lto-streamer.c (struct tree_hash_entry): Likewise.
32225 (struct tree_entry_hasher): Likewise.
32226 (tree_entry_hasher::hash): Likewise.
32227 (tree_entry_hasher::equal): Likewise.
32228 (lto_streamer_init): Likewise.
32229 (lto_orig_address_map): Likewise.
32230 (lto_orig_address_get): Likewise.
32231 (lto_orig_address_remove): Likewise.
32232 * lto-streamer.h: Likewise.
32233 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32234 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32236 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32238 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32239 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32240 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32241 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32242 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32243 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32244 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32245 (mm_maskz_sub_ss): New intrinsics.
32246 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32247 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32248 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32249 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32250 (__builtin_ia32_subss_mask_round): New builtins.
32251 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32252 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32253 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32255 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32256 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32258 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32261 2017-05-02 Martin Jambor <mjambor@suse.cz>
32263 PR tree-optimization/78687
32264 * tree-sra.c (access): New field parent.
32265 (process_subtree_disqualification): New function.
32266 (disqualify_candidate): Call it.
32267 (build_accesses_from_assign): Reset write flag if creating an
32269 (build_access_subtree): Fill in parent field and also prpagate
32270 down grp_write flag.
32271 (create_artificial_child_access): New parameter set_grp_write, set
32272 grp_write to its value.
32273 (propagate_subaccesses_across_link): Also propagate grp_write flag
32275 (propagate_all_subaccesses): Push the closest parent back to work
32276 queue if add_access_to_work_queue returned true.
32278 2017-05-02 Richard Biener <rguenther@suse.de>
32280 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32281 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32282 -fstrict-overflow documentation.
32283 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32284 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32285 flag_strict_overflow.
32286 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32287 * lto-opts.c (lto_write_options): Do not stream it.
32288 * lto-wrapper.c (merge_and_complain): Do not handle it.
32289 * opts.c (default_options_table): Do not set -fstrict-overflow.
32290 (finish_options): Likewise do not clear it when sanitizing.
32291 * simplify-rtx.c (simplify_const_relational_operation): Do not
32292 test flag_strict_overflow.
32294 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32296 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32297 using enabled attribute.
32298 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32299 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32300 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32301 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32302 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32303 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32304 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32305 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32306 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32307 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32309 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32311 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32313 2017-05-02 Richard Biener <rguenther@suse.de>
32315 PR tree-optimization/80591
32317 2017-04-10 Richard Biener <rguenther@suse.de>
32319 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32322 2017-05-02 Richard Biener <rguenther@suse.de>
32324 PR tree-optimization/80549
32325 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32326 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32327 headers if they do not have a preheader.
32329 2017-05-02 Martin Liska <mliska@suse.cz>
32332 * common.opt: Fix typo.
32333 * doc/invoke.texi: Likewise.
32335 2017-05-01 Jan Beulich <jbeulich@suse.com>
32337 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32338 swapping, add (x,x,m,x,n) alternative.
32340 2017-05-01 Nathan Sidwell <nathan@acm.org>
32342 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32343 unnecessary unadjusted_alignment check.
32345 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32348 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32350 * gimplify.c (gimplify_cilk_detach): New function.
32351 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32352 * tree-core.h: Document EXPR_CILK_SPAWN.
32353 * tree.h (EXPR_CILK_SPAWN): Define.
32355 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32357 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32358 to use new fixit_hint representation, using the "replace" logic.
32359 (get_line_span_for_fixit_hint): Likewise.
32360 (layout::print_any_fixits): Likewise.
32361 (selftest::test_one_liner_many_fixits): Rename to...
32362 (selftest::test_one_liner_many_fixits_1): ...this, and update
32363 comment and expected output to reflect that the multiple fix-it
32364 hints are now consolidated into one insertion.
32365 (selftest::test_one_liner_many_fixits_2): New test.
32366 (selftest::test_diagnostic_show_locus_one_liner): Update for
32368 (selftest::test_fixit_consolidation): Update for fix-it API
32370 * diagnostic.c (print_parseable_fixits): Likewise.
32371 * edit-context.c (edited_line::m_line_events): Convert from
32372 auto_vec <line_event *> to auto_vec <line_event>.
32373 (class line_event): Convert from abstract base class to a concrete
32374 class, taking over the role of replace_event.
32375 (class insert_event): Delete.
32376 (class replace_event): Rename to class line_event. Convert to
32378 (edit_context::add_fixits): Reimplement.
32379 (edit_context::apply_insert): Delete.
32380 (edit_context::apply_replace): Rename to...
32381 (edit_context::apply_fixit): ...this. Convert to half-open range.
32382 (edited_file::apply_insert): Delete.
32383 (edited_file::apply_replace): Rename to...
32384 (edited_file::apply_fixit): ...this.
32385 (edited_line::~edited_line): Drop deletion of events.
32386 (edited_line::apply_insert): Delete.
32387 (edited_line::apply_replace): Rename to...
32388 (edited_line::apply_fixit): ...this. Convert to half-open range.
32389 Update for change to type of m_line_events.
32390 * edit-context.h (edit_context::apply_insert): Delete.
32391 (edit_context::apply_replace): Rename to...
32392 (edit_context::apply_fixit): ...this.
32394 2017-05-01 Martin Sebor <msebor@redhat.com>
32396 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32399 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32402 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32403 __get_cpuid_max returns 0.
32404 (__get_cpuid_count): Ditto.
32406 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32408 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32409 replacement expression is another instance of one of its arguments.
32411 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32414 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32415 check for stack push/pop autoinc.
32416 * config/i386/i386.c (ix86_agi_dependent): Return false
32417 if the only reason why modified_in_p returned true is that
32418 addr is SP based and set_insn is a push or pop.
32420 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32422 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32425 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32428 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32429 (account_size_time): Use two predicates - exec_pred and
32431 (evaluate_conditions_for_known_args): Compute both clause and
32433 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32434 (inline_summary_t::duplicate): Update.
32435 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32437 (compute_inline_parameters): Likewise.
32438 (estimate_edge_size_and_time): Update caluclation of time.
32439 (estimate_node_size_and_time): Compute both time and nonspecialized
32441 (estimate_ipcp_clone_size_and_time): Update.
32442 (inline_merge_summary): Update.
32443 (do_estimate_edge_time): Update.
32444 (do_estimate_edge_size): Update.
32445 (do_estimate_edge_hints): Update.
32446 (inline_read_section, inline_write_summary): Stream both new predicates.
32447 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32449 (compute_inlined_call_time): Cleanup.
32450 (big_speedup_p): Update.
32451 (edge_badness): Update.
32452 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32453 (size_time_entry): Replace predicate by exec_predicate and
32454 nonconst_predicate.
32455 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32456 (estimate_edge_time): Return also nonspec_time.
32457 (reset_edge_growth_cache): Update.
32459 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32461 PR rtl-optimization/80491
32462 * ifcvt.c (noce_process_if_block): When looking for x setter
32463 with missing else_bb, don't check only the insn right before
32464 cond_earliest, but look for the last insn that x is modified in
32465 within the same bb.
32467 PR rtl-optimization/80491
32468 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32470 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32472 PR tree-optimization/80487
32473 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32475 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32477 PR tree-optimization/79697
32478 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32479 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32480 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32482 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32483 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32485 2017-04-28 Martin Sebor <msebor@redhat.com>
32487 PR tree-optimization/80523
32488 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32489 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32491 (maybe_warn, format_directive, parse_directive): Use new functions.
32492 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32494 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32496 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32498 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32500 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32501 target_header_dir): Set correctly.
32502 * configure: Regenerated.
32503 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32504 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32505 instead of SYSTEM_HEADER_DIR.
32507 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32509 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32510 (estimate_local_effects): Likewise.
32511 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32512 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32513 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32514 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32515 do_estimate_edge_time, estimate_edge_time): Likewise.
32516 * ipa-inline-analysis.c (estimate_node_size_and_time,
32517 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32518 (estimate_time_after_inlining): Remove.
32520 2017-04-28 Martin Liska <mliska@suse.cz>
32522 * doc/gcov.texi: Enhance documentation of gcov.
32524 2017-04-28 Martin Liska <mliska@suse.cz>
32526 * doc/gcov.texi: Sort options in alphabetic order.
32527 * doc/gcov-dump.texi: Likewise.
32528 * doc/gcov-tool.texi: Likewise.
32529 * gcov.c (print_usage): Likewise.
32530 * gcov-dump.c (print_usage): Likewise.
32531 * gcov-tool.c (print_merge_usage_message): Likewise.
32532 (print_rewrite_usage_message): Likewise.
32533 (print_overlap_usage_message): Likewise.
32535 2017-04-28 Martin Liska <mliska@suse.cz>
32537 PR gcov-profile/53915
32538 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32540 2017-04-28 Martin Liska <mliska@suse.cz>
32542 PR gcov-profile/79891
32543 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32544 is marked by compiler as living on a line.
32545 (get_cycles_count): Remove usage of the union.
32546 (output_intermediate_file): Likewise.
32547 (find_source): Fix GNU coding style.
32548 (accumulate_line_counts): Remove old non-all block mode.
32549 (output_lines): Remove usage of the union.
32550 * profile.c (output_location): Include all BBs, even if
32551 belonging to a same line (and file) as a previous BB.
32553 2017-04-28 Martin Liska <mliska@suse.cz>
32555 * gcov.c (process_args): Handle new argument 'w'.
32556 (read_graph_file): Assign ID to BBs.
32557 (output_branch_count): Display BB # if verbose flag is set.
32558 (output_lines): Likewise for arcs.
32559 (print_usage): Add '--verbose' option help.
32560 * doc/gcov.texi: Document --verbose (-w) option.
32562 2017-04-28 Martin Liska <mliska@suse.cz>
32564 * gcov.c (struct block_location_info): New struct.
32565 (process_file): Fill up the new structure.
32566 (read_graph_file): Replace usage of encoding by the newly added
32568 (add_line_counts): Likewise.
32569 (accumulate_line_counts): Remove usage of the union.
32570 (function_info::function_info): New function.
32571 (function_info::~function_info): Likewise.
32572 (process_file): Call delete instead of release_function.
32573 (release_function): Release the function.
32574 (release_structures): Call delete instead of release_function.
32575 (solve_flow_graph): Replace usage of num_blocks.
32576 (find_exception_blocks): Likewise.
32577 (output_lines): Fix GNU coding style.
32579 2017-04-28 Martin Liska <mliska@suse.cz>
32582 * coverage.c (coverage_remove_note_file): New function.
32583 * coverage.h: Declare the function.
32584 * toplev.c (finalize): Clean if an error has been seen.
32586 2017-04-28 Martin Liska <mliska@suse.cz>
32588 PR gcov-profile/80031
32589 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32590 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32591 * gcov.c (read_graph_file): Read just number of blocks.
32592 * profile.c (branch_prob): Do not stream 0 flags per a basic
32595 2017-04-28 Martin Liska <mliska@suse.cz>
32597 * gcov-dump.c (tag_*): Add new argument to declarations.
32598 (dump_gcov_file): Likewise.
32599 (tag_blocks): Add and use new argument depth.
32600 (tag_arcs): Likewise.
32601 (tag_lines): Likewise.
32602 (tag_counters): Likewise.
32603 (tag_summary): Likewise.
32604 (dump_working_sets): Use depth to do a proper indentation.
32606 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32609 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32610 * symtab.c (symtab_node::debug_symtab): Move definition here.
32612 2017-04-28 Richard Biener <rguenther@suse.de>
32614 * lto-streamer.h (LTO_major_version): Bump to 7.
32616 2017-04-28 Richard Biener <rguenther@suse.de>
32618 * tree-vrp.c (assert_info): New struct.
32619 (add_assert_info): New helper.
32620 (register_edge_assert_for_2): Refactor to add asserts to a vector
32622 (register_edge_assert_for_1): Likewise.
32623 (register_edge_assert_for): Likewise.
32624 (finish_register_edge_assert_for): New helper actually registering
32625 asserts where live on edge.
32626 (find_conditional_asserts): Adjust.
32627 (find_switch_asserts): Likewise.
32628 (evrp_dom_walker::try_find_new_range): Generalize.
32629 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32631 2017-04-27 Marek Polacek <polacek@redhat.com>
32634 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32635 arg10 and arg11 to itype.
32637 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32639 * doc/extend.texi (Object Size Checking): Improve grammar.
32641 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32644 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32645 that the logic for permitting reciprocal estimates matches that
32648 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32651 * tree.c (type_cache_hasher::equal): Only compare
32652 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32653 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32654 non-aggregate element types.
32655 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32656 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32658 2017-04-27 Richard Biener <rguenther@suse.de>
32660 PR middle-end/80533
32661 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32662 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32663 keeping a reference to a trailing array.
32665 2017-04-27 Richard Biener <rguenther@suse.de>
32667 PR middle-end/80539
32668 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32669 being in loop-closed SSA form conservatively.
32670 (chrec_fold_multiply_poly_poly): Likewise.
32672 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32674 PR middle-end/79665
32675 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32676 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32678 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32681 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32682 (aarch64_function_arg_alignment): Return unsigned int again, but still
32683 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32684 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32685 Don't emit -Wpsabi note.
32686 (aarch64_function_arg_boundary): Likewise.
32687 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32690 2017-04-26 Nathan Sidwell <nathan@acm.org>
32692 * tree.h (crc32_unsigned_n): Declare.
32693 (crc32_unsigned, crc32_unsigned): Make inline.
32694 * tree.c (crc32_unsigned_bits): Replace with ...
32695 (crc32_unsigned_n): ... this.
32696 (crc32_unsigned, crc32_byte): Remove.
32697 (crc32_string): Remove unnecessary braces.
32699 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32701 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32702 * ipa-inline-analysis.c (MAX_TIME): Remove.
32703 (account_size_time): Use sreal for time.
32704 (dump_inline_summary): Update.
32705 (estimate_function_body_sizes): Update.
32706 (estimate_edge_size_and_time): Update.
32707 (estimate_calls_size_and_time): Update.
32708 (estimate_node_size_and_time): Update.
32709 (inline_merge_summary): Update.
32710 (inline_update_overall_summary): Update.
32711 (estimate_time_after_inlining): Update.
32712 (inline_read_section): Update.
32713 (inline_write_summary): Update.
32714 * ipa-inline.c (compute_uninlined_call_time): Update.
32715 (compute_inlined_call_time): Update.
32716 (recursive_inlining): Update.
32717 (inline_small_functions): Update.
32718 (dump_overall_stats): Update.
32719 * ipa-inline.h: Include sreal.h.
32720 (size_time_entry): Turn time to sreal.
32721 (inline_summary): Turn self_time nad time to sreal.
32723 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32725 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32727 (sreal::stream_out, sreal::stream_in): New.
32728 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32730 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32732 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32735 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32738 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32740 (dimode_scalar_chain::compute_convert_gain): Ditto.
32741 (dimode_scalar_chain::make_vector_copies): Ditto.
32742 (dimode_scalar_chain::convert_reg): Ditto.
32743 (dimode_scalar_chain::convert_insn): Ditto.
32744 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32745 (VI248_AVX512BW_1): New mode iterator.
32746 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32747 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32750 2017-04-25 Martin Sebor <msebor@redhat.com>
32752 PR tree-optimization/80497
32753 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32754 constants are representable in HOST_WIDE_INT.
32755 (parse_directive): Ditto.
32757 2017-04-25 Martin Sebor <msebor@redhat.com>
32760 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32761 (new_zero_array): Adjust signature.
32762 (dom_info::dom_init): Used unsigned rather that size_t.
32763 (dom_info::dom_info): Same.
32765 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32766 Jakub Jelinek <jakub@redhat.com>
32769 * config/arm/arm.c: Include gimple.h.
32770 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32771 returns negative, increment ncrn only if it returned positive.
32772 (arm_needs_doubleword_align): Return int instead of bool,
32773 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32774 members, but if there is any such non-FIELD_DECL
32775 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32776 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32777 returns negative, increment nregs only if it returned positive.
32778 (arm_setup_incoming_varargs): Likewise.
32779 (arm_function_arg_boundary): Emit -Wpsabi note if
32780 arm_needs_doubleword_align returns negative, return
32781 DOUBLEWORD_ALIGNMENT only if it returned positive.
32783 2017-04-25 Marek Polacek <polacek@redhat.com>
32786 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32787 first argument to type.
32789 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32792 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32793 type checks to test for compatibility instead of equality.
32795 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32796 Jakub Jelinek <jakub@redhat.com>
32799 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32801 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32802 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32803 the alignment computation, but return their maximum in warn_alignment.
32804 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32805 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32807 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32808 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32811 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32813 * config/arc/simdext.md (dmpyh): Fix typo.
32815 2017-04-25 Richard Biener <rguenther@suse.de>
32817 PR tree-optimization/80492
32818 * alias.c (compare_base_decls): Handle registers with asm
32819 specification conservatively.
32820 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32821 compare_base_decls returning dont-know properly.
32823 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32825 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32826 (legitimate_offset_address_p): New function.
32827 (arc_legitimate_address_p): Use above function.
32829 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32831 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32833 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32835 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32836 ACCH registers whenever they are available.
32838 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32840 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32841 double regs fix when not used.
32843 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32845 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32847 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32848 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32850 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32852 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32853 when emitting short ADD instructions.
32855 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32857 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32859 (cmpsi_cc_c_insn): Likewise.
32860 (cbranchsi4_scratch): Compute proper instruction length using
32861 compact_hreg_operand.
32862 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32864 2017-04-25 Richard Biener <rguenther@suse.de>
32866 PR middle-end/80509
32867 * passes.c (pass_manager::pass_manager): Initialize
32868 m_name_to_pass_map.
32870 2017-04-25 Richard Biener <rguenther@suse.de>
32872 PR tree-optimization/79201
32873 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32875 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32878 * config/s390/vector.md: Split MEM->GPR vector moves for
32879 non-s_operand addresses.
32881 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32884 * config/s390/predicates.md (reload_const_wide_int_operand): New
32886 * config/s390/s390.md ("movti"): Remove d/P alternative.
32887 ("movti_bigconst"): New pattern definition.
32889 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32892 * s390-protos.h (s390_expand_cs_hqi): Removed.
32893 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32894 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32895 modes as well as CCZ1mode and CCZmode.
32896 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32897 signature of s390_emit_compare_and_swap.
32898 (s390_expand_cs_hqi): Likewise, make static.
32899 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32900 compare-and-swap, in some cases.
32901 (s390_expand_cs): Wrapper function.
32902 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32904 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32905 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32906 patterns for small and large integers. Forbid symref memory operands.
32907 Move expander to s390.c. Require cc register.
32908 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32909 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32910 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32911 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32912 symref memory operands. Remove CC mode and call s390_match_ccmode
32914 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32916 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32918 * config/s390/s390.md (define_peephole2): New peephole to help
32919 combining the load-and-test pattern with volatile memory.
32921 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32923 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32924 with CCZmode for TARGET_Z196.
32926 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32928 PR rtl-optimization/80501
32929 * combine.c (make_compound_operation_int): Set subreg_code to SET
32930 even for AND with mask of the sign bit of mode.
32932 PR rtl-optimization/80500
32933 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32934 sum's initial value.
32936 2017-04-25 Julian Brown <julian@codesourcery.com>
32937 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32939 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32941 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32943 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32945 2017-04-25 Julian Brown <julian@codesourcery.com>
32946 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32948 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32949 (thunderx2t99_sha): New Reservation.
32951 2017-04-25 Julian Brown <julian@codesourcery.com>
32952 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32954 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32955 type for 1-element load.
32957 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32959 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32961 2017-04-24 Martin Jambor <mjambor@suse.cz>
32963 PR tree-optimization/80293
32964 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32965 char arrays not totally scalarizable if it is false.
32966 (analyze_all_variable_accesses): Pass correct value in the new
32967 parameter. Add a statistics counter.
32969 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32971 PR middle-end/79931
32972 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32974 2017-04-24 Richard Biener <rguenther@suse.de>
32976 PR tree-optimization/80494
32977 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32978 out for complex types.
32980 2017-04-24 Richard Biener <rguenther@suse.de>
32982 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32983 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32984 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32985 (DFS): Adjust and never fail.
32986 (sccvn_dom_walker::fail): Remove.
32987 (sccvn_dom_walker::before_dom_children): Adjust.
32988 (run_scc_vn): Likewise and never fail.
32989 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32990 (pass_fre::execute): Likewise.
32992 2017-04-24 Richard Biener <rguenther@suse.de>
32994 PR tree-optimization/79725
32995 * tree-ssa-sink.c (statement_sink_location): Return whether
32996 failure reason was zero uses. Move that check later.
32997 (sink_code_in_bb): Deal with zero uses by removing the stmt
33000 2017-04-24 Richard Biener <rguenther@suse.de>
33003 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33004 pointer-based references.
33006 2017-04-24 Richard Biener <rguenther@suse.de>
33009 * pass_manager.h (pass_manager::operator new): Remove.
33010 (pass_manager::operator delete): Likewise.
33011 * passes.c (pass_manager::operator new): Remove.
33012 (pass_manager::operator delete): Likewise.
33013 (pass_manager::pass_manager): Zero individual pass members.
33015 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33018 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33019 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33020 Check "XEXP (src, 1)" operand here.
33021 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33022 Check "XEXP (src, 1)" operand here.
33023 (dimode_scalar_chain::make_vector_copies): Detect count register
33024 of a shift instruction. Zero extend count register from QImode
33025 to DImode to satisfy vector shift pattern count operand predicate.
33026 Substitute vector shift count operand with a DImode copy.
33027 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33030 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33032 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33033 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33034 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33035 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33036 (UNSPEC_NOREX_MEM): Remove definition.
33038 2017-04-21 Richard Biener <rguenther@suse.de>
33040 PR tree-optimization/79547
33041 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33042 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33043 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33044 without any constraints.
33046 2017-04-21 Richard Biener <rguenther@suse.de>
33048 PR tree-optimization/78847
33049 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33051 2017-04-21 Richard Biener <rguenther@suse.de>
33053 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33054 (build_distinct_type_copy): Likewise.
33055 (build_variant_type_copy): Likewise.
33056 * tree.c (build_qualified_type): Pass down mem-stat info.
33057 (build_distinct_type_copy): Likewise.
33058 (build_variant_type_copy): Likewise.
33060 2017-04-21 Richard Biener <rguenther@suse.de>
33062 PR tree-optimization/80237
33063 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33065 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33066 for a simplified result.
33068 2016-04-21 Richard Biener <rguenther@suse.de>
33070 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33071 sth as strict as a simple_iv but a chrec without symbols and an
33072 operand defined in the loop we are peeling (and not some subloop).
33073 (propagate_constants_for_unrolling): Propagate all constants.
33075 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33078 * config/i386/i386.c (print_reg): Remove assert for disalowed
33079 regno values, call output_operand_lossage instead.
33081 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33084 * config/i386/constraints.md (Yc): New register constraint.
33085 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33086 Use Yc constraint for alternative 2 of operand 0. Remove
33087 preferred_for_speed attribute.
33089 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33091 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33092 lastprivate clauses in SIMT case.
33094 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33096 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33098 2017-04-20 Richard Biener <rguenther@suse.de>
33100 PR tree-optimization/57796
33101 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33102 as N scalar stores.
33103 (vect_model_load_cost): Cost gathers as N scalar loads.
33105 2017-04-20 Richard Biener <rguenther@suse.de>
33107 * ggc-page.c (ggc_allocated_p): Rename to ...
33108 (safe_lookup_page_table_entry): ... this and return the lookup
33110 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33112 2017-04-20 Richard Biener <rguenther@suse.de>
33114 PR tree-optimization/80453
33115 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33116 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33117 from the conditions.
33118 (vn_phi_eq): Pass them down.
33119 (vn_phi_lookup): Record them.
33120 (vn_phi_insert): Likewise.
33122 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33124 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33125 uninitialized variable warning to avoid buffer overrun.
33127 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33130 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33131 is suppressed for '{ 0 }' in C.
33133 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33135 * BASE-VER: Set to 8.0.0.
33137 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33139 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33140 priority .init_array and .fini_array section with SECTION_NOTYPE
33143 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33145 PR middle-end/80423
33146 * tree.h (build_array_type): Add typeless_storage default argument.
33147 * tree.c (type_cache_hasher::equal): Also compare
33148 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33149 (build_array_type): Add typeless_storage argument, set
33150 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33152 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33153 (build_array_type): Likewise. Add typeless_storage argument.
33155 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33156 Jakub Jelinek <jakub@redhat.com>
33158 PR tree-optimization/80426
33159 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33160 operation on symbolic operands, also compute the overflow for the
33161 invariant part when the operation degenerates into a negation.
33163 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33166 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33167 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33170 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33172 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33175 * config/avr/avr.c (tree.h): Include it.
33176 (cgraph.h): Include it.
33177 (avr_encode_section_info): Don't warn for uninitialized progmem
33178 variable if it's just an alias.
33180 2017-04-19 Richard Biener <rguenther@suse.de>
33183 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33184 when needed by AutoPGO.
33186 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33189 * doc/lto.texi: Remove an extra 'that'.
33191 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33193 PR rtl-optimization/80429
33194 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33195 are only used in debug insns.
33197 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33198 Vladimir Makarov <vmakarov@redhat.com>
33200 * config/sparc/predicates.md (input_operand): Add comment. Return
33201 true for any memory operand when LRA is in progress.
33202 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33204 2017-04-18 Jeff Law <law@redhat.com>
33207 * mips.md ({return,simple_return}_internal): Do not overwrite
33210 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33212 PR tree-optimization/80443
33213 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33214 instead of adding 1, subtract -1 and similarly instead of subtracting
33217 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33219 PR rtl-optimization/80357
33220 * haifa-sched.c (tmp_bitmap): New variable.
33221 (model_recompute): Handle duplicate use records.
33222 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33223 (free_global_sched_pressure_data): Free it.
33225 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33228 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33229 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33230 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33231 instead of SYSTEM_HEADER_DIR.
33233 2017-04-18 Jeff Law <law@redhat.com>
33235 PR middle-end/80422
33236 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33237 predecessors after walking up the insn chain.
33239 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33242 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33243 sizetype type into debug info.
33245 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33248 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33249 unneeded test for TARGET_UPPER_REGS_SF.
33250 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33252 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33255 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33256 instead of gsi_after_labels.
33258 2017-04-18 Jeff Law <law@redhat.com>
33260 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33264 2017-04-13 Jeff Law <law@redhat.com>
33265 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33266 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33268 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33271 * config/avr/avr.c (intl.h): Include it.
33272 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33274 2017-04-18 Martin Liska <mliska@suse.cz>
33276 PR gcov-profile/78783
33277 * gcov-tool.c (gcov_output_files): Validate that destination
33278 file is either removed by the tool or by a user.
33280 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33281 Guy Benyei <guybe@mellanox.com>
33283 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33284 block, and do not negate it, the stored id is already negative.
33286 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33288 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33290 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33293 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33294 masks of options that should be turned off if the VSX vector
33295 options are turned off.
33296 (OTHER_P8_VECTOR_MASKS): Likewise.
33297 (OTHER_VSX_VECTOR_MASKS): Likewise.
33298 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33299 rs6000_disable_incompatible_switches to validate no type switches
33301 (rs6000_incompatible_switch): New function to disallow turning on
33302 other vector options if -mno-vsx, -mno-power8-vector, or
33303 -mno-power9-vector are specified.
33305 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33307 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33309 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33311 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33312 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33313 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33314 (ARG_POINTER_CFA_OFFSET): Likewise.
33316 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33318 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33319 conditions to take advantage of various optimizations.
33321 2017-04-13 Jeff Law <law@redhat.com>
33323 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33324 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33325 (zero_extendsidi2_dext): Likewise.
33327 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33330 * fold-const.c (fold_ternary_loc): Revert
33331 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33334 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33336 PR rtl-optimization/80343
33337 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33338 new scratch pseudo.
33340 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33343 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33344 to ubsan_encode_value.
33346 2017-04-13 Jeff Law <law@redhat.com>
33348 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33349 appearing in DEBUG_INSNs.
33351 2017-04-13 Martin Liska <mliska@suse.cz>
33353 PR gcov-profile/80413
33354 * gcov-io.c (gcov_write_string): Copy to buffer just when
33355 allocated size is greater than zero.
33357 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33360 * dwarf2out.c (decls_for_scope): Ignore declarations of
33361 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33363 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33366 * ipa-visibility.c (non_local_p): Fix typos.
33367 (localize_node): When localizing symbol in same comdat group,
33368 dissolve the group only when we know external symbols are going
33370 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33372 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33374 PR tree-optimization/79390
33375 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33376 order does not result in usable sequence, retry with reversed operand
33382 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33383 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33384 op0 instead of fold_convert_loc (loc, type, arg0).
33386 2017-04-12 Jeff Law <law@redhat.com>
33388 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33389 has a delay slot in the generated code.
33391 * config/cris/cris.md (cris_preferred_reload_class): Return
33392 GENNONACR_REGS rather than GENERAL_REGS.
33394 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33397 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33398 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33399 signedness of the result type.
33401 2017-04-12 Richard Biener <rguenther@suse.de>
33402 Jeff Law <law@redhat.com>
33404 PR tree-optimization/80359
33405 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33406 trim stores to TARGET_MEM_REFs.
33408 2017-04-12 Richard Biener <rguenther@suse.de>
33410 PR tree-optimization/79390
33411 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33412 threading case even more.
33414 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33417 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33418 for quad_address_p for TImode, instead of just not indexed_address.
33420 2017-04-12 Richard Biener <rguenther@suse.de>
33421 Bernd Edlinger <bernd.edlinger@hotmail.de>
33423 PR middle-end/79671
33424 * alias.c (component_uses_parent_alias_set_from): Handle
33425 TYPE_TYPELESS_STORAGE.
33426 (get_alias_set): Likewise.
33427 * tree-core.h (tree_type_common): Add typeless_storage flag.
33428 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33429 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33430 for types containing members with TYPE_TYPELESS_STORAGE.
33431 (place_field): Likewise.
33432 (layout_type): Likewise for ARRAY_TYPE.
33433 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33434 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33435 TYPE_TYPELESS_STORAGE.
33436 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33438 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33441 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33442 first argument to type.
33444 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33448 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33449 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33450 (rs6000_expand_binop_builtin): Likewise.
33451 (rs6000_expand_ternop_builtin): Likewise; also add missing
33452 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33453 vshasigma built-ins.
33454 * doc/extend.texi: Document that vec_xxpermdi's third argument
33455 must be a constant.
33457 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33459 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33460 Use shift_const cost parameter when calculating gain of STV shifts.
33462 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33464 PR rtl-optimization/70478
33465 * lra-constraints.c (process_alt_operands): Check memory for
33466 disfavoring memory insn operand.
33468 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33470 PR middle-end/80100
33471 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33472 left shift in unsigned HOST_WIDE_INT type.
33474 PR rtl-optimization/80385
33475 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33476 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33479 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33480 if they have any depend clauses.
33482 2017-04-11 Martin Liska <mliska@suse.cz>
33485 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33486 * ipa-split.c (split_function): Create a local comdat symbol
33487 if caller is in a comdat group.
33489 2017-04-11 Martin Liska <mliska@suse.cz>
33492 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33495 2017-04-11 Martin Sebor <msebor@redhat.com>
33497 PR middle-end/80364
33498 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33499 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33501 (directive::set_width, directive::set_precision, format_character):
33503 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33506 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33509 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33510 conflict, set target->arch_name instead of target->cpu_name.
33512 2017-04-11 Richard Biener <rguenther@suse.de>
33514 PR tree-optimization/80374
33515 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33516 build_zero_cst, remove fold_convertible_p check again.
33518 2017-04-11 Martin Liska <mliska@suse.cz>
33521 * ubsan.c (instrument_object_size): Do not instrument register
33524 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33527 * config/i386/i386-builtin-types.def
33528 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33529 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33530 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33531 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33532 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33533 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33534 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33535 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33536 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33537 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33538 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33539 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33540 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33541 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33542 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33543 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33544 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33545 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33546 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33547 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33548 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33549 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33550 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33551 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33552 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33553 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33554 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33555 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33556 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33557 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33558 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33559 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33560 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33561 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33562 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33563 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33564 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33565 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33566 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33567 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33568 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33569 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33570 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33571 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33572 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33573 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33575 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33576 flag to second_arg_count, handle 4 argument function type _COUNT
33577 aliases, handle second_arg_count on second argument rather than last.
33579 2017-04-10 Jeff Law <law@redhat.com>
33581 PR tree-optimization/80374
33582 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33583 record anything if we can not convert integer_zero_node to the
33586 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33589 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33590 Enhance special handling given to the TARGET_P9_MINMAX option in
33591 relation to certain other options.
33593 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33595 PR tree-optimization/80153
33596 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33597 remove POINTER_PLUS_EXPR's base part directly, rather than through
33600 2017-04-10 Richard Biener <rguenther@suse.de>
33601 Bin Cheng <bin.cheng@arm.com>
33603 PR tree-optimization/80153
33604 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33605 the first element of pointer type aff_tree. Build result expr in
33607 (add_elt_to_tree): Convert to type unconditionally. Remove other
33608 fold_convert calls.
33609 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33610 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33612 2017-04-10 Richard Biener <rguenther@suse.de>
33614 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33617 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33619 PR rtl-optimization/70478
33620 * lra-constraints.c (curr_small_class_check): New.
33621 (update_and_check_small_class_inputs): New.
33622 (process_alt_operands): Update curr_small_class_check. Disfavor
33623 alternative insn memory operands. Check available regs for small
33626 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33629 * config/mips/mips.opt (-mvirt): Update description.
33630 * doc/invoke.texi (-mvirt): Likewise.
33632 2017-04-10 Richard Biener <rguenther@suse.de>
33634 PR middle-end/80362
33635 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33636 looking for NEGATE_EXPR in -A / -B to A / B folding.
33638 2017-04-10 Martin Liska <mliska@suse.cz>
33640 PR gcov-profile/80224
33641 * gcov.c (print_usage): Fix usage string.
33642 (get_gcov_intermediate_filename): Remove.
33643 (output_gcov_file): Use both for normal and intermediate format.
33644 (generate_results): Do not initialize special file for
33645 intermediate format.
33647 2017-04-10 Richard Biener <rguenther@suse.de>
33649 PR tree-optimization/80304
33650 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33653 2017-04-10 Nathan Sidwell <nathan@acm.org>
33656 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33657 (rs6000_init_builtins): Use it.
33659 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33661 * config/arm/arm.md (<mrc>): Add mode to SET source.
33662 (<mrrc>): Likewise.
33664 2017-04-10 Richard Biener <rguenther@suse.de>
33666 PR middle-end/80344
33667 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33669 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33672 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33673 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33674 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33675 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33676 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33677 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33678 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33679 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33680 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33681 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33682 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33683 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33684 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33685 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33686 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33687 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33688 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33689 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33690 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33691 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33692 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33693 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33694 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33696 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33698 PR rtl-optimization/70478
33699 * lra-constraints.c: Reverse the last patch.
33701 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33703 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33704 Add comment for WCHAR_T.
33706 2017-04-08 Martin Liska <mliska@suse.cz>
33709 2017-04-07 Martin Liska <mliska@suse.cz>
33712 * ipa-split.c (split_function): Add function part to a same comdat
33715 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33718 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33720 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33722 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33723 (vec_pairing): ...to this.
33724 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33725 (rs6000_sched_init): Adjust for name change.
33726 (struct rs6000_sched_context): Likewise.
33727 (rs6000_init_sched_context): Likewise.
33728 (rs6000_set_sched_context): Likewise.
33730 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33736 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33738 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33739 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33740 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33742 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33744 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33746 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33748 PR rtl-optimization/70703
33749 * ira-color.c (update_conflict_hard_regno_costs): Use
33750 int64_t instead of HOST_WIDE_INT.
33752 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33754 PR rtl-optimization/70478
33755 * lra-constraints.c (process_alt_operands): Disfavor alternative
33756 insn memory operands.
33758 2017-04-07 Jeff Law <law@redhat.com>
33760 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33761 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33763 2017-04-07 Martin Liska <mliska@suse.cz>
33766 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33767 Show error message instead of an ICE.
33769 2017-04-07 Martin Liska <mliska@suse.cz>
33772 * ipa-split.c (split_function): Add function part to a same comdat
33775 2017-04-07 Richard Biener <rguenther@suse.de>
33777 PR middle-end/80341
33778 * tree.c (get_unwidened): Also handle ! for_type case for
33780 * convert.c (do_narrow): Split out from ...
33781 (convert_to_integer_1): ... here. Do not pass final truncation
33782 type to get_unwidened for TRUNC_DIV_EXPR.
33784 2017-04-07 Richard Biener <rguenther@suse.de>
33786 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33788 (aff_combination_const): Adjust.
33789 (aff_combination_scale): Likewise.
33790 (aff_combination_add_elt): Likewise.
33791 (aff_combination_add_cst): Likewise.
33792 (aff_combination_convert): Likewise.
33793 (add_elt_to_tree): Likewise. Remove unused argument.
33794 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33796 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33798 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33800 * config/arm/arm.c (arm_default_short_enums): Use
33801 ARM_DEFAULT_SHORT_ENUMS.
33802 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33804 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33807 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33808 members with redundant out-of-class redeclaration.
33810 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33813 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33814 * config/i386/i386.md (*zero_extendsidi2):
33815 Add (?*x,*x) and (?*v,*v) alternatives.
33817 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33820 * config/i386/i386.c (ix86_expand_builtin)
33821 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33822 mode from insn data. Convert operands to insn operand mode.
33823 Copy operands that don't satisfy insn predicate to a register.
33825 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33827 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33830 2017-04-06 Richard Biener <rguenther@suse.de>
33832 PR tree-optimization/80334
33833 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33834 preserve alignment of accesses.
33836 2017-04-06 Richard Biener <rguenther@suse.de>
33838 PR tree-optimization/80262
33839 * tree-sra.c (build_ref_for_offset): Preserve address-space
33841 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33842 Drop useless address-space information on MEM_REF offsets.
33844 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33846 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33848 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33850 PR rtl-optimization/70703
33851 * ira-color.c (update_conflict_hard_regno_costs): Use
33852 HOST_WIDE_INT instead of long.
33854 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33857 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33858 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33860 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33861 for x86_64 target. Handle -m3dnowa option.
33863 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33865 PR rtl-optimization/70703
33866 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33867 (update_conflict_hard_regno_costs): Use long instead of unsigned
33868 arithmetic for cost calculation.
33870 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33871 Bernd Edlinger <bernd.edlinger@hotmail.de>
33874 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33877 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33880 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33881 ptr_mode with Pmode throughout.
33882 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33883 into probe_stack_range and use DImode.
33885 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33888 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33889 call_eh_return is true.
33891 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33893 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33894 Initialize last_match_fntype_index.
33896 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33899 * tree-nvr.c: Include internal-fn.h.
33900 (pass_return_slot::execute): Ignore internal calls without
33903 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33904 Richard Biener <rguenther@suse.de>
33907 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33908 captures used multiple times, except for the last use.
33909 * generic-match-head.c: Include gimplify.h.
33911 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33913 PR tree-optimization/79390
33914 * target.h (struct noce_if_info): Declare.
33915 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33916 * target.def (noce_conversion_profitable_p): New target hook.
33917 * ifcvt.h (struct noce_if_info): New type, moved from ...
33918 * ifcvt.c (struct noce_if_info): ... here.
33919 (noce_conversion_profitable_p): Renamed to ...
33920 (default_noce_conversion_profitable_p): ... this. No longer
33922 (noce_try_store_flag_constants, noce_try_addcc,
33923 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33924 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33925 instead of noce_conversion_profitable_p.
33926 * config/i386/i386.c: Include ifcvt.h.
33927 (ix86_option_override_internal): Don't override
33928 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33929 (ix86_noce_conversion_profitable_p): New function.
33930 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33931 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33932 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33933 * doc/tm.texi: Regenerated.
33935 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33937 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33940 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33943 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33944 instructions for small multiply cores.
33946 2017-04-04 Jeff Law <law@redhat.com>
33948 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33950 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33951 that are not set by the loop over the elements.
33953 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33956 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33957 int mode, convert_modes it to mode as unsigned, otherwise use
33958 lowpart_subreg to mode rather than SImode.
33959 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33960 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33961 Use DImode instead of SImode for the shift count operand.
33962 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33965 2017-04-04 Richard Biener <rguenther@suse.de>
33967 PR middle-end/80281
33968 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33969 arithmetic done for the negate or the plus. Simplify.
33970 (A - (-B) -> A + B): Likewise.
33971 * fold-const.c (split_tree): Make sure to not negate pointers.
33973 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33975 PR rtl-optimization/60818
33976 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33977 a compare of comparisons with the thing compared if this results
33978 in a different machine mode.
33980 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33982 * alias.c (base_alias_check): Fix typo in comment.
33983 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33984 * cgraphunit.c (symbol_table::compile): Likewise.
33985 * collect2.c (maybe_run_lto_and_relink): Likewise.
33986 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33987 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33988 * config/avr/avr.c (avr_map_op_t): Likewise.
33989 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33990 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33991 * config/epiphany/epiphany.md (movcc): Likewise.
33992 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33993 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33995 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33996 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33997 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33998 * config/sh/sh.c (sh_rtx_costs): Likewise.
33999 * fold-const.c (fold_truth_andor): Likewise.
34000 * genautomata.c (collapse_flag): Likewise.
34001 * gengtype.h (struct type::u::s): Likewise.
34002 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34003 * input.c (FORMAT_AMOUNT): Likewise.
34004 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34005 (known_aggs_to_agg_replacement_list): Likewise.
34006 * ipa-inline-analysis.c: Likewise.
34007 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34008 * ipa-polymorphic-call.c
34009 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34010 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34011 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34013 * modulo-sched.c (apply_reg_moves): Likewise.
34014 * omp-expand.c (build_omp_regions_1): Likewise.
34015 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34016 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34017 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34018 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34019 * value-prof.c: Likewise.
34020 * var-tracking.c (val_reset): Likewise.
34022 2017-04-03 Richard Biener <rguenther@suse.de>
34024 PR tree-optimization/80275
34025 * fold-const.c (split_address_to_core_and_offset): Handle
34028 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34030 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34031 descriptors is at least equal to that of functions.
34033 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34035 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34037 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34040 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34041 (mov<IMOD4:mode>): New expander.
34042 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34044 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34046 PR rtl-optimization/79405
34047 * fwprop.c (propagations_left): New variable.
34048 (forward_propagate_into): Decrement it.
34049 (fwprop_init): Initialize it.
34050 (fw_prop): If the variable has reached zero, stop propagating.
34051 (fwprop_addr): Ditto.
34053 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34056 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34057 a FUNCTION_DECL, pass it as decl instead of origin to
34060 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34062 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34065 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34068 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34069 TARGET_VSX_SMALL_INTEGER.
34071 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34073 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34074 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34076 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34078 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34079 extraction from odd-numbered MSA register.
34081 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34083 PR middle-end/80173
34084 * expmed.c (store_bit_field_1): Don't attempt to create
34085 a word subreg out of hard registers wider than word if they
34086 have HARD_REGNO_NREGS of 1 for their mode.
34088 PR middle-end/80163
34089 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34090 conversions to integer types wider than word and pointer.
34093 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34094 (rtx_equal_for_cselib_p): Pass 0 to it.
34095 * cselib.c (cselib_hasher::equal): Likewise.
34096 (rtx_equal_for_cselib_1): Add depth argument. If depth
34097 is 128, don't look up VALUE locs and punt. Increment
34098 depth in recursive calls when walking VALUE locs.
34100 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34102 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34103 (make_gcov_file_name): Use the canonical path name for generating
34105 (read_line): Fix handling of files with ascii null bytes.
34107 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34109 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34110 to initialise a vector register instead
34111 of using a const_int.
34113 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34115 PR translation/80189
34116 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34117 diagnostic messages.
34119 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34122 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34123 (dfp_diex_<mode>): Update mode of operand 1.
34124 * doc/extend.texi (dxex, dxexq): Document change to return type.
34125 (diex, diexq): Document change to argument type.
34127 2017-03-30 Martin Jambor <mjambor@suse.cz>
34130 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34131 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34132 it reflects the signature changes performed at the callee side.
34133 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34134 to cgraph_build_function_type_skip_args.
34135 (build_function_decl_skip_args): Adjust call to the above function.
34137 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34140 * config/i386/sse.md
34141 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34142 register as dest whenever it is a MEM not rtx_equal_p to the
34143 corresponding dup operand, and when forcing into reg move the
34144 reg into the memory afterwards.
34145 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34146 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34147 for the force_reg mode.
34148 (avx512vl_vextractf128<mode>): Use register as dest either
34149 always when a MEM, or when it is a MEM not rtx_equal_p to the
34150 corresponding dup operand, or even not when it is a CONST_VECTOR
34151 depending on the mode and lo vs. hi.
34152 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34154 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34155 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34156 Likewise. Require that operands[2] is even.
34157 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34158 Remove extraneous parens. Require that operands[2] is a multiple
34160 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34161 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34162 disallow memory then.
34164 2017-03-30 Richard Biener <rguenther@suse.de>
34166 PR tree-optimization/77498
34167 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34168 to non-constants over backedges.
34170 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34172 PR rtl-optimization/80233
34173 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34174 as last_combined_insn. Do not test for BARRIER_P separately.
34176 2017-03-29 Andreas Schwab <schwab@suse.de>
34179 * calls.c (prepare_call_address): Convert funexp to Pmode before
34180 copying to temp reg.
34182 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34184 PR tree-optimization/80158
34185 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34186 Handle possible future case of more than one alternate
34188 (replace_rhs_if_not_dup): Likewise.
34189 (replace_one_candidate): Likewise.
34191 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34193 PR rtl-optimization/80193
34194 * ira.c (ira): Do not check allocation for LRA.
34196 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34198 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34199 (nvptx_output_simt_exit): Declare.
34200 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34201 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34202 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34203 (nvptx_declare_function_name): ...here. Emit declaration of local
34204 memory space buffer for omp_simt_enter insn.
34205 (nvptx_output_unisimt_switch): New.
34206 (nvptx_output_softstack_switch): New.
34207 (nvptx_output_simt_enter): New.
34208 (nvptx_output_simt_exit): New.
34209 * config/nvptx/nvptx.h (struct machine_function): New fields
34210 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34211 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34212 (UNSPECV_SIMT_EXIT): Ditto.
34213 (omp_simt_enter_insn): New insn.
34214 (omp_simt_enter): New expansion.
34215 (omp_simt_exit): New insn.
34216 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34218 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34219 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34220 (expand_GOMP_SIMT_EXIT): New.
34221 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34222 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34223 (GOMP_SIMT_EXIT): Ditto.
34224 * target-insns.def (omp_simt_enter): New insn.
34225 (omp_simt_exit): Ditto.
34226 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34228 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34229 (lower_rec_input_clauses): Likewise.
34230 (lower_lastprivate_clauses): Handle SIMT privatization.
34232 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34233 (ompdevlow_adjust_simt_enter): New.
34234 (find_simtpriv_var_op): New.
34235 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34236 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34238 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34239 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34240 (copy_decl_for_dup_finish): Ditto.
34242 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34244 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34247 * config/i386/i386.c (ix86_option_override_internal): Always
34248 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34250 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34252 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34254 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34256 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34257 mark new edge's irreducible flag accordign to it.
34258 (vect_do_peeling): Check loop preheader edge's irreducible flag
34259 and pass it to function slpeel_add_loop_guard.
34261 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34263 PR tree-optimization/80218
34264 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34265 Update block frequencies and counts.
34267 2017-03-28 Richard Biener <rguenther@suse.de>
34269 PR tree-optimization/78644
34270 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34271 of a simplification result we may not use it at all.
34273 2017-03-28 Richard Biener <rguenther@suse.de>
34276 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34277 without arguments, generate default definition of a SSA name.
34279 2017-03-28 Richard Biener <rguenther@suse.de>
34281 PR middle-end/80222
34282 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34283 TYPE_REF_CAN_ALIAS_ALL references.
34284 * fold-const.c (fold_indirect_ref_1): Likewise.
34286 2017-03-28 Martin Liska <mliska@suse.cz>
34289 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34290 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34292 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34293 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34295 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34296 (EXTRA_SPECS): Define.
34297 (SUBTARGET_EXTRA_SPECS): Likewise.
34298 (SUBTARGET_CPP_SPEC): Likewise.
34299 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34300 SUBTARGET_EXTRA_SPECS.
34301 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34303 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34305 * config/arc/simdext.md (vst64_insn): Update pattern.
34306 (vld32wh_insn): Likewise.
34307 (vld32wl_insn): Likewise.
34308 (vld64_insn): Likewise.
34309 (vld32_insn): Likewise.
34311 2017-03-28 Marek Polacek <polacek@redhat.com>
34314 * fold-const.c (fold_comparison): Use protected_set_expr_location
34315 instead of SET_EXPR_LOCATION.
34317 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34319 * tree.c (add_expr): Avoid name lookup warning.
34321 2017-03-27 Jeff Law <law@redhat.com>
34323 PR tree-optimization/80216
34324 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34325 function name. Limit recursion depth.
34326 (record_temporary_equivalences): Corresponding changes.
34328 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34330 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34333 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34336 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34338 * cfgcleanup.c (reg_note_cfa_p): New array.
34339 (insns_have_identical_cfa_notes): New function.
34340 (old_insns_match_p): Don't cross-jump in between /f
34341 and non-/f instructions. If both i1 and i2 are frame related,
34342 verify all CFA notes, their order and content.
34344 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34347 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34348 HImode and SImode with zero extend to DImode to one insn.
34349 (bswap<mode>2_extenddi): Likewise.
34350 (bswapsi2_extenddi): Likewise.
34351 (bswaphi2_extendsi): Likewise.
34352 (bswaphi2): Combine bswap HImode and SImode into one insn.
34353 Separate memory insns from swapping register.
34354 (bswapsi2): Likewise.
34355 (bswap<mode>2): Likewise.
34356 (bswaphi2_internal): Delete, no longer used.
34357 (bswapsi2_internal): Likewise.
34358 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34359 store, and gpr<-gpr swap insns.
34360 (bswap<mode>2_store): Likewise.
34361 (bswaphi2_reg): Register only splitter, combine with the splitter.
34362 (bswaphi2 splitter): Likewise.
34363 (bswapsi2_reg): Likewise.
34364 (bswapsi2 splitter): Likewise.
34365 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34366 the insns into load, store, and register/register insns.
34367 (bswapdi2_ldbrx): Likewise.
34368 (bswapdi2_load): Likewise.
34369 (bswapdi2_store): Likewise.
34370 (bswapdi2_reg): Likewise.
34372 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34374 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34375 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34377 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34380 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34382 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34383 special handling for target option conflicts between dform
34384 options (-mpower9-dform, -mpower9-dform-vector,
34385 -mpower9-dform-scalar) and -mno-direct-move.
34387 2017-03-27 Richard Biener <rguenther@suse.de>
34389 PR tree-optimization/80181
34390 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34392 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34394 * config/arc/predicates.md (move_double_src_operand): Replace the
34395 call to move_double_src_operand with a call to address_operand.
34397 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34399 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34400 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34401 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34403 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34405 * config/arc/predicates.md (long_immediate_loadstore_operand):
34406 Consider scaled addresses cases.
34408 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34410 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34411 restored when in interrupt.
34412 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34413 doesn't have delay slot.
34415 2017-03-27 Richard Biener <rguenther@suse.de>
34418 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34419 inlined thunk clones.
34421 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34424 * asan.c (instrument_derefs): Copy over last operand from
34425 original COMPONENT_REF to the new COMPONENT_REF with
34426 DECL_BIT_FIELD_REPRESENTATIVE.
34427 * ubsan.c (instrument_object_size): Likewise.
34429 2017-03-27 Richard Biener <rguenther@suse.de>
34431 PR tree-optimization/80170
34432 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34433 sure DR/SCEV didnt fold in constants we do not see when looking
34434 at the reference base alignment.
34436 2017-03-27 Richard Biener <rguenther@suse.de>
34438 PR middle-end/80171
34439 * gimple-fold.c (fold_ctor_reference): Properly guard against
34440 NULL return value from canonicalize_constructor_val.
34442 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34445 * config/i386/i386.c (ix86_expand_builtin)
34446 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34447 flags reg setting and flags reg using instructions.
34448 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34449 clobbering instructions to zero extend op2.
34451 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34453 * doc/install.texi (Configuration) <--with-aix-soname>:
34454 Update link to AIX ld.
34456 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34458 PR rtl-optimization/80160
34459 PR rtl-optimization/80159
34460 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34461 reg_alternate_class into account.
34463 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34466 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34467 to consider in curr_insn_transform.
34469 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34471 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34472 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34473 and emit_mode_inner.
34475 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34477 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34478 argument to the overloaded builtin variants. Use the new flag to
34479 deprecate certain builtin variants.
34480 * config/s390/s390-builtin-types.def: Add new builtin types.
34481 * config/s390/s390-builtins.h: Support new flags field for
34482 overloaded builtins.
34483 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34484 (s390_macro_to_expand): Enable vector float data type.
34485 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34486 builtins by incrementing the __VEC__ version number.
34487 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34489 (s390_resolve_overloaded_builtin): Emit error messages depending
34490 on the builtin flags.
34491 * config/s390/s390.c (s390_expand_builtin): Support additional
34492 flags argument. Change error message to match the messages
34493 emitted in s390-c.c.
34494 * config/s390/s390.md: New UNSPEC_* constants.
34495 (op_type): Add new instruction types.
34496 * config/s390/vecintrin.h: Add new builtins and test data class
34498 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34499 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34500 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34501 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34502 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34503 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34505 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34506 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34507 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34508 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34510 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34511 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34512 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34513 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34515 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34516 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34517 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34518 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34519 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34520 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34521 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34523 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34524 ("vec_scatter_element<V_HW_4:mode>_DI")
34525 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34526 ("vec_fpint<mode>", "vflls")
34527 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34528 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34529 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34530 ("*vec_cmphe<mode>_cc"): ... these.
34532 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34533 mode constant instead of magic value.
34535 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34537 * config/s390/s390.c (s390_expand_vec_compare): Support other
34538 vector floating point modes than just V2DF.
34539 (s390_expand_vcond): Likewise.
34540 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34541 (s390_cannot_change_mode_class): Prevent mode changes between TF
34542 and V1TF in vector registers.
34543 * config/s390/s390.md (DF, SF): New mode attributes.
34544 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34545 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34546 SFmode support for VRs.
34547 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34549 (VFT, VF_HW): New mode iterators.
34550 (vw, sdx): New mode attributes.
34551 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34552 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34553 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34554 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34555 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34556 also the new vector floating point modes. Renaming to ...
34558 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34559 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34560 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34561 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34562 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34563 ("vec_unordered<mode>"): ... these.
34565 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34566 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34567 ("*vec_extendv2df"): New insn definitions.
34569 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34571 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34572 ("mulditi3_2", "*muldi3_sign"): New patterns.
34573 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34574 rename the pattern definition.
34576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34578 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34580 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34582 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34584 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34585 instruction if possible.
34586 * config/s390/vector.md (vec_halfnumelts): New mode
34588 ("*vec_vllezlf<mode>"): New pattern.
34590 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34592 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34593 ("popcountv4si2", "popcountv2di2"): Rename to ...
34594 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34595 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34597 ("popcount<mode>2_vxe"): New pattern.
34599 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34601 * common/config/s390/s390-common.c (processor_flags_table): Add
34603 * config.gcc: Add arch12.
34604 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34605 Default to arch12 for unknown CPU model numbers.
34606 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34607 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34608 PROCESSOR_max sanity check.
34609 * config/s390/s390-opts.h (enum processor_type): Add
34611 * config/s390/s390.c (processor_table): Add arch12.
34612 (s390_expand_builtin): Add check for B_VXE flag.
34613 (s390_issue_rate): Add PROCESSOR_ARCH12.
34614 (s390_get_sched_attrmask): Likewise.
34615 (s390_get_unit_mask): Likewise.
34616 (s390_sched_score): Enable z13 scheduling for arch12.
34617 (s390_sched_reorder): Likewise.
34618 (s390_sched_variable_issue): Likewise.
34619 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34621 (s390_tune_attr): Use z13 scheduling also for arch12.
34622 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34623 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34624 (TARGET_VXE_P): New macros.
34625 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34626 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34627 * config/s390/s390.opt: Add arch12 as processor_type.
34629 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34631 * config/s390/s390.md
34632 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34633 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34634 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34636 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34637 Rename expanders to ...
34639 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34640 ("fixuns_truncdddi2_emu"): ... these.
34642 ("fixuns_trunc<mode>si2_emu"): New expander.
34644 ("*fixuns_truncdfdi2_z13"): Rename to ...
34645 ("*fixuns_truncdfdi2_vx"): ... this.
34647 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34649 * config/s390/2964.md: Remove the single element vector compare
34650 instructions which are no longer used.
34651 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34653 (s390_canonicalize_comparison): Remove handling of DFmode
34655 (s390_expand_vec_compare_scalar): Remove function.
34656 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34657 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34659 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34661 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34663 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34665 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34666 will anyway by matched by mov<mode>_64dfp.
34668 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34670 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34671 vlef/vstef. Add missing operand to vleif.
34673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34675 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34676 pair for all vector types with 64 bit elements.
34677 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34678 * config/s390/vector.md (V_HW_64): ... here.
34679 (V_128_NOSINGLE): New mode iterator.
34680 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34681 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34682 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34683 ("*vec_load_pairv2di"): Change to ...
34684 ("*vec_load_pair<mode>"): ... this one.
34686 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34688 * config/s390/constraints.md: Add comments.
34689 (jKK): Reject element sizes > 8 bytes.
34690 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34692 * config/s390/s390.md: Add the s_operand checks formerly in
34693 s390_split_ok_p to various splitters where they are still
34695 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34696 for 128 bit vectors. Plus two splitters.
34698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34700 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34703 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34706 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34707 error if the boundary argument is not constant.
34709 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34711 PR rtl-optimization/80112
34712 * loop-doloop.c (doloop_condition_get): Don't check condition
34713 if cmp isn't SET with IF_THEN_ELSE src.
34715 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34717 PR tree-optimization/80158
34718 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34719 replacing a candidate statement, also replace it for the
34720 candidate's alternate interpretation.
34721 (replace_rhs_if_not_dup): Likewise.
34722 (replace_one_candidate): Likewise.
34724 2017-03-24 Richard Biener <rguenther@suse.de>
34726 PR tree-optimization/80167
34727 * graphite-isl-ast-to-gimple.c
34728 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34730 (translate_isl_ast_to_gimple::get_rename): Likewise.
34732 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34735 handling of certain combinations of target options, including the
34736 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34737 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34739 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34742 * config/arm/arm.md (*load_multiple): Add reload_completed to
34743 matching condition.
34745 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34746 Richard Biener <rguenth@suse.de>
34748 PR tree-optimization/79908
34749 PR tree-optimization/80136
34750 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34751 been cast away, gimplify_and_add suffices.
34753 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34755 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34757 2017-03-23 Richard Biener <rguenther@suse.de>
34759 PR tree-optimization/80032
34760 * gimplify.c (gimple_push_cleanup): Forced unconditional
34761 cleanups still have to go to the conditional_cleanups
34764 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34766 PR tree-optimization/80072
34767 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34769 (next_operand_entry_id): Change type to unsigned int.
34770 (sort_by_operand_rank): Make sure to return the right return value
34771 even if unsigned fields are bigger than INT_MAX.
34772 (struct oecount): Change cnt and id type to unsigned int.
34773 (oecount_hasher::equal): Formatting fix.
34774 (oecount_cmp): Make sure to return the right return value
34775 even if unsigned fields are bigger than INT_MAX.
34776 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34779 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34780 TREE_READONLY on result if writing it more than once.
34783 * doc/invoke.texi (-fsanitize=thread): Document that with
34784 -fnon-call-exceptions atomics are not able to throw
34788 * tsan.c: Include tree-eh.h.
34789 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34790 maybe_clean_or_replace_eh_stmt where needed.
34791 (instrument_memory_accesses): Add cfg_changed argument.
34792 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34793 if it returned true.
34794 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34796 PR rtl-optimization/63191
34797 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34798 wrapper function, moved the whole old content into ...
34799 (ix86_delegitimize_address_1): ... this. New inline function.
34800 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34801 true as last argument instead of ix86_delegitimize_address.
34803 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34805 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34806 cortexa57_branch_cost.
34808 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34810 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34812 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34815 * doc/md.texi (Constraints): Document wA constraint.
34816 * config/rs6000/constraints.md (wA): New.
34817 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34818 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34819 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34820 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34822 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34825 * gimplify.c (is_oacc_declared): New function.
34826 (oacc_default_clause): Use it to set default flags for acc declared
34827 variables inside parallel regions.
34828 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34829 declared variables.
34830 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34831 declare attribute to any decl as necessary.
34833 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34836 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34837 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34838 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34839 (arm_arch_lpae): This.
34840 * config/arm/arm.c (arm_arch7ve): Rename into ...
34841 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34842 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34845 2017-03-22 Martin Liska <mliska@suse.cz>
34848 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34849 error message instead of an ICE.
34851 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34853 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34855 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34857 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34859 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34862 2017-03-21 Martin Sebor <msebor@redhat.com>
34864 * doc/extend.texi: Use "cannot" instead of "can't."
34865 * doc/hostconfig.texi: Same.
34866 * doc/install.texi: Same.
34867 * doc/invoke.texi: Same.
34868 * doc/loop.texi: Same.
34869 * doc/md.texi: Same.
34870 * doc/objc.texi: Same.
34871 * doc/rtl.texi: Same.
34872 * doc/tm.texi: Same.
34873 * doc/tm.texi.in: Same.
34874 * doc/trouble.texi: Same.
34876 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34879 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34880 (collect_checksum_attributes): Set it.
34881 (die_checksum_ordered): Use it.
34883 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34885 PR tree-optimization/79908
34886 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34887 change: For a VA_ARG whose LHS has been cast away, use
34888 force_gimple_operand to construct the side effects.
34890 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34892 PR translation/80001
34893 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34894 more amenable to translation.
34895 (oacc_loop_auto_partitions): Likewise.
34897 2017-03-21 Marek Polacek <polacek@redhat.com>
34898 Martin Sebor <msebor@redhat.com>
34900 PR tree-optimization/80109
34901 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34902 on INTEGRAL_TYPE_P.
34904 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34905 Segher Boessenkool <segher@kernel.crashing.org>
34908 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34909 check reg_used_between_p between insn and one of succ or succ2
34910 depending on if succ is artificial insn not inserted into insn
34913 2017-03-21 Martin Liska <mliska@suse.cz>
34915 PR gcov-profile/80081
34916 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34917 * doc/gcc.texi: Include gcov-dump stuff.
34918 * doc/gcov-dump.texi: New file.
34920 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34922 PR rtl-optimization/79150
34923 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34924 conditional jump, if the jump is the last insn of the loop.
34926 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34927 Richard Biener <rguenth@suse.de>
34929 PR tree-optimization/79908
34930 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34931 been cast away, use force_gimple_operand to construct the side
34934 2017-03-21 Martin Liska <mliska@suse.cz>
34936 PR libfortran/79956
34937 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34940 2017-03-21 Brad Spengler <spender@grsecurity.net>
34943 * plugin.c (htab_hash_plugin): New function.
34944 (add_new_plugin): Use it and adjust.
34945 (parse_plugin_arg_opt): Adjust.
34946 (init_one_plugin): Likewise.
34948 2017-03-21 Richard Biener <rguenther@suse.de>
34950 PR tree-optimization/80032
34951 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34952 if set force the cleanup to happen unconditionally.
34953 (gimplify_target_expr): Push inserted clobbers with force_uncond
34954 to avoid them being removed by control-dependent DCE.
34956 2017-03-21 Richard Biener <rguenther@suse.de>
34958 PR tree-optimization/80122
34959 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34960 va_arg_pack_len when the inlined call stmt requires pack
34962 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34964 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34967 * tsan.c (instrument_builtin_call): If the memory model argument
34968 is not a constant, assume it is valid.
34971 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34974 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34976 PR rtl-optimization/79910
34977 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34978 if its dest is used by an insn before I2 (other than the combined
34979 insns themselves, which are properly handled already).
34981 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34984 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34986 * combine.c (record_used_regs): New static function.
34987 (try_combine): Handle situations where there is an additional
34988 instruction between I2 and I3 which needs to have a LOG_LINK
34992 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34994 * combine.c (try_combine): Delete redundant i1 test. Call
34995 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34997 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35000 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35001 alternatives 13/14.
35003 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35005 PR tree-optimization/80054
35006 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35007 the optimization if a PHI or any of its arguments is not dominated
35008 by the candidate's basis. Use gphi* rather than gimple* as
35010 (replace_profitable_candidates): Clean up a gimple* variable that
35011 should be a gphi* variable.
35013 2017-03-20 Martin Sebor <msebor@redhat.com>
35016 * doc/extend.texi (attribute constructor): Document present limitation.
35018 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35021 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35022 __POWER9_VECTOR__ #ifdef control, change template definition to
35023 use Power9-specific built-in function.
35024 (vec_any_eq): Likewise.
35025 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35026 to control outcomes from this test.
35027 (vector_ae_<mode>p): For VEC_F modes, likewise.
35029 2017-03-20 Ian Lance Taylor <iant@google.com>
35031 * config/i386/i386.c (ix86_function_regparm): Save an extra
35032 register for -fsplit-stack with DECL_STATIC_CHAIN.
35034 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35037 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35038 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35040 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35042 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35044 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35047 2017-03-20 Marek Polacek <polacek@redhat.com>
35050 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35052 2017-03-20 Richard Biener <rguenther@suse.de>
35054 PR tree-optimization/80113
35055 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35056 allocate extra SSA name for PHI def.
35057 (add_close_phis_to_outer_loops): Likewise.
35058 (add_close_phis_to_merge_points): Likewise.
35059 (copy_loop_close_phi_args): Likewise.
35060 (copy_cond_phi_nodes): Likewise.
35062 2017-03-20 Martin Liska <mliska@suse.cz>
35064 PR middle-end/79753
35065 * tree-chkp.c (chkp_build_returned_bound): Do not build
35066 returned bounds for a LHS that's not a BOUNDED_P type.
35068 2017-03-20 Martin Liska <mliska@suse.cz>
35072 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35073 COMPLEX_CST and VECTOR_CST.
35075 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35078 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35079 target operand. A new splitter adds the clobber statement in case
35080 the target operand is dead anyway.
35082 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35084 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35085 to age-old versions of binutils and glibc.
35087 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35089 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35091 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35093 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35095 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35097 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35098 requirement for binutils 2.13.
35100 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35102 * combine.c (try_combine): Delete redundant i1 test. Call
35103 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35105 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35107 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35108 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35110 <riscv64-*-elf>: Re-arrange section
35111 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35112 <riscv32-*-linux>: Likewise.
35113 <riscv64-*-elf>: Likewise
35114 <riscv64-*-linux>: Likewise.
35116 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35119 * aarch64.opt(verbose-cost-dump): Fix typo.
35121 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35124 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35125 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35127 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35129 * reload.c (find_reloads): When reloading a nonoffsettable address,
35130 use RELOAD_OTHER for it and its address reloads.
35132 PR rtl-optimization/79910
35133 * combine.c (record_used_regs): New static function.
35134 (try_combine): Handle situations where there is an additional
35135 instruction between I2 and I3 which needs to have a LOG_LINK
35138 2017-03-17 Jeff Law <law@redhat.com>
35140 PR tree-optimization/71437
35141 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35142 conditional in the hash table first.
35143 (vrp_dom_walker::before_dom_children): Extract condition from
35144 ASSERT_EXPR. Record condition, its inverion and any implied
35145 conditions as well.
35147 2017-03-17 Marek Polacek <polacek@redhat.com>
35148 Markus Trippelsdorf <markus@trippelsdorf.de>
35150 PR tree-optimization/80079
35151 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35154 2017-03-17 Richard Biener <rguenther@suse.de>
35156 PR middle-end/80075
35157 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35158 Properly verify the LHS before the RHS possibly claims to be
35160 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35163 2017-03-17 Martin Jambor <mjambor@suse.cz>
35165 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35166 (List of -O2 options): Likewise.
35167 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35170 2017-03-17 Tom de Vries <tom@codesourcery.com>
35172 * gcov-dump.c (print_usage): Print bug_report_url.
35174 2017-03-17 Richard Biener <rguenther@suse.de>
35176 PR middle-end/80050
35177 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35178 (parser::peek): Likewise.
35180 2017-03-17 Richard Biener <rguenther@suse.de>
35182 PR tree-optimization/80048
35183 * sese.c (free_sese_info): Properly release rename_map and
35184 copied_bb_map elements.
35186 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35188 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35189 Add linked-list forward and backlinks. Insert on
35190 construction, remove on destruction.
35191 (class pass_store_merging): Add m_stores_head field.
35192 (pass_store_merging::terminate_and_process_all_chains):
35193 Iterate over m_stores_head list.
35194 (pass_store_merging::terminate_all_aliasing_chains):
35196 (pass_store_merging::execute): Check for debug stmts first.
35197 Push new chains onto the m_stores_head stack.
35199 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35202 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35203 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35204 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35206 2017-03-16 Jeff Law <law@redhat.com>
35208 PR tree-optimization/71437
35209 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35210 member function. Implementation moved into after_dom_children
35211 member function and into the threader's thread_outgoing_edges
35213 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35214 some code into new thread_outgoing_edges.
35215 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35216 definition. Simplify marker handling (do it here). Assume we always
35217 have the available expression and the const/copies tables.
35218 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35220 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35221 * tree-vrp.c (equiv_stack): No longer file scoped.
35222 (vrp_dom_walker): New class.
35223 (vrp_dom_walker::before_dom_children): New member function.
35224 (vrp_dom_walker::after_dom_children): Likewise.
35225 (identify_jump_threads): Setup domwalker. Use it rather than
35226 walking edges in a random order by hand. Simplify setup/finalization.
35227 (finalize_jump_threads): Remove.
35228 (vrp_finalize): Do not call identify_jump_threads here.
35229 (execute_vrp): Do it here instead and call thread_through_all_blocks
35232 PR tree-optimization/71437
35233 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35235 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35237 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35238 (dom_opt_dom_walker::thread_across_edge): Remove
35239 handle_dominating_asserts argument. All callers changed.
35240 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35241 changes. Remove calls to lhs_of_dominating_assert. Other
35242 uses of handle_dominating_asserts turn into unconditional code
35243 (simplify_control_stmt_condition_1): Likewise.
35244 (simplify_control_stmt_condition): Likewise.
35245 (thread_through_normal_block, thread_across_edge): Likewise.
35246 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35247 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35248 object if it is not an SSA_NAME.
35249 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35250 before calling into the VRP specific simplifiers.
35251 (identify_jump_threads): Remove handle_dominating_asserts
35254 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35257 * tree-diagnostic.c (default_tree_printer): No longer static.
35258 * tree-diagnostic.h (default_tree_printer): New prototype.
35260 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35262 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35263 Change ins into fmov.
35265 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35267 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35268 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35269 Use h_con constraint for operand 1.
35270 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35271 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35273 2017-03-15 Jeff Law <law@redhat.com>
35275 PR tree-optimization/71437
35276 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35277 (record_temporary_equivalences): Use it.
35279 PR tree-optimization/71437
35280 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35281 tree-ssa-scopedtables.
35282 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35283 (record_conditions, record_cond, vuse_eq): Likewise.
35284 (record_edge_info): Adjust to API tweak of record_conditions.
35285 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35286 (record_temporary_equivalences, optimize_stmt): Likewise.
35287 (eliminate_redundant_computations): Likewise.
35288 (record_equivalences_from_stmt): Likewise.
35289 * tree-ssa-scopedtables.c: Include options.h and params.h.
35290 (vuse_eq): New function, moved from tree-ssa-dom.c
35291 (build_and_record_new_cond): Likewise.
35292 (record_conditions): Likewise. Accept vector of conditions rather
35293 than edge_equivalence structure for first argument.
35294 for the first argument.
35295 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35296 from tree-ssa-dom.c.
35297 (avail_exprs_stack::record_cond): Likewise.
35298 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35299 from tree-ssa-dom.c.
35300 (avail_exprs_stack): Add new member functions lookup_avail_expr
35302 (record_conditions): Declare.
35304 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35307 * lra-constraints.c (process_alt_operands): Increase reject for
35308 reloading an input/output operand.
35310 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35313 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35314 insns to convert from signed/unsigned char/short to IEEE 128-bit
35316 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35318 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35321 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35322 subreg of inner mode for values already in registers.
35324 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35326 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35327 iteration reg is used after the loop.
35329 2017-03-14 Martin Sebor <msebor@redhat.com>
35331 PR tree-optimization/79800
35332 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35333 precision in negative-positive range.
35334 (format_floating): Call non-const overload with adjusted precision.
35336 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35339 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35342 2017-03-14 Martin Sebor <msebor@redhat.com>
35344 PR middle-end/80020
35345 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35346 * builtins.def (aligned_alloc): Use it.
35349 * Makefile.in (GTFILES): Add calls.c.
35350 * calls.c: Include "gt-calls.h".
35352 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35354 PR rtl-optimization/79728
35355 * regs.h (struct target_regs): New field
35356 x_contains_allocatable_regs_of_mode.
35357 (contains_allocatable_regs_of_mode): New macro.
35358 * reginfo.c (init_reg_sets_1): Initialize it, and change
35359 contains_reg_of_mode so it includes global regs as well.
35360 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35361 rather than contains_regs_of_mode.
35363 2017-03-14 Martin Liska <mliska@suse.cz>
35365 * doc/invoke.texi: Document options that can't be combined with
35366 -fcheck-pointer-bounds.
35368 2017-03-14 Martin Liska <mliska@suse.cz>
35370 PR middle-end/79831
35371 * doc/invoke.texi (-Wchkp): Document the option.
35373 2017-03-14 Martin Liska <mliska@suse.cz>
35375 * Makefile.in: Install gcov-dump.
35377 2017-03-14 Martin Liska <mliska@suse.cz>
35379 * multiple_target.c (expand_target_clones): Bail out for
35380 an invalid attribute.
35382 2017-03-14 Richard Biener <rguenther@suse.de>
35384 * alias.c (struct alias_set_entry): Pack properly.
35385 * cfgloop.h (struct loop): Likewise.
35386 * cse.c (struct set): Likewise.
35387 * ipa-utils.c (struct searchc_env): Likewise.
35388 * loop-invariant.c (struct invariant): Likewise.
35389 * lra-remat.c (struct cand): Likewise.
35390 * recog.c (struct change_t): Likewise.
35391 * rtl.h (struct address_info): Likewise.
35392 * symbol-summary.h (function_summary): Likewise.
35393 * tree-loop-distribution.c (struct partition): Likewise.
35394 * tree-object-size.c (struct object_size_info): Likewise.
35395 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35396 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35397 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35398 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35399 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35400 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35401 (struct _stmt_vec_info): Likewise.
35403 2017-03-14 Martin Liska <mliska@suse.cz>
35406 * multiple_target.c (create_dispatcher_calls): Check that
35407 a target can create a function dispatcher.
35409 2017-03-14 Martin Liska <mliska@suse.cz>
35412 * multiple_target.c (expand_target_clones): Drop local.local
35413 flag for default implementation.
35415 2017-03-14 Richard Biener <rguenther@suse.de>
35417 PR tree-optimization/80030
35418 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35420 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35422 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35423 gcc_fallthrough() instead of __attribute__((fallthrough));
35425 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35427 * doc/gcc.texi: Remove "up" link to (DIR).
35428 * doc/gccint.texi: Ditto.
35430 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35432 * doc/install.texi (Specific) <avr>: Remove reference to
35435 2017-03-13 Jeff Law <law@redhat.com>
35437 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35438 attribute rather than comments.
35440 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35441 match_scratch operand is highest.
35443 2017-03-13 Martin Liska <mliska@suse.cz>
35445 PR middle-end/78339
35446 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35447 is a CHKP clone, use original declaration.
35449 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35451 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35452 (arc_conditional_register_usage): Use a different allocation order
35453 when optimizing for size.
35454 * common/config/arc/arc-common.c (arc_option_optimization_table):
35455 Section anchors default on when optimizing for size.
35457 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35459 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35461 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35463 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35464 * config/arc/arc.md (cpu_facility): Add cd variant.
35465 (*movqi_insn): Add code density variant.
35466 (*movhi_insn): Likewise.
35467 (*movqi_insn): Likewise.
35468 (*addsi3_mixed): Likewise.
35469 (subsi3_insn): Likewise.
35471 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35473 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35475 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35477 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35478 expressions with MINUS and UNARY ops.
35480 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35483 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35485 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35486 between vec_select and vector argument.
35487 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35488 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35489 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35490 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35491 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35492 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35494 2017-03-13 Richard Biener <rguenther@suse.de>
35497 * params.def (vect-max-peeling-for-alignment): Fix typo.
35499 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35501 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35502 issue that only occurred with binutils below 2.18.
35504 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35506 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35507 refer to binutils 2.11/2.12 minimum.
35509 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35511 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35512 ftp.kernel.org and simplify binutils requirement.
35514 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35516 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35518 (Optimize Options): Ditto. Also remove redundancy.
35520 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35522 PR translation/79848
35523 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35525 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35526 to G_ to avoid double translation.
35528 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35530 PR translation/79923
35531 * auto-profile.c (get_combined_location): Convert leading
35532 character of diagnostics to lower case and remove trailing period.
35533 (read_profile): Likewise for various diagnostics.
35534 * config/arm/arm.c (arm_option_override): Remove trailing period
35535 from various diagnostics.
35536 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35537 (msp430_expand_delay_cycles): Likewise.
35539 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35542 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35543 full command-line argument, rather than just "str".
35544 (aarch64_validate_march): Likewise.
35545 (aarch64_validate_mtune): Likewise.
35547 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35549 PR rtl-optimization/78911
35550 * lra-assigns.c (must_not_spill_p): New function.
35551 (spill_for): Use it.
35553 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35555 PR tree-optimization/79981
35556 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35557 ATOMIC_COMPARE_EXCHANGE ifn result.
35558 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35559 IFN_ATOMIC_COMPARE_EXCHANGE.
35561 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35564 * opts.c (parse_sanitizer_options): Add missing question mark to
35565 "did you mean" message.
35567 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35569 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35571 (VMULEUH_UNS): Likewise.
35572 (VMULOUB_UNS): Likewise.
35573 (VMULOUH_UNS): Likewise.
35574 * config/rs6000/rs6000.c (builtin_function_type): Remove
35575 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35577 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35580 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35581 x with result of extra_parsing_for_operand_code_0.
35582 (function_reader::extra_parsing_for_operand_code_0): Convert
35583 return type from void to rtx, returning x. When reading
35584 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35585 larger size containing struct block_symbol.
35587 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35590 -mfloat128-hardware without -m64.
35592 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35595 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35596 entries to the case statement that marks unsigned arguments to
35597 overloaded functions.
35599 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35601 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35602 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35604 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35607 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35608 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35610 2017-03-10 Martin Liska <mliska@suse.cz>
35614 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35615 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35618 2017-03-10 Olivier Hainque <hainque@adacore.com>
35620 * tree-switch-conversion (array_value_type): Start by resetting
35621 candidate type to it's main variant.
35623 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35625 PR rtl-optimization/79909
35626 * combine.c (try_combine): Use simplify_replace_rtx on individual
35627 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35628 of the whole CALL_INSN_FUNCTION_USAGE.
35630 PR tree-optimization/79972
35631 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35632 get_range_info on SSA_NAMEs. Formatting fixes.
35634 2017-03-10 Richard Biener <rguenther@suse.de>
35635 Jakub Jelinek <jakub@redhat.com>
35637 PR tree-optimization/77975
35638 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35639 edge to be constant.
35640 (get_val_for): For constant x return it. Formatting fix.
35641 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35642 would use the same bases as the current one.
35644 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35646 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35647 instead of vec_select for V1TImode.
35648 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35650 (VSX_LE_128): Add V1TI to this mode iterator.
35651 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35652 (*vsx_le_perm_store_<mode>): Likewise.
35653 (pre-reload splitter for VSX stores): Likewise.
35654 (post-reload splitter for VSX stores): Likewise.
35655 (*vsx_xxpermdi2_le_<mode>): Likewise.
35656 (*vsx_lxvd2x2_le_<mode>): Likewise.
35657 (*vsx_stxvd2x2_le_<mode>): Likewise.
35659 2017-03-09 Michael Eager <eager@eagercon.com>
35661 Correct failures with --enable-checking=yes,rtl.
35663 * config/microblaze/microblaze.c (microblaze_expand_shift):
35664 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35665 test for const0_rtx.
35666 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35667 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35669 2017-03-09 Richard Biener <rguenther@suse.de>
35671 PR tree-optimization/79977
35672 * graphite-scop-detection.c (scop_detection::merge_sese):
35673 Handle the case of extra exits to blocks dominating the entry.
35675 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35677 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35680 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35682 PR rtl-optimization/79949
35683 * lra-constraints.c (process_alt_operands): Check memory when
35684 trying to predict a cycle. Print about the overall increase.
35686 2017-03-09 Richard Biener <rguenther@suse.de>
35688 PR middle-end/79971
35689 * gimple-expr.c (useless_type_conversion_p): Preserve
35690 TYPE_SATURATING for fixed-point types.
35692 2017-03-09 Richard Biener <rguenther@suse.de>
35695 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35696 alignment of BLKmode params.
35698 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35701 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35702 (VALL_NO_V2Q): Likewise.
35704 * config/aarch64/aarch64-simd.md
35705 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35707 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35708 VALL_NO_V2Q mode iterator.
35709 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35711 2017-03-09 Martin Liska <mliska@suse.cz>
35713 PR tree-optimization/79631
35714 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35715 tree_int_cst_sign_bit just for INTEGER constants.
35717 2017-03-09 Martin Liska <mliska@suse.cz>
35721 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35724 2017-03-09 Marek Polacek <polacek@redhat.com>
35727 * tree.c (inchash::add_expr): Handle TREE_VEC.
35729 2017-03-09 Martin Liska <mliska@suse.cz>
35732 (chkp_narrow_size_and_offset): New function.
35733 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35734 (void chkp_parse_bit_field_ref): New function.
35735 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35736 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35738 2017-03-09 Martin Liska <mliska@suse.cz>
35741 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35742 (chkp_find_bounds_1): Remove gcc_unreachable.
35744 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35747 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35748 BUILT_IN_SYNC*, determine the access type from the size suffix and
35749 always build a MEM_REF with that type. Handle forgotten
35750 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35753 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35754 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35755 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35756 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35757 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35758 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35759 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35760 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35761 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35762 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35763 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35764 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35765 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35766 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35767 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35768 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35769 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35770 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35771 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35772 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35773 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35774 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35775 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35776 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35777 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35778 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35779 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35780 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35781 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35782 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35783 definitions outside of __OPTIMIZE__ guarded section.
35786 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35787 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35788 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35789 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35792 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35794 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35795 ("vfenez<mode>"): Add missing constraints.
35797 2017-03-08 Martin Sebor <msebor@redhat.com>
35800 * config/nds32/nds32.c (nds32_option_override):
35801 Fix misspelled diagnostic.
35803 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35806 * gimplify.c (gimplify_omp_for): Replace index var in outer
35807 taskloop statement with an artificial variable and add
35808 OMP_CLAUSE_PRIVATE clause for it.
35810 2017-03-08 Richard Biener <rguenther@suse.de>
35812 PR tree-optimization/79955
35813 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35814 for accesses that are completely outside of the variable.
35816 2017-03-08 Andrew Haley <aph@redhat.com>
35818 PR tree-optimization/79943
35819 * tree-ssa-loop-split.c (compute_new_first_bound): When
35820 calculating the new upper bound, (END-BEG) should be added, not
35823 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35825 * config/avr/avr.md (setmemhi): Make sure match_dup
35826 operand number comes before match_scratch.
35828 2017-03-08 Richard Biener <rguenther@suse.de>
35830 PR tree-optimization/79920
35831 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35832 with ncopies == 1 to ...
35833 (vect_transform_slp_perm_load): ... here. Properly compute
35834 all element loads by iterating VF times over the group. Do
35835 not handle ncopies (computed in a broken way) in
35836 vect_create_mask_and_perm.
35838 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35841 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35842 is a uniform vector, use uniform_vector_p return value instead of
35843 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35845 2017-03-07 Marek Polacek <polacek@redhat.com>
35847 PR middle-end/79809
35848 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35849 (alloca_call_type): Likewise.
35851 2017-03-07 Martin Liska <mliska@suse.cz>
35853 * gcov.c (process_args): Put comment to correct location.
35855 2017-03-07 Martin Liska <mliska@suse.cz>
35857 PR middle-end/68270
35858 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35859 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35860 (chkp_narrow_bounds_for_field): Likewise.
35861 (chkp_parse_array_and_component_ref): Pass one more argument to
35864 2017-03-07 Richard Biener <rguenther@suse.de>
35866 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35869 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35871 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35872 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35874 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35877 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35878 to end of description.
35879 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35881 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35883 PR rtl-optimization/79901
35884 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35886 (*avx512f_<code><mode>3<mask_name>): ... this.
35887 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35888 iterator instead of VI8_AVX2_AVX512BW.
35890 PR rtl-optimization/79901
35891 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35892 min/max expander, expand it using expand_vec_cond_expr.
35895 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35898 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35901 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35902 to void * for PCH reasons.
35903 * dwarf2out.c (output_loc_operands, output_die): Cast
35904 v.val_vec.array to unsigned char *.
35906 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35909 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35912 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35914 PR rtl-optimization/79571
35915 * lra-constraints.c (process_alt_operands): Calculate static
35916 reject and subtract it from overall when only addresses will be
35919 2017-03-06 Julia Koval <julia.koval@intel.com>
35922 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35923 incoming stack boundary to 128 for 64-bit targets.
35925 2017-03-06 Richard Biener <rguenther@suse.de>
35927 PR tree-optimization/79894
35928 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35929 to NULL after folding it.
35931 2017-03-06 Richard Biener <rguenther@suse.de>
35933 PR tree-optimization/79824
35934 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35935 check disabling peeling for gaps.
35937 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35939 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35940 attributes): Document gettimeofday.
35942 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35944 * config/s390/s390.c (s390_option_override_internal): Set
35945 PARAM_MIN_VECT_LOOP_BOUND
35947 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35949 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35950 * config/s390/s390.md: Likewise.
35952 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35955 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35956 (<avx2_avx512>_perm<mode>): Rename to ...
35957 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35959 (<avx512>_perm<mode>_mask): Rename to ...
35960 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35962 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35963 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35964 instead of VI8F_256_512.
35965 (avx512f_perm<mode>): New define_expand.
35966 (avx512f_perm<mode>_mask): Likewise.
35967 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35968 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35970 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35972 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35973 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35975 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35977 2017-03-06 Martin Liska <mliska@suse.cz>
35980 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35981 when having a SSA NAME w/o VAR_DECL assigned to it.
35983 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35985 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35986 msa_dpsub_<su>_d): Fix MODE for vec_select.
35988 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35990 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35992 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35994 2017-03-06 Richard Biener <rguenther@suse.de>
35996 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35997 * plugin.c (register_plugin_info): Likewise.
35998 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36000 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36002 * config/i386/sse.md (sse_storehps, sse_storelps,
36003 avx_<castmode><avxsizesuffix>_<castmode>,
36004 avx512f_<castmode><avxsizesuffix>_<castmode>,
36005 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36006 in condition that at least one operand is not a MEM.
36008 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36010 PR middle-end/79805
36011 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36012 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36014 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36015 gimple_call_nothrow_p flag based on whether original builtin can throw.
36016 If it can, emit following stmts on the fallthrough edge.
36017 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36018 don't create new bb if inserting just debug stmts on the edge, try to
36019 insert them on the fallthru bb or just reset debug stmts.
36021 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36024 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36025 restore recog_data (including the operand rtxes inside it) around
36026 the call to get_insn_template.
36028 2017-03-03 Martin Sebor <msebor@redhat.com>
36030 PR tree-optimization/79699
36031 * context.c (context::~context): Free MPFR caches to avoid
36032 a memory leak on program exit.
36034 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36036 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36037 Use wide_int::ulow () instead of .elt (0).
36039 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36041 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36042 (*pushxf): Limit oF constraint to 32bit targets and add oC
36043 constraint for 64bit targets.
36044 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36045 (*pushdf): Change rmF constraint to rmC.
36047 2017-03-03 Martin Liska <mliska@suse.cz>
36049 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36050 Remove unused variable.
36052 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36055 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36056 is a memory operand, increase num_memory.
36057 (ix86_expand_args_builtin): Likewise.
36059 2017-03-03 Jan Hubicka <jh@suse.cz>
36062 * ipa-devirt.c (maybe_record_node): Properly handle
36063 __cxa_pure_virtual visibility.
36065 2017-03-03 Martin Liska <mliska@suse.cz>
36067 PR tree-optimization/79803
36068 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36070 (pass_loop_prefetch::execute): Disabled optimization if an
36071 assumption about L1 cache size is not met.
36073 2017-03-03 Martin Liska <mliska@suse.cz>
36075 PR rtl-optimization/79574
36076 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36077 (hash_scan_set): Likewise.
36078 (dump_hash_table): Likewise.
36079 (hoist_code): Likewise.
36081 2017-03-03 Richard Biener <rguenther@suse.de>
36083 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36084 in place of uhigh ().
36085 (fixed_convert_from_real): Likewise.
36087 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36090 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36092 2017-03-03 Richard Biener <rguenther@suse.de>
36094 PR middle-end/79818
36095 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36096 TYPE_OVERFLOW_UNDEFINED check.
36098 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36100 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36102 (vector_ae_<mode>_p): Likewise.
36103 (vector_nez_<mode>_p): Likewise.
36104 (vector_ne_v2di_p): Likewise.
36105 (vector_ae_v2di_p): Likewise.
36106 (vector_ne_<mode>_p): Likewise.
36107 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36109 (vsx_tsqrt<mode>2_fe): Likewise.
36111 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36114 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36116 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36118 PR rtl-optimization/79780
36119 * cprop.c (one_cprop_pass): When second and further conditional trap
36120 in a single basic block is turned into an unconditional trap, turn it
36121 into a deleted note to avoid RTL verification failures.
36123 2017-03-02 Richard Biener <rguenther@suse.de>
36125 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36127 2017-03-02 Richard Biener <rguenther@suse.de>
36129 PR tree-optimization/79345
36131 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36132 param and abort the walk, returning -1 if it is hit.
36133 (walk_aliased_vdefs): Take a limit param and pass it on.
36134 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36135 defaulting to 0 and return a signed int.
36136 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36137 (check_defs): New helper.
36138 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36139 about uninitialized memory.
36140 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36141 bogus uninitialized warning.
36142 (fixed_convert_from_real): Likewise.
36144 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36146 PR tree-optimization/66768
36147 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36148 iv_use if base object can't be determined.
36150 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36152 PR tree-optimization/79345
36153 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36154 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36155 (get_pattern_stats): Initialize it.
36156 * genemit.c (gen_expand): Verify match_scratch numbers come after
36157 match_operand/match_dup numbers.
36158 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36159 match_scratch numbers.
36160 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36162 * config/s390/s390.md (trunctdsd2): Likewise.
36164 2017-03-02 Richard Biener <rguenther@suse.de>
36166 * wide-int.h (wide_int_storage::operator=): Implement in terms
36169 2017-03-02 Richard Biener <rguenther@suse.de>
36171 PR tree-optimization/79777
36172 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36173 the to insert expression to sth existing.
36175 2017-03-01 Martin Sebor <msebor@redhat.com>
36177 PR middle-end/79692
36178 * gimple-ssa-sprintf.c
36179 (directive::known_width_and_precision): New function.
36180 (format_integer): Use it.
36181 (get_mpfr_format_length): Consider the full range of precision
36182 when computing %g output with the # flag. Set the likely byte
36183 count to 3 rather than 1 when precision is indeterminate.
36184 (format_floating): Correct the lower bound of precision.
36186 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36188 * doc/invoke.texi: Document default code model for 64-bit Linux.
36190 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36193 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36194 udiv rather than div since input pattern is unsigned.
36196 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36198 * config/i386/i386.c (print_reg): Warn for values of
36199 unsupported size in integer register.
36201 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36204 * config/rs6000/predicates.md (current_file_function_operand): Do
36205 not allow self calls to be local if the function is replaceable.
36207 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36210 * config/rs6000/altivec.h (vec_ctz and others): Change the
36211 preprocessor macro that controls conditional compilation from
36212 _ARCH_PWR9 to __POWER9_VECTOR__.
36213 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36214 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36215 control (instead of _ARCH_PWR9 control) so that template
36216 definition uses power9-specific function.
36217 (vec_any_eq): Likewise.
36218 (vec_all_ne): Change macro definition to use a power9-specific
36219 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36220 _ARCH_PWR9 control).
36221 (vec_any_eq) Likewise.
36222 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36223 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36224 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36225 support for xvcmpnedp instruction.
36226 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36227 macro expansion so that Power9 implementation of vec_all_ne does
36228 not use the AltiVec predicate framework.
36229 (VCMPNEH_P): Likewise.
36230 (VCMPNEW_P): Likewise.
36231 (VCMPNED_P): Likewise.
36232 (VCMPNEFP_P): Likewise.
36233 (VCMPNEDP_P): Likewise.
36234 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36235 implementation of vec_any_eq to not use AltiVec predicate
36237 (VCMPAEH_P): Likewise.
36238 (VCMPAEW_P): Likewise.
36239 (VCMPAED_P): Likewise.
36240 (VCMPAEFP_P): Likewise.
36241 (VCMPAEDP_P): Likewise.
36242 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36243 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36244 not use the AltiVec predicate framework.
36245 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36246 of vec_any_eq to not use AltiVec predicate framework.
36247 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36248 support for predefined __POWER9_VECTOR__ macro to indicate that
36249 Power9 instruction selection is enabled.
36250 (altivec_overloaded_builtins): Remove extraneous
36251 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36252 function argument types RS6000_BTI_bool_V16QI and
36253 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36254 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36255 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36256 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36257 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36258 Power9 for implementations of vec_cmpne. Change the signature for
36259 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36260 (representing vec_all_ne) to remove the previously described first
36261 argument of type RS6000_BTI_INTSI, as this was an artifact of
36262 reliance on the AltiVec predicate framework, which is no longer
36263 used in the implementation of these functions. Add
36264 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36265 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36266 since, unlike the AltiVec predicate framework implementation, we
36267 do not share function descriptors between vec_alle and vec_anyeq.
36268 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36269 set of modes that receive special treatment even when
36270 TARGET_P9_VECTOR is true. The special treatment emits code that
36271 does not depend on Power9 instructions.
36272 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36273 define_expand to not rely on AltiVec predicate framework.
36274 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36276 (vector_ne_v2di_p): Change this define_expand to not rely on
36277 AltiVec predicate framework.
36278 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36280 (vector_ne_<mode>_p): Change this define_expand to not rely on
36281 AltiVec predicate framework.
36282 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36284 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36285 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36286 define_insn pattern.
36287 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36288 define_insn pattern because the xvcmpne<VSs>. instruction is not
36290 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36291 instruction is not supported.
36293 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36295 * config/nvptx/nvptx.c: Include intl.h.
36297 2017-03-01 Martin Jambor <mjambor@suse.cz>
36300 * ipa-prop.h (ipa_bits): Removed field known.
36301 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36302 to pointers. Adjusted their comments to warn about their sharing.
36303 (ipcp_transformation_summary): Change bits to a vector of pointers.
36304 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36305 (ipa_get_ipa_bits_for_value): Declare.
36306 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36307 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36308 (ipa_bits_hash_table): Likewise.
36309 (ipa_vr_ggc_hash_traits): Likewise.
36310 (ipa_vr_hash_table): Likewise.
36311 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36312 being pointers and vr_known being removed.
36313 (ipa_set_jf_unknown): Likewise.
36314 (ipa_get_ipa_bits_for_value): New function.
36315 (ipa_set_jfunc_bits): Likewise.
36316 (ipa_get_value_range): New overloaded functions.
36317 (ipa_set_jfunc_vr): Likewise.
36318 (ipa_compute_jump_functions_for_edge): Use the above functions to
36319 construct bits and vr parts of jump functions.
36320 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36321 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36323 (ipcp_grow_transformations_if_necessary): Also allocate
36324 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36326 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36327 them. Fix too long lines.
36328 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36329 vr_known being removed.
36330 (ipa_read_jump_function): Use new setter functions to construct bits
36331 and vr parts of jump functions or set them to NULL.
36332 (write_ipcp_transformation_info): Adjust for bits being pointers.
36333 (read_ipcp_transformation_info): Likewise.
36334 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36336 Include gt-ipa-prop.h.
36337 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36339 (ipcp_store_bits_results): Likewise.
36340 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36341 Do not write to existing jump functions but use a temporary instead.
36343 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36346 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36347 attempt to use its first operand as BIT_FIELD_REF base.
36349 2017-03-01 Richard Biener <rguenther@suse.de>
36351 PR middle-end/79721
36352 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36353 interpolating formula in wrapping arithmetic.
36354 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36356 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36358 PR tree-optimization/79734
36359 * tree-vect-generic.c (expand_vector_condition): Optimize
36360 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36361 Handle VEC_COND_EXPR where comparison has different inner width from
36362 type's inner width.
36364 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36366 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36367 markup, and similar issues. Remove @opindex entries for things
36368 that aren't options. Add missing -mmpy-option entries.
36370 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36372 PR tree-optimization/79737
36373 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36374 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36375 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36376 instead of byte_size. Formatting fix.
36377 (shift_bytes_in_array_right): Formatting fix.
36379 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36382 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36383 condition on optimize for the leaf function test.
36385 2017-02-28 Martin Liska <mliska@suse.cz>
36388 * read-rtl-function.c (function_reader::handle_unknown_directive):
36389 Bail out when one uses -flto.
36391 2017-02-28 Martin Liska <mliska@suse.cz>
36393 * common.opt: Replace space with tabular for options of <number>
36395 * config/i386/i386.opt: Show <number> value for
36396 -mlarge-data-threshold.
36397 * opts.c (print_filtered_help): Do not display number in hexadecimal
36400 2017-02-28 Martin Liska <mliska@suse.cz>
36402 * common.opt: Fix --help=option -Q for options which are of
36405 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36407 * config/i386/i386.c (print_reg): Error out for values
36408 of 8-bit size in invalid integer register.
36410 2017-02-28 Martin Sebor <msebor@redhat.com>
36412 PR tree-optimization/79691
36413 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36415 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36418 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36419 gcc_unreachable with output_operand_lossage.
36421 2017-02-28 Richard Biener <rguenther@suse.de>
36423 PR tree-optimization/79740
36424 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36426 (visit_nary_op): Insert the nary into the hashtable if we
36427 pattern-matched sth.
36428 * tree-ssa-pre.c (eliminate_insert): Robustify.
36430 2017-02-28 Richard Biener <rguenther@suse.de>
36432 PR middle-end/79731
36433 * fold-const.c (decode_field_reference): Reject out-of-bound
36436 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36438 * config/i386/i386.c: Include intl.h.
36439 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36440 instead of just cond ? "..." : "...".
36441 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36442 * coverage.c (read_counts_file): Likewise.
36443 * omp-offload.c: Include intl.h.
36444 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36445 of just cond ? "..." : "...".
36446 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36447 of just cond ? "..." : "...".
36449 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36452 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36454 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36455 'tune for' CPU name.
36456 * config/arm/arm-cpu-data.h: Regenerated.
36458 2017-02-28 Richard Biener <rguenther@suse.de>
36460 PR tree-optimization/79732
36461 * tree-inline.c (expand_call_inline): Do not shadow var.
36463 2017-02-28 Richard Biener <rguenther@suse.de>
36465 PR tree-optimization/79723
36466 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36467 address-space properly.
36469 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36471 * doc/optinfo.texi (Optimization groups): Fix option used for
36473 * doc/invoke.texi (-fopt-info): Document "omp".
36474 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36475 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36476 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36477 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36478 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36480 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36482 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36483 OPTGROUP_OMP with "omp".
36485 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36488 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36489 for arithmetic shift of unsigned V2DI.
36491 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36493 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36494 arc/linux.h headers.
36495 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36496 (LINK_SPEC): Likewise.
36497 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36498 (EXTRA_SPECS): Likewise.
36499 (STARTFILE_SPEC): Likewise.
36500 (ENDFILE_SPEC): Likewise.
36501 (LIB_SPEC): Likewise.
36502 (TARGET_SDATA_DEFAULT): Likewise.
36503 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36504 (MULTILIB_DEFAULTS): Likewise.
36505 (DWARF2_UNWIND_INFO): Likewise.
36506 * config/arc/big.h: New file.
36507 * config/arc/elf.h: Likewise.
36508 * config/arc/linux.h: Likewise.
36509 * config/arc/t-uClibc: Remove.
36511 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36513 PR tree-optimization/77536
36514 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36515 (tree_transform_and_unroll_loop): Use above function to compute the
36516 estimated niter of unrolled loop and use it when scaling profile.
36517 Also use count info rather than frequency if it's non-zero.
36518 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36519 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36520 (vect_transform_loop): Call above function.
36522 2017-02-27 Richard Biener <rguenther@suse.de>
36524 PR tree-optimization/45397
36525 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36526 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36527 (visit_nary_op): Add pattern matching for CSEing sign-changed
36528 or truncated operations with wider ones.
36530 2017-02-27 Richard Biener <rguenther@suse.de>
36532 PR tree-optimization/79690
36533 * tree-vect-stmts.c (vectorizable_store): Use vector type
36534 built from the DR with address-space.
36536 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36538 * doc/invoke.texi (Optimize Options): Refine the description
36539 of asan-use-after-return.
36541 2017-02-25 Alan Modra <amodra@gmail.com>
36543 PR rtl-optimization/79584
36544 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36545 base, not ad->base_term, the reg within base. Remove assertion
36546 that ad->base == ad->base_term. Replace gen_int_mode using
36547 bogus mode with const0_rtx.
36549 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36551 PR middle-end/79396
36552 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36553 FMA_EXPR like tcc_binary or tcc_unary.
36555 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36558 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36560 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36561 (output_loc_operands): Handle DW_OP_call_ref and
36562 DW_OP_GNU_variable_value.
36563 (struct variable_value_struct): New type.
36564 (struct variable_value_hasher): Likewise.
36565 (variable_value_hash): New variable.
36566 (string_types): Remove.
36567 (copy_loc_descr): New function.
36568 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36569 (prepend_loc_descr_to_each): New function.
36570 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36571 instead of add_loc_descr_to_each if the first argument is single
36572 location list and the second has multiple.
36573 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36574 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36575 when looking for variable value which doesn't have other location info.
36576 (loc_list_from_tree): Formatting fix.
36577 (gen_array_type_die): Simplify DW_AT_string_length handling.
36578 (adjust_string_types): Remove.
36579 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36580 string_types. Call resolve_variable_values.
36581 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36582 (resolve_addr_in_expr): Likewise. Add A argument.
36583 (copy_deref_exprloc): Remove deref argument. Adjust for the
36584 original expression being DW_OP_GNU_variable_value with optionally
36585 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36586 optionally after it.
36587 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36588 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36589 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36590 (variable_value_hasher::hash, variable_value_hasher::equal): New
36592 (resolve_variable_value_in_expr, resolve_variable_value,
36593 resolve_variable_values, note_variable_value_in_expr,
36594 note_variable_value): New functions.
36595 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36598 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36601 * opts.h (handle_generated_option): Add GENERATED_P argument.
36602 * opts-common.c (handle_option): Adjust function comment.
36603 (handle_generated_option): Add GENERATED_P argument, pass it to
36605 (control_warning_option): Pass false to handle_generated_option
36607 * opts.c (maybe_default_option): Pass true to handle_generated_option
36609 * optc-gen.awk: Likewise.
36611 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36613 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36614 a REG, look at the REG it is a SUBREG of.
36615 (splitter for cmpeqsi_t): Ditto.
36617 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36619 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36620 the special USEs with the pattern of the insn, not the insn itself.
36622 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36625 * doc/invoke.texi: Document -mload-store-pairs.
36627 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36628 Sandra Loosemore <sandra@codesourcery.com>
36630 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36631 argument isn't a CONST_INT.
36632 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36633 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36634 (nios2_validate_compare): Bypass alternate compare logic if *op2
36635 is not a CONST_INT.
36636 (ldstwm_operation_p): Return false if first_base is not a REG or
36637 if first_offset is not a CONST_INT.
36639 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36641 * config/cris/cris.md: Use correct operand in a define_peephole2.
36643 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36645 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36647 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36649 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36650 this_insn if it is an INSN or JUMP_INSN.
36651 (force_offsettable): Look at base, not at addr.
36652 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36653 on things that aren't necessarily CONST_INTs.
36655 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36657 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36658 -mfpmath=sse is the default also for x86-32 targets with SSE2
36659 instruction set when @option{-ffast-math} is enabled
36661 2017-02-24 Jeff Law <law@redhat.com>
36663 PR rtl-optimizatoin/79286
36664 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36667 2017-02-24 Richard Biener <rguenther@suse.de>
36669 PR tree-optimization/79389
36670 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36673 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36675 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36676 function comment to reflect reality.
36677 (loop_exits_before_overflow): Fix typo in function description.
36679 2017-02-24 Richard Biener <rguenther@suse.de>
36681 PR tree-optimization/79389
36682 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36683 properly that a threading opportunity exists. Detect conditional
36684 copy/constant propagation opportunities.
36686 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36688 * config/visium/visium.md (type): Add trap.
36689 (b): New mode attribute.
36690 (*btst): Rename into...
36691 (*btst<mode>): ...this and adjust.
36692 (*cbranchsi4_btst_insn): Rename into...
36693 (*cbranch<mode>4_btst_insn): ...this and adjust.
36694 (trap): New define_insn.
36696 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36698 PR tree-optimization/79389
36699 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36700 (noce_reversed_cond_code): New function.
36701 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36702 reversed_comparison_code. Formatting fix.
36703 (noce_try_store_flag): Test rev_cond != NULL in addition to
36704 reversed_comparison_code.
36705 (noce_try_store_flag_constants): Likewise.
36706 (noce_try_store_flag_mask): Likewise.
36707 (noce_try_addcc): Use rev_cond if non-NULL instead of
36708 reversed_comparison_code.
36709 (noce_try_cmove_arith): Likewise. Formatting fixes.
36710 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36711 (noce_find_if_block): Initialize rev_cond.
36712 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36713 instead of false as last argument never attempt to reverse it
36716 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36718 PR tree-optimization/79663
36719 * tree-predcom.c (combine_chains): Process refs in reverse order
36720 only for ZERO length chains, and add explaining comment.
36722 2017-02-23 Jeff Law <law@redhat.com>
36724 PR tree-optimization/79578
36725 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36726 in call to operand_equal_p.
36728 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36731 * config/i386/cpuid.h: Fix another undefined behavior.
36733 2017-02-23 Richard Biener <rguenther@suse.de>
36735 PR tree-optimization/79683
36736 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36737 vector types for data-refs.
36739 2017-02-23 Martin Liska <mliska@suse.cz>
36741 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36743 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36745 PR middle-end/79665
36746 * internal-fn.c (get_range_pos_neg): Moved to ...
36747 * tree.c (get_range_pos_neg): ... here. No longer static.
36748 * tree.h (get_range_pos_neg): New prototype.
36749 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36750 are known to be in between 0 and signed maximum inclusive, try to
36751 expand both unsigned and signed divmod and use the cheaper one from
36754 2017-02-22 Jeff Law <law@redhat.com>
36756 PR tree-optimization/79578
36757 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36758 to compare base operands.
36760 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36763 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36764 gpc_reg_operand instead of fpr_reg_operand.
36766 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36768 * config/mips/mips.c (mips_return_in_memory): Force FP
36769 vector types to be returned in memory for o32 ABI.
36771 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36773 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36774 instead of DW_TAG_member for static data member declarations and don't
36775 set no_linkage_name for static inline data members.
36776 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36779 2017-02-22 Martin Liska <mliska@suse.cz>
36781 * doc/invoke.texi: Replace inequality signs with square brackets
36784 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36787 * lra-constraints.c (simplify_operand_subreg): Handle
36788 WORD_REGISTER_OPERATIONS targets.
36790 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36793 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36794 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36795 elimination by swapping fld*.
36797 2017-02-22 Richard Biener <rguenther@suse.de>
36799 PR tree-optimization/79673
36800 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36801 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36802 irrelevant address-space qualifiers and avoiding a
36803 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36805 2017-02-22 Richard Biener <rguenther@suse.de>
36807 PR tree-optimization/79666
36808 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36809 to not symbolically negate if that may introduce undefined
36812 2017-02-22 Martin Liska <mliska@suse.cz>
36815 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36816 * data-streamer-out.c (streamer_write_gcov_count_stream):
36818 * value-prof.c (stream_out_histogram_value): Make assert more
36819 precise based on type of counter.
36821 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36824 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36825 Use nonimmediate_operand instead of memory_operand for operand 1.
36826 (float-extend standard_x87sse_constant_load splitter): Ditto.
36828 2017-02-21 Jeff Law <law@redhat.com>
36830 PR tree-optimization/79621
36831 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36832 blocks with edges to themselves.
36834 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36837 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36838 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36839 Use gimple_call_builtin_p.
36842 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36843 on temporarily removed DEBUG_INSNs.
36845 PR tree-optimization/79649
36846 * tree-loop-distribution.c (classify_partition): Give up on
36847 non-generic address space loads/stores.
36849 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36851 * doc/loop.texi (Loop manipulation): Remove nonexistent
36852 tree_ssa_loop_version from the documentation.
36853 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36855 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36858 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36859 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36860 * config/rs6000/rs6000.c: Include except.h.
36861 (rs6000_expand_split_stack_prologue): Call
36862 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36864 2017-02-21 Martin Jambor <mjambor@suse.cz>
36867 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36868 have been analyzed.
36870 2017-02-21 Martin Jambor <mjambor@suse.cz>
36872 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36873 for backward compatibility only.
36874 * doc/invoke.texi (Option Summary): Remove all references to
36875 -fipa-cp-alignment.
36877 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36881 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36883 * lra-constraints.c (curr_insn_transform): Handle
36884 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36886 2017-02-21 Martin Liska <mliska@suse.cz>
36888 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36890 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36892 PR translation/79638
36893 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36895 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36898 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36899 (arm_function_ok_for_sibcall): Return false for an indirect call by
36900 descriptor if all the argument registers are used.
36901 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36902 alignment of the function.
36904 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36906 PR tree-optimization/61441
36907 * simplify-rtx.c (simplify_const_unary_operation): For
36908 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36909 the sNaN unmodified.
36911 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36913 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36914 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36915 instead of SYSTEM_HEADER_DIR.
36917 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36918 Martin Liška <mliska@suse.cz>
36920 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36921 Fix typos and grammar, use active voice, and clarify.
36923 2017-02-20 Marek Polacek <polacek@redhat.com>
36925 PR middle-end/79537
36926 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36929 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36931 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36934 * config/i386/i386.c (ix86_expand_builtin): Handle
36935 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36936 ix86_builtins_isa[fcode].isa as a requirement of those
36937 flags and any other flag in the bitmask.
36938 (ix86_init_mmx_sse_builtins): Use 0 instead of
36939 ~OPTION_MASK_ISA_64BIT as mask.
36940 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36941 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36942 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36943 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36945 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36948 * lra-constraints.c (split_reg): Check requested split mode
36949 is supported by the register.
36951 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36953 * lra-constraints.c (simplify_operand_subreg): Remove early
36956 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36959 * lra-constraints.c (curr_insn_transform): Tighten condition
36960 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36962 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36965 * lra-constraints.c (curr_insn_transform): Handle
36966 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36968 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36971 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36973 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36975 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36978 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36981 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36983 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36986 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36988 * final.c (last_columnnum, override_columnnum): New variables.
36989 (final_start_function): Set last_columnnum, pass it to begin_prologue
36990 hook and pass 0 to dwarf2out_begin_prologue.
36991 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36992 to source_line debug hook.
36993 (notice_source_line): Compute last_columnnum and for debug_column_info
36994 return true on column changes.
36995 * debug.h (struct gcc_debug_hooks): Add column argument to
36996 source_line and begin_prologue hooks.
36997 (debug_nothing_int_charstar_int_bool): Remove prototype.
36998 (debug_nothing_int_int_charstar,
36999 debug_nothing_int_int_charstar_int_bool): New prototypes.
37000 (dwarf2out_begin_prologue): Add column argument.
37001 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37002 begin_prologue hooks.
37003 (debug_nothing_int_charstar_int_bool): Remove.
37004 (debug_nothing_int_int_charstar,
37005 debug_nothing_int_int_charstar_int_bool): New functions.
37006 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37007 through to dwarf2out_source_line.
37008 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37009 (dwarf2out_source_line): Add column argument, emit it if requested.
37010 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37012 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37013 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37014 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37015 through to dwarf2out_begin_prologue.
37016 (vmsdbgout_source_line): Add column argument, pass it through to
37017 dwarf2out_source_line.
37018 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37019 dbxout_source_line caller.
37020 (dbxout_source_line): Add column argument.
37022 * common.opt (gno-column-info, gcolumn-info): New options.
37023 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37024 (check_die): Also test for multiple DW_AT_decl_column attributes.
37025 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37026 DW_AT_decl_column if requested.
37027 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37029 (gen_variable_die): Likewise.
37030 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37031 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37034 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37035 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37036 (ix86_handle_option): Handle OPT_m3dnowa.
37037 * doc/invoke.texi (-m3dnowa): Document.
37038 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37039 -m3dnowa instead of -m3dnow -march=athlon.
37042 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37043 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37045 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37048 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37049 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37050 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37051 generator for vsx_xxpermdi_<mode>_be.
37052 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37053 force big-endian semantics.
37054 (vsx_xxpermdi_<mode>_be): New define_expand with same
37055 implementation as previous version of vsx_xxpermdi_<mode>.
37057 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37059 PR tree-optimization/79327
37060 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37061 variable, its initialization and use.
37063 2017-02-17 Julia Koval <julia.koval@intel.com>
37065 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37066 (OPTION_MASK_ISA_PKU_UNSET): New.
37067 (ix86_handle_option): Handle -mrdpid.
37068 * config/i386/cpuid.h (bit_RDPID): New.
37069 * config/i386/driver-i386.c (host_detect_local_cpu):
37070 Detect RDPID feature.
37071 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37072 * config/i386/i386-c.c (ix86_target_macros_internal):
37074 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37075 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37076 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37077 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37078 * config/i386/i386.md (define_insn "rdpid"): New.
37079 * config/i386/i386.opt Add -mrdpid.
37080 * config/i386/immintrin.h (_rdpid_u32): New.
37082 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37084 PR rtl-optimization/79541
37085 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37086 instead of transforming it into USE.
37088 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37090 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37091 If HONOR_SNANS (SFmode) force the input to a register.
37092 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37093 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37094 an frsp or similar insn.
37096 2017-02-17 Martin Liska <mliska@suse.cz>
37098 PR rtl-optimization/79577
37099 * params.def (selsched-max-sched-times): Increase minimum to 1.
37101 2017-02-17 Martin Liska <mliska@suse.cz>
37103 PR rtl-optimization/79574
37104 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37106 2017-02-17 Martin Liska <mliska@suse.cz>
37108 PR tree-optimization/79529
37109 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37110 ssa_defined_default_def_p to handle cases which are implicitly
37112 * tree-ssa.c (ssa_defined_default_def_p): New function.
37113 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37114 which are implicitly defined.
37115 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37117 2017-02-17 Richard Biener <rguenther@suse.de>
37119 PR middle-end/79576
37120 * params.def (max-ssa-name-query-depth): Limit to 10.
37122 2017-02-17 Richard Biener <rguenther@suse.de>
37124 PR tree-optimization/79552
37125 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37128 2017-02-17 Richard Biener <rguenther@suse.de>
37131 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37133 2017-02-17 Marek Polacek <polacek@redhat.com>
37135 PR middle-end/79536
37136 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37137 (fold_negate_expr): New wrapper.
37139 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37141 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37142 Correct terminology and de-emphasize pre-standard behavior.
37144 2017-02-16 Alan Modra <amodra@gmail.com>
37146 PR rtl-optimization/79286
37147 * ira.c (def_dominates_uses): New function.
37148 (update_equiv_regs): Don't create an equivalence for insns that
37149 may trap where the register def does not dominate the use.
37151 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37153 PR rtl-optimization/78127
37154 * lra.c (lra): Call lra_eliminate before finish the loop after
37157 2017-02-16 Richard Biener <rguenther@suse.de>
37159 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37161 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37162 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37163 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37164 (isl_val_int_from_wi): New function.
37165 (extract_affine_gmp): Rename to ...
37166 (extract_affine_wi): ... this, take a widest_int.
37167 (extract_affine_int): Just wrap extract_affine_wi.
37168 (add_param_constraints): Use isl_val_int_from_wi.
37169 (add_loop_constraints): Likewise, and extract_affine_wi.
37171 2017-02-15 Jeff Law <law@redhat.com>
37173 PR middle-end/79521
37174 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37175 ira_init_register_move_cost_if_necessary.
37177 2017-02-15 Martin Sebor <msebor@redhat.com>
37179 PR middle-end/32003
37180 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37181 removed in a prior commit.
37183 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37185 PR tree-optimization/79347
37186 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37187 counters during peeling.
37189 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37191 * Makefile.in (site.exp): Remove "set ISLVER".
37193 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37196 * real.c (real_from_integer): Call real_convert even for decimal.
37198 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37201 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37203 2017-02-14 Andrew Pinski <apinski@cavium.com>
37205 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37206 cores and change the partno/implementer to be correct.
37207 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37208 the 'B" as the implementer.
37209 * config/aarch64/aarch64-tune.md: Regenerate.
37211 2017-02-14 Carl Love <cel@us.ibm.com>
37213 * config/rs6000/rs6000.c: Add case statement entry to make the
37214 xvcvuxdsp built-in argument unsigned.
37215 * config/rs6000/vsx.md: Fix the source and return operand types so they
37216 match the instruction definitions from the ISA document. Fix typo
37217 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37220 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37223 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37224 member early_clobber_alts.
37225 * lra-lives.c (reg_early_clobber_p): New.
37226 (process_bb_lives): Use it.
37227 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37228 (debug_operand_data): Initialize early_clobber_alts.
37229 (setup_operand_alternative): Set up early_clobber_alts.
37230 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37231 alternatives to new_insn_reg.
37232 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37234 (lra_update_insn_regno_info): Pass the new arg.
37236 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37238 PR middle-end/79505
37239 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37240 (new_oacc_loop_raw): Don't clear already cleared fields.
37243 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37244 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37245 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37247 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37250 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37252 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37255 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37256 the extra instruction to the right place to store 128-bit constant
37259 2017-02-14 Martin Sebor <msebor@redhat.com>
37261 PR middle-end/79448
37262 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37263 warning for strings of unknown length.
37265 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37267 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37269 2017-02-14 Jeff Law <law@redhat.com>
37272 * ira-costs.c (scan_one_insn): Initialize register move costs
37273 for pseudos seen in USE/CLOBBER insns.
37275 PR tree-optimization/79095
37276 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37277 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37278 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37279 if the operands are known to be not equal, then the resulting range
37281 (intersect_ranges): If the new range is ~[0,0] and the old range is
37282 wide, then prefer ~[0,0].
37283 * tree-vrp.c (overflow_comparison_p_1): New function.
37284 (overflow_comparison_p): New function.
37285 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37286 if NAME is used in an overflow test.
37287 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37288 overflow check that can be expressed as an equality test, then adjust
37289 ops to be that equality test.
37291 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37293 * config/s390/s390-builtin-types.def: Remove flags argument.
37294 * config/s390/s390.c (s390_init_builtins): Likewise.
37296 2017-02-14 Martin Liska <mliska@suse.cz>
37298 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37299 vector. Fix trailing white spaces.
37301 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37303 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37306 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37308 PR rtl-optimization/68664
37309 * config/arm/arm.c (arm_sched_can_speculate_insn):
37310 New function. Declare prototype.
37311 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37313 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37315 PR rtl-optimization/68664
37316 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37318 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37320 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37322 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37323 max skip bytes for function, loop and jump.
37325 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37327 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37328 ABS_EXPR for gimple dump.
37330 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37333 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37335 PR tree-optimization/79408
37336 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37337 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37338 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37339 also if rhs1 is INTEGER_CST.
37341 2017-02-14 Richard Biener <rguenther@suse.de>
37343 PR middle-end/79432
37344 * tree-into-ssa.c (insert_phi_nodes): When the function can
37345 have abnormal edges rewrite SSA names with broken use-def
37346 dominance out of SSA and register them for PHI insertion.
37348 2017-02-13 Martin Sebor <msebor@redhat.com>
37350 PR middle-end/79496
37351 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37352 clearing info.nowrite flag when snprintf size argument is a range.
37354 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37356 * cprop.c (cprop_jump): Add missing space in string literal.
37357 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37358 (get_constraint_for_component_ref): Likewise.
37359 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37360 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37361 * lra-constraints.c (process_alt_operands): Likewise.
37362 * ipa-inline.c (inline_small_functions): Likewise.
37363 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37364 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37365 * trans-mem.c (diagnose_tm_1_op): Likewise.
37366 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37367 (grid_parallel_clauses_gridifiable): Likewise.
37369 * config/nvptx/mkoffload.c (process): Add space in between
37372 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37373 "MOD4_SSE_REGS" and "ALL_REGS".
37375 * spellcheck.c (test_data): Add , in between "foo" and "food".
37377 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37380 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37381 boundary crossing check and subsequent code generation agree.
37383 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37385 * config/aarch64/aarch64.c (has_memory_op): Delete.
37386 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37389 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37391 PR rtl-optimization/79388
37392 PR rtl-optimization/79450
37393 * combine.c (distribute_notes): When removing TEM_INSN for which
37394 corresponding dest has last value recorded, invalidate that last
37397 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37399 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37400 of explicit '@'. Add missing assembly comment marker on branch costs
37403 2017-02-13 Nathan Sidwell <nathan@acm.org>
37405 * gengtype-lex.l (<in_struct>): Add '/'.
37407 2017-02-13 Martin Liska <mliska@suse.cz>
37410 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37412 2017-02-13 Richard Biener <rguenther@suse.de>
37414 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37416 * configure: Re-generate.
37417 * config.in: Likewise.
37418 * graphite-dependences.c: Simplify as if
37419 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37420 * graphite-isl-ast-to-gimple.c: Likewise.
37421 * graphite-optimize-isl.c: Likewise.
37422 * graphite-poly.c: Likewise.
37423 * graphite-sese-to-poly.c: Likewise.
37424 * graphite.h: Likewise.
37425 * toplev.c: Include isl/version.h and use isl_version () for
37426 printing the ISL version.
37427 * doc/install.texi: Update ISL requirement.
37429 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37431 * doc/standards.texi (Standards): Update reference to
37434 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37436 * doc/extend.texi (Named Address Spaces): sourceware.org now
37438 * doc/install.texi (Binaries): Ditto.
37441 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37443 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37444 standard terminology "stringize"/"stringizing" throughout.
37445 * doc/cppinternals.texi: Likewise.
37447 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37449 * doc/extend.texi: Fix some spelling mistakes and typos.
37450 * doc/invoke.texi: Likewise.
37452 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37455 * params.def (inline-min-speedup) Change from 10 to 8.
37457 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37459 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37462 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37465 * ipa-inline-analysis.c (get_minimal_bb): New function.
37466 (record_modified): Use it.
37467 (remap_edge_change_prob): Handle also ancestor functions.
37469 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37471 * doc/contrib.texi (Contributors): Remove broken link into
37472 the Mauve CVS repository.
37474 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37476 PR middle-end/79454
37477 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37478 result computation whenever lhs doesn't have vector mode, not
37479 just when it has BLKmode.
37481 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37483 * doc/makefile.texi (profiledbootstrap): Refer to the
37484 installation instructions only in textual form.
37486 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37489 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37491 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37493 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37494 (Specific): Update mingw-w64 reference.
37496 (Specific): Remove broken link to Renesas RX processor.
37498 2017-02-10 Richard Biener <rguenther@suse.de>
37500 * toplev.c (process_options): Do not mention obsolete graphite
37501 options when printing sorry message about missing graphite support.
37502 Mention -floop-nest-optimize.
37504 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37506 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37507 (vtst_p16): Likewise.
37508 (vtstq_p8): Likewise.
37509 (vtstq_p16): Likewise.
37511 (vtstq_p64): Likewise.
37512 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37513 (vset_lane_p64): New.
37514 (vsetq_lane_p64): New.
37516 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37518 PR tree-optimization/79411
37519 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37520 stmt operands are SSA_NAMEs used in abnormal phis.
37521 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37524 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37527 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37530 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37532 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37534 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37536 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37540 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37541 not arbitrary TREE_CONSTANT.
37544 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37545 "omp declare target link" attribute unless is_global_var.
37546 * omp-offload.c (find_link_var_op): Likewise.
37548 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37549 Chung-Lin Tang <cltang@codesourcery.com>
37551 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37553 (gimplify_adjust_omp_clauses): Don't delete TILE.
37554 (gimplify_omp_for): Deal with TILE.
37555 * internal-fn.c (expand_GOACC_TILE): New function.
37556 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37558 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37559 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37561 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37562 avoid DIV for outermost collapse var.
37563 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37564 Remove out of date comments, fix whitespace.
37565 * omp-general.c (omp_extract_for_data): Deal with tiling.
37566 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37567 adjust OLF_DIM_BASE value.
37568 (struct omp_for_data): Add tiling field.
37569 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37570 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37571 for auto loops. Remove default auto determining, moved to
37572 oacc_loop_fixed_partitions.
37573 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37574 stmts, add e_mask field.
37575 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37576 (oacc_thread_numbers): Use oacc_dim_call.
37577 (oacc_xform_tile): New.
37578 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37579 (finish_oacc_loop): Adjust for ifns vector.
37580 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37581 add case for GOACC_TILE fns.
37582 (oacc_loop_xform_loop): Delete.
37583 (oacc_loop_process): Iterate over call list directly, and add
37584 handling for GOACC_TILE fns.
37585 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37587 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37588 vector partitioning to outer loops. Assign 2 partitions to loops
37589 when available. Add TILE handling.
37590 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37591 (execite_oacc_device_lower): Process GOACC_TILE fns,
37592 ignore unknown specs.
37593 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37594 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37595 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37597 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37599 * configure.ac (ACX_BUGURL): Update.
37600 * configure: Regenerate.
37602 2017-02-09 Richard Biener <rguenther@suse.de>
37604 PR tree-optimization/69823
37605 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37606 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37608 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37610 * config/arc/arc-c.def: Add __NPS400__ definition.
37611 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37612 (TARGET_NPS400): Define.
37614 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37616 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37618 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37619 pointer, arch_info.
37620 (arc_cpu_types): Fill the arch_info field with a pointer into the
37621 arc_arch_types table.
37622 (arc_selected_cpu): Declare.
37623 * config/arc/arc.c (arc_selected_cpu): Make global.
37624 (arc_selected_arch): Delete.
37625 (arc_base_cpu): Delete.
37626 (arc_override_options): Remove references to deleted variables,
37627 update access to arch information.
37628 (ARC_OPT): Update access to arch information.
37629 (ARC_OPTX): Likewise.
37630 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37631 (TARGET_ARC600): Update access to arch information.
37632 (TARGET_ARC601): Likewise.
37633 (TARGET_ARC700): Likewise.
37634 (TARGET_EM): Likewise.
37635 (TARGET_HS): Likewise.
37636 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37639 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37642 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37643 condition/operands for integer GE/LE/GEU/LEU operations.
37645 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37647 PR translation/79397
37648 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37651 2017-02-08 Martin Jambor <mjambor@suse.cz>
37654 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37655 whether allocation happened.
37656 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37657 nothing was allocated.
37659 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37661 PR tree-optimization/79408
37662 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37663 constant, but SSA_NAME with a known integer range, use the minimum
37664 of that range instead of op1 to determine if modulo can be replaced
37665 with its first operand.
37667 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37669 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37671 2017-02-08 Richard Biener <rguenther@suse.de>
37673 PR tree-optimization/71824
37674 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37675 Check all loops contained in the merged region.
37677 2017-02-07 Andrew Pinski <apinski@cavium.com>
37679 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37681 2017-02-07 Andrew Pinski <apinski@cavium.com>
37683 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37684 (thunderxt88): Likewise.
37685 (thunderxt81): Disable LSE and change v8.1 to v8.
37686 (thunderxt83): Likewise.
37688 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37689 Richard Biener <rguenther@suse.de>
37691 PR middle-end/79399
37692 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37693 type from int to size_t.
37694 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37696 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37698 PR rtl-optimization/79386
37699 * cprop.c (bypass_conditional_jumps): Initialize
37700 bypass_last_basic_block already before splitting bbs after
37701 unconditional traps...
37702 (bypass_conditional_jumps): ... rather than here.
37705 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37706 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37707 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37708 fix -masm=intel patterns.
37710 2017-02-07 Richard Biener <rguenther@suse.de>
37712 PR tree-optimization/79256
37713 PR middle-end/79278
37714 * builtins.c (get_object_alignment_2): Use min_align_of_type
37715 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37716 and ADJUST_FIELD_ALIGN.
37718 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37720 * doc/tm.texi: Regenerate.
37721 * stor-layout.c (layout_decl): Adjust.
37722 (update_alignment_for_field): Likewise.
37723 (place_field): Likewise.
37724 (min_align_of_type): Likewise.
37725 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37726 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37727 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37728 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37729 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37730 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37731 * config/i386/i386.c (x86_field_alignment): Likewise.
37732 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37733 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37734 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37735 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37736 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37737 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37741 2017-01-30 Richard Biener <rguenther@suse.de>
37743 PR tree-optimization/79256
37744 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37745 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37748 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37750 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37751 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37752 builtins to SImode and emit a zero-extend, if necessary.
37754 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37756 * docs/invoke.texi (RISC-V Options): Alphabetize.
37758 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37760 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37763 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37765 * config/riscv/riscv.c: New file.
37766 * common/config/riscv/riscv-common.c: Likewise.
37767 * config.gcc: Likewise.
37768 * config/riscv/constraints.md: Likewise.
37769 * config/riscv/elf.h: Likewise.
37770 * config/riscv/generic.md: Likewise.
37771 * config/riscv/linux.h: Likewise.
37772 * config/riscv/multilib-generator: Likewise.
37773 * config/riscv/peephole.md: Likewise.
37774 * config/riscv/pic.md: Likewise.
37775 * config/riscv/predicates.md: Likewise.
37776 * config/riscv/riscv-builtins.c: Likewise.
37777 * config/riscv/riscv-c.c: Likewise.
37778 * config/riscv/riscv-ftypes.def: Likewise.
37779 * config/riscv/riscv-modes.def: Likewise.
37780 * config/riscv/riscv-opts.h: Likewise.
37781 * config/riscv/riscv-protos.h: Likewise.
37782 * config/riscv/riscv.h: Likewise.
37783 * config/riscv/riscv.md: Likewise.
37784 * config/riscv/riscv.opt: Likewise.
37785 * config/riscv/sync.md: Likewise.
37786 * config/riscv/t-elf-multilib: Likewise.
37787 * config/riscv/t-linux: Likewise.
37788 * config/riscv/t-linux-multilib: Likewise.
37789 * config/riscv/t-riscv: Likewise.
37790 * configure.ac: Likewise.
37791 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37792 Waterman as RISC-V maintainers.
37793 * doc/install.texi: Add RISC-V entries.
37794 * doc/invoke.texi: Add RISC-V options section.
37795 * doc/md.texi: Add RISC-V constraints section.
37796 * configure: Regenerated.
37798 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37801 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37802 false values to be constant vectors with all 0 or all 1 bits set.
37803 (vcondu<mode><mode>): Likewise.
37804 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37806 (fpmask_comparison_operator): Update comment.
37807 (vecint_comparison_operator): New predicate.
37808 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37809 vector conditionals when the true and false values are constant
37810 vectors with all 0 bits or all 1 bits set.
37812 2017-02-06 Martin Sebor <msebor@redhat.com>
37814 PR tree-optimization/79376
37815 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37817 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37819 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37820 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37821 to simplify split condition.
37823 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37825 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37826 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37829 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37831 PR rtl-optimization/68664
37832 * target.def (can_speculate_insn): New hook.
37833 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37834 * doc/tm.texi: Regenerate.
37835 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37836 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37837 (rs6000_sched_can_speculate_insn): New function.
37839 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37841 PR tree-optimization/79284
37842 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37843 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37844 vectorizable_mask_load_store, vectorizable_operation,
37845 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37846 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37847 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37848 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37849 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37850 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37851 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37852 is_gimple_assign (stmt). Replace another such test with
37853 is_gimple_assign (stmt).
37855 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37858 * config/avr/avr.c (rtl-iter.h): Include it.
37859 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37860 (avr_legitimate_combined_insn): ...and implementation.
37862 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37864 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37865 * config/s390/s390.c (s390_const_operand_ok)
37866 (s390_canonicalize_comparison, s390_extract_part)
37867 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37868 (s390_contiguous_bitmask_p, s390_rtx_costs)
37869 (legitimize_pic_address): Likewise.
37870 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37871 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37872 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37873 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37874 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37876 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37878 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37879 REGNO($0) == REGNO($1).
37881 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37883 * config/s390/linux.h(SIZE_TYPE): Add comment.
37885 2017-02-06 Julian Brown <julian@codesourcery.com>
37886 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37887 Virendra Pathak <virendra.pathak@broadcom.com>
37889 * config/aarch64/aarch64-cores.def: Change the scheduler
37891 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37892 * config/aarch64/thunderx2t99.md: New file.
37894 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37896 * doc/standards.texi (Go Language): Update link to language
37899 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37901 * tree-eh.c (lower_resx): Sanitize profile.
37902 (cleanup_empty_eh_move_lp): Likewise.
37904 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37907 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37909 * cfgloopmanip.h (loop_version): Update prototype.
37910 * modulo-sched.c (sms_schedule): Update call of loop_version.
37911 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37912 * tree-parloops.c (gen_parallel_loop): Likewise.
37913 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37914 * tree-ssa-loop-split.c (split_loop): Likewise.
37915 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37916 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37918 2017-02-05 Martin Liska <mliska@suse.cz>
37921 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37923 (print_operand_address): Initialize a struct to zero.
37925 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37927 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37928 garbage collector only in textual form.
37930 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37932 * doc/extend.texi (x86 specific memory model extensions for
37933 transactional memory): Simplify a phrase.
37935 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37938 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37939 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37940 (atomic_storedi_1): Likewise.
37942 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37944 PR tree-optimization/79338
37945 * tree-parloops.c (gather_scalar_reductions): Don't call
37946 vect_analyze_loop_form for loop->inner before destroying loop's
37949 2017-02-03 Martin Sebor <msebor@redhat.com>
37951 PR tree-optimization/79327
37952 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37953 when precision has resulted in leading zeros.
37954 (format_integer): Adjust the likely counter to assume an unknown
37955 argument that may be zero is non-zero.
37957 2017-02-03 Jason Merrill <jason@redhat.com>
37960 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37961 avoid copying non-taken branch.
37963 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37965 PR tree-optimization/79340
37966 * tree-vect-loop.c (vectorizable_reduction): Release
37967 vec_defs elements after safe_splicing them into other vectors.
37970 PR tree-optimization/79327
37971 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37972 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37974 (format_integer): Use wide_int_to_tree instead of build_int_cst
37975 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37976 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37977 of shortest and longest sequence.
37979 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37981 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37982 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37984 2017-02-03 Walter Lee <walt@tilera.com>
37987 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37988 after initial stackframe link reg save.
37989 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37991 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37994 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37995 wu for stxssp alternative.
37997 2017-02-03 Martin Sebor <msebor@redhat.com>
37999 PR tree-optimization/79352
38000 * gimple-fold.c (get_range_strlen): Add argument.
38001 (get_range_strlen): Change return type to bool.
38002 (get_maxval_strlen): Pass in a dummy argument.
38003 * gimple-fold.h (get_range_strlen): Change return type to bool.
38004 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38005 * tree.h (array_at_struct_end_p): Add argument.
38006 * tree.c (array_at_struct_end_p): Handle it.
38008 2017-02-03 Martin Liska <mliska@suse.cz>
38011 * multiple_target.c (create_dispatcher_calls): Redirect edge
38012 from a caller of a dispatcher.
38013 (expand_target_clones): Make the clones local.
38014 (ipa_target_clone): Do both target clones and resolvers.
38015 (ipa_dispatcher_calls): Remove the pass.
38016 (pass_dispatcher_calls::gate): Likewise.
38017 (make_pass_dispatcher_calls): Likewise.
38018 * passes.def (pass_target_clone): Put as very first IPA early
38021 2017-02-03 Martin Liska <mliska@suse.cz>
38023 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38024 in case of a function with ifunc attribute.
38026 2017-02-03 Martin Liska <mliska@suse.cz>
38028 * cgraph.c (cgraph_node::dump): Dump function version info.
38029 * symtab.c (symtab_node::dump_base): Add missing new line.
38031 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38033 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38034 (ifcombine_ifandif): Use it.
38036 2017-02-03 Martin Liska <mliska@suse.cz>
38038 * doc/invoke.texi: Document default value for
38039 use-after-scope-direct-emission-threshold.
38041 2017-02-03 Martin Liska <mliska@suse.cz>
38043 PR tree-optimization/79339
38044 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38045 (format_floating): Likewise.
38047 2017-02-03 Martin Liska <mliska@suse.cz>
38050 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38052 (ipa_node_params_t::remove): Likewise.
38053 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38054 initialization from removed ipa_node_params_t::insert.
38055 (ipa_node_params::~ipa_node_params): Move from removed
38056 ipa_node_params_t::release.
38057 * symbol-summary.h (symbol_summary::m_released): New member.
38058 Do not release a summary twice. Do not allow to call finalizer
38059 for types of a summary that live in GGC memory.
38061 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38063 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38066 2017-02-02 Martin Sebor <msebor@redhat.com>
38068 PR middle-end/79275
38069 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38070 (format_string): Tighten up the range of output for non-constant
38071 strings and correct the expected range for wide non-constant strings.
38073 2017-02-02 Martin Sebor <msebor@redhat.com>
38075 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38077 PR middle-end/32003
38078 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38080 (-fdump-tree-@var): Add to index and document how to come up
38081 with pass-specific option and dump file names.
38082 (-fdump-passes): Clarify where to look for output.
38084 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38086 PR middle-end/77445
38087 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38088 statistics of the analyzed path; allow threading for speed when
38089 any of BBs along the path are optimized for speed.
38091 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38093 PR middle-end/78468
38094 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38095 settings of the virtual registers.
38098 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38100 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38101 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38104 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38106 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38107 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38109 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38111 * config/s390/s390.md: Add missing comments with the expanded
38113 * config/s390/vector.md: Likewise.
38114 * config/s390/vx-builtins.md: Likewise.
38116 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38119 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38120 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38121 conditions on a single line.
38123 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38125 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38126 __S390_VX__ to __VX__.
38128 2017-02-01 Andrew Pinski <apinski@cavium.com>
38130 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38131 stmt_info to record_stmt_cost.
38132 (vect_get_known_peeling_cost): Pass stmt_info if known to
38134 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38135 cpu_vector_cost field into
38136 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38137 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38138 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38139 splitting of scalar_stmt_cost and vec_stmt_cost.
38140 (thunderx_vector_cost): Likewise.
38141 (cortexa57_vector_cost): LIkewise.
38142 (exynosm1_vector_cost): Likewise.
38143 (xgene1_vector_cost): Likewise.
38144 (thunderx2t99_vector_cost): Improve after the splitting of the two
38146 (aarch64_builtin_vectorization_cost): Update for the splitting of
38147 scalar_stmt_cost and vec_stmt_cost.
38149 2017-02-01 Torvald Riegel <triegel@redhat.com>
38150 Richard Henderson <rth@redhat.com>
38152 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38153 conditional on existance of a fast atomic load.
38154 * optabs-query.c (can_atomic_load_p): New function.
38155 * optabs-query.h (can_atomic_load_p): Declare it.
38156 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38157 no fast atomic load is available for the particular size of access.
38158 (expand_atomic_compare_and_swap): Likewise.
38159 (expand_atomic_load): Likewise.
38160 (expand_atomic_store): Likewise.
38161 (expand_atomic_fetch_op): Likewise.
38162 * testsuite/lib/target-supports.exp
38163 (check_effective_target_sync_int_128): Remove x86 because it provides
38164 no fast atomic load.
38165 (check_effective_target_sync_int_128_runtime): Likewise.
38167 2017-02-01 Richard Biener <rguenther@suse.de>
38169 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38170 (graphite_transform_loops): Provide opt-info for optimized nests.
38171 * tree-parloop.c (parallelize_loops): Provide opt-info for
38172 parallelized loops.
38174 2017-02-01 Richard Biener <rguenther@suse.de>
38176 PR middle-end/79315
38177 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38178 was not set before.
38180 2017-02-01 Richard Biener <rguenther@suse.de>
38182 PR tree-optimization/71824
38183 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38184 Verify the loops are valid in the merged SESE region.
38185 (scop_detection::can_represent_loop_1): Check analyzing the
38186 evolution of the number of iterations in the region succeeds.
38188 2017-01-31 Ian Lance Taylor <iant@golang.org>
38190 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38191 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38193 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38195 PR preprocessor/79210
38196 * input.c (get_substring_ranges_for_loc): Replace line_width
38197 assertion with error-handling.
38199 2017-01-31 Richard Biener <rguenther@suse.de>
38201 PR tree-optimization/77318
38202 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38203 (create_pw_aff_from_tree): Take loop parameter.
38204 (add_condition_to_pbb): Pass loop of the condition to
38205 create_pw_aff_from_tree.
38207 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38209 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38210 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38212 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38216 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38218 (convert_int_to_float128): Likewise.
38219 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38220 (convert_int_to_float128): Likewise.
38221 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38222 (UNSPEC_IEEE128_CONVERT): Likewise.
38223 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38224 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38225 Use local variables for IBM extended format.
38226 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38227 (fix_trunc<mode>si2_fprs): Likewise.
38228 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38229 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38230 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38231 to know that we can now have integers of all sizes in vector
38233 (fix<uns>_<mode>di2_hw): Likewise.
38234 (float<uns>_<mode>si2_hw): Likewise.
38235 (fix_<mode>si2_hw): Likewise.
38236 (fixuns_<mode>si2_hw): Likewise.
38237 (float<uns>_<mode>di2_hw): Likewise.
38238 (float_<mode>di2_hw): Likewise.
38239 (float_<mode>si2_hw): Likewise.
38240 (floatuns_<mode>di2_hw): Likewise.
38241 (floatuns_<mode>si2_hw): Likewise.
38242 (xscvqp<su>wz_<mode>): Delete, no longer used.
38243 (xscvqp<su>dz_<mode>): Likewise.
38244 (xscv<su>dqp_<mode>): Likewise.
38245 (ieee128_mfvsrd_64bit): Likewise.
38246 (ieee128_mfvsrd_32bit): Likewise.
38247 (ieee128_mfvsrwz): Likewise.
38248 (ieee128_mtvsrw): Likewise.
38249 (ieee128_mtvsrd_64bit): Likewise.
38250 (ieee128_mtvsrd_32bit): Likewise.
38252 2017-01-31 Martin Liska <mliska@suse.cz>
38255 * ipa-prop.c (ipa_free_all_node_params): Call release method
38256 instead of ~sumbol_summary to not to trigger double times
38259 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38261 PR tree-optimization/71691
38262 * bitmap.h (class auto_bitmap): New.
38263 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38264 is_maybe_undefined instead of ssa_undefined_value_p.
38266 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38268 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38269 __S390_ARCH_LEVEL__ to __ARCH__.
38271 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38273 PR tree-optimization/79267
38274 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38275 if should_remove_lhs_p is true.
38277 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38280 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38281 (add_alignment_attribute): New.
38282 (base_type_die): Add alignment attribute.
38283 (subrange_type_die): Likewise.
38284 (modified_type_die): Likewise.
38285 (gen_array_type_die): Likewise.
38286 (gen_descr_array_type_die: Likewise.
38287 (gen_enumeration_type_die): Likewise.
38288 (gen_subprogram_die): Likewise.
38289 (gen_variable_die): Likewise.
38290 (gen_field_die): Likewise.
38291 (gen_ptr_to_mbr_type_die): Likewise.
38292 (gen_struct_or_union_type_die): Likewise.
38293 (gen_subroutine_type_die): Likewise.
38294 (gen_typedef_die): Likewise.
38295 (base_type_cmp): Compare alignment attribute.
38297 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38300 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38301 (setb_unsigned) New pattern for setb with CCUNS.
38302 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38303 subfc./subfe sequence to avoid overflow problems. Generate a
38304 shorter sequence with cmpld/setb for power9.
38305 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38306 for generating subfc. instruction.
38307 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38308 now uses this instruction.
38310 2017-01-30 Ian Lance Taylor <iant@google.com>
38313 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38314 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38316 2017-01-30 Martin Sebor <msebor@redhat.com>
38318 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38319 Move constant to the right of a relational operator.
38320 (get_mpfr_format_length, format_character, format_string): Ditto.
38321 (should_warn_p, maybe_warn): Same.
38323 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38325 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38328 * asan.c (get_translation_unit_decl): Remove function.
38329 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38331 2017-01-30 Martin Liska <mliska@suse.cz>
38333 PR gcov-profile/79259
38334 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38335 -fprofile-generate.
38337 2017-01-30 Martin Liska <mliska@suse.cz>
38340 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38341 Initialize variables with NULL value.
38343 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38346 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38348 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38350 2017-01-30 Richard Biener <rguenther@suse.de>
38352 PR tree-optimization/79276
38353 * tree-vrp.c (process_assert_insertions): Properly adjust common
38354 when removing a duplicate.
38356 2017-01-30 Richard Biener <rguenther@suse.de>
38358 PR tree-optimization/79256
38359 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38360 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38362 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38364 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38367 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38368 ("*r<noxa>sbg_<mode>_sll_bitmask")
38369 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38370 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38371 Use contiguous_bitmask_nowrap_operand.
38373 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38376 * config/rs6000/altivec.h (vec_xl): Revise #define.
38377 (vec_xst): Likewise.
38379 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38381 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38383 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38385 PR rtl-optimization/79194
38386 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38387 traps before call to bypass_conditional_jumps.
38389 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38391 PR tree-optimization/71374
38392 * lra-constraints.c (check_conflict_input_operands): New.
38393 (match_reload): Use it.
38395 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38398 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38399 account to calculate conflict_set.
38401 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38403 PR rtl-optimization/78559
38404 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38405 other_insn in combine.
38407 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38409 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38410 uint16_type_node for BT_UINT16.
38412 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38414 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38415 "RTL Tests" to menu.
38416 (GIMPLE Tests): New node.
38417 (RTL Tests): New node.
38419 2017-01-27 Richard Biener <rguenther@suse.de>
38421 PR tree-optimization/79245
38422 * tree-loop-distribution.c (distribute_loop): Apply cost
38423 modeling also to detected patterns.
38425 2017-01-27 Richard Biener <rguenther@suse.de>
38427 PR tree-optimization/71433
38428 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38429 (compare_assert_loc): New function.
38430 (process_assert_insertions): Sort and optimize assert locations
38431 to remove duplicates and push down identical assertions on
38432 edges to their destination block.
38434 2017-01-27 Richard Biener <rguenther@suse.de>
38436 PR tree-optimization/79244
38437 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38438 out SSA names even if abnormal.
38440 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38442 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38443 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38444 instead of MPFR_RNDN.
38446 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38449 * arm.c (arm_option_override): Don't call build_target_option_node
38450 until after doing all option overrides.
38451 (arm_valid_target_attribute_tree): Likewise.
38453 2017-01-27 Martin Liska <mliska@suse.cz>
38455 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38456 for {cd}tors and C++ {cd}tors.
38458 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38460 * config/s390/s390.md ("*setmem_long_and")
38461 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38463 2017-01-26 Martin Sebor <msebor@redhat.com>
38465 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38468 2017-01-26 Martin Sebor <msebor@redhat.com>
38470 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38471 HAVE_DFmode before using XFmode or DFmode.
38472 (parse_directive): Avoid using the z length modifier to avoid
38473 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38475 PR middle-end/78703
38476 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38477 to accept adjustment as an array.
38478 (get_int_range): New function.
38479 (struct directive): Make width and prec arrays.
38480 (directive::set_width, directive::set_precision): Call get_int_range.
38481 (format_integer, format_floating): Handle width and precision ranges.
38482 (format_string, parse_directive): Same.
38484 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38487 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38488 comdat_type_p set, just clone them, but keep the children in the
38492 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38493 which have direct callers with -fvar-tracking-assignments enabled
38495 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38496 inside of type units.
38498 2017-01-26 Martin Sebor <msebor@redhat.com>
38500 PR middle-end/78703
38501 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38503 (struct format_result): Replace number_chars, number_chars_min,
38504 and number_chars_max with a single member of struct result_range.
38506 (format_result::operator+=): Adjust.
38507 (struct fmtresult): Remove bounded. Handle likely and unlikely
38509 (fmtresult::adjust_for_width_or_precision): New function.
38510 (fmtresult:type_max_digits): New function.
38511 (bytes_remaining): Handle likely and unlikely counters.
38512 (min_bytes_remaining): Remove.
38513 (format_percent): Simplify.
38514 (format_integer, format_floating): Set likely and unlikely counters.
38515 (get_string_length, format_character, format_string): Same.
38516 (format_plain, should_warn_p): New function.
38517 (maybe_warn): Call should_warn_p. Update diagnostic messages
38518 and handle those for all directives, including plain strings.
38519 (format_directive): Handle likely and unlikely counters.
38520 Remove unnecessary quoting from diagnostics. Add an informational
38522 (add_bytes): Remove.
38523 (pass_sprintf_length::compute_format_length): Simplify.
38524 (try_substitute_return_value): Handle likely and unlikely counters.
38526 2017-01-26 Carl Love <cel@us.ibm.com>
38528 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38529 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38531 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38534 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38535 endianess for subregs into account.
38536 * lra-constraints.c (lra_constraints): Do risky transformations
38537 always on the first iteration.
38538 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38539 last_call_used_reg_set.
38540 (process_bb_lives): Define and use last_call_used_reg_set.
38541 * lra.c (lra): Always continue after lra_constraints on the first
38544 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38546 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38548 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38550 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38552 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38553 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38554 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38555 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38556 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38557 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38558 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38559 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38560 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38562 2017-01-26 Marek Polacek <polacek@redhat.com>
38565 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38566 for the third operand.
38568 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38570 PR middle-end/79236
38571 * omp-low.c (struct omp_context): Add simt_stmt field.
38572 (scan_omp_for): Return omp_context *.
38573 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38574 context to the _simt_ SIMD stmt.
38575 (lower_omp_for): For combined SIMD with sibling _simt_
38576 SIMD, make sure to use the same decls in _looptemp_
38577 clauses as in the sibling.
38579 2017-01-26 David Sherwood <david.sherwood@arm.com>
38581 PR middle-end/79212
38582 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38585 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38588 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38589 emit fld b; fld a; if possible.
38591 * brig-builtins.def: Update copyright years.
38592 * config/arm/arm_acle_builtins.def: Update copyright years.
38594 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38597 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38598 constraint instead of o for the stxsd instruction.
38600 2017-01-25 Carl Love <cel@us.ibm.com>
38602 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38603 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38605 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38607 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38609 2017-01-25 Richard Biener <rguenther@suse.de>
38611 PR tree-optimization/69264
38612 * target.def (vector_alignment_reachable): Improve documentation.
38613 * doc/tm.texi: Regenerate.
38614 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38616 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38617 earlier changes with respect to TYPE_USER_ALIGN.
38618 (vector_alignment_reachable_p): Likewise. Improve dumping.
38620 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38623 * config/arm/arm.md (xordi3): Force constant operand into a register
38626 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38628 * doc/invoke.texi (-fstore-merging): Correct default optimization
38629 levels at which it is enabled.
38630 (-O): Move -fstore-merging from list to...
38633 2017-01-25 Richard Biener <rguenther@suse.de>
38636 * omp-expand.c: Include debug.h.
38637 (expand_omp_taskreg): Make sure to generate early debug before
38638 outlining anything from a function.
38639 (expand_omp_target): Likewise.
38640 (grid_expand_target_grid_body): Likewise.
38642 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38645 * asan.c (get_translation_unit_decl): New function.
38646 (asan_add_global): Extract modules file name from globals
38647 TRANSLATION_UNIT_DECL name.
38649 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38652 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38653 for long calls with APCS frame and VFP.
38655 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38657 * cfg.c (original_copy_tables_initialized_p): New function.
38658 * cfg.h (original_copy_tables_initialized_p): New decl.
38659 * cfgrtl.c (relink_block_chain): Guard the call to
38660 free_original_copy_tables with a call to
38661 original_copy_tables_initialized_p.
38662 * cgraph.h (symtab_node::native_rtl_p): New decl.
38663 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38664 (symtab_node::needed_p): Don't assert for early assembly output
38665 for __RTL functions.
38666 (cgraph_node::finalize_function): Set "force_output" for __RTL
38668 (cgraph_node::analyze): Bail out early for __RTL functions.
38669 (analyze_functions): Update assertion to support __RTL functions.
38670 (cgraph_node::expand): Bail out early for __RTL functions.
38671 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38673 * function.h (struct function): Update comment for field
38675 * gimple-expr.c: Include "tree-pass.h".
38676 (gimple_has_body_p): Return false for __RTL functions.
38677 * Makefile.in (OBJS): Add run-rtl-passes.o.
38678 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38680 (gcc::pass_manager::get_clean_slate): New accessor.
38681 * passes.c: Include "insn-addr.h".
38682 (should_skip_pass_p): Add logging. Update logic for running
38683 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38684 property-provider override so it is only done for gimple passes.
38686 (skip_pass): New function.
38687 (execute_one_pass): Call skip_pass when skipping passes.
38688 * read-md.c (md_reader::read_char): Support filtering
38689 the input to a subset of line numbers.
38690 (md_reader::md_reader): Initialize fields
38691 m_first_line and m_last_line.
38692 (md_reader::read_file_fragment): New function.
38693 * read-md.h (md_reader::read_file_fragment): New decl.
38694 (md_reader::m_first_line): New field.
38695 (md_reader::m_last_line): New field.
38696 * read-rtl-function.c (function_reader::create_function): Only
38697 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38698 curr_properties. Set DECL_INITIAL to a dummy block.
38699 (read_rtl_function_body_from_file_range): New function.
38700 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38702 * run-rtl-passes.c: New file.
38703 * run-rtl-passes.h: New file.
38705 2017-01-24 Jeff Law <law@redhat.com>
38707 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38710 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38712 PR tree-optimization/79159
38713 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38714 (record_nonwrapping_iv): Improve boundary using above function if no
38715 value range information.
38717 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38718 Martin Jambor <mjambor@suse.cz>
38720 * brig-builtins.def: New file.
38721 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38722 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38723 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38724 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38725 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38726 * builtin-types.def (BT_INT8): New.
38727 (BT_INT16): Likewise.
38728 (BT_UINT8): Likewise.
38729 (BT_UINT16): Likewise.
38730 (BT_FN_ULONG): Likewise.
38731 (BT_FN_UINT_INT): Likewise.
38732 (BT_FN_UINT_ULONG): Likewise.
38733 (BT_FN_UINT_LONG): Likewise.
38734 (BT_FN_UINT_PTR): Likewise.
38735 (BT_FN_ULONG_PTR): Likewise.
38736 (BT_FN_INT8_FLOAT): Likewise.
38737 (BT_FN_INT16_FLOAT): Likewise.
38738 (BT_FN_UINT32_FLOAT): Likewise.
38739 (BT_FN_UINT16_FLOAT): Likewise.
38740 (BT_FN_UINT8_FLOAT): Likewise.
38741 (BT_FN_UINT64_FLOAT): Likewise.
38742 (BT_FN_UINT16_UINT32): Likewise.
38743 (BT_FN_UINT32_UINT16): Likewise.
38744 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38745 (BT_FN_INT_PTR_INT): Likewise.
38746 (BT_FN_UINT_PTR_UINT): Likewise.
38747 (BT_FN_LONG_PTR_LONG): Likewise.
38748 (BT_FN_ULONG_PTR_ULONG): Likewise.
38749 (BT_FN_VOID_UINT64_UINT64): Likewise.
38750 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38751 (BT_FN_INT8_INT8_INT8): Likewise.
38752 (BT_FN_INT16_INT16_INT16): Likewise.
38753 (BT_FN_INT_INT_INT): Likewise.
38754 (BT_FN_UINT_FLOAT_UINT): Likewise.
38755 (BT_FN_FLOAT_UINT_UINT): Likewise.
38756 (BT_FN_ULONG_UINT_UINT): Likewise.
38757 (BT_FN_ULONG_UINT_PTR): Likewise.
38758 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38759 (BT_FN_UINT_UINT_UINT): Likewise.
38760 (BT_FN_VOID_UINT_PTR): Likewise.
38761 (BT_FN_UINT_UINT_PTR: Likewise.
38762 (BT_FN_UINT32_UINT64_PTR): Likewise.
38763 (BT_FN_INT_INT_UINT_UINT): Likewise.
38764 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38765 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38766 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38767 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38768 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38769 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38770 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38771 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38772 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38773 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38774 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38775 * doc/frontends.texi: List BRIG FE.
38776 * doc/install.texi (Testing): Add BRIG tesring requirements.
38777 * doc/invoke.texi (Overall Options): Mention BRIG.
38778 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38780 2017-01-24 Richard Biener <rguenther@suse.de>
38782 PR translation/79208
38783 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38785 2017-01-24 Martin Jambor <mjambor@suse.cz>
38788 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38789 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38790 and known_contexts.
38792 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38794 PR middle-end/79123
38795 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38796 casts from signed to unsigned really don't have a range.
38798 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38800 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38801 GMP_RNDx for compatiblity.
38803 2017-01-24 Martin Liska <mliska@suse.cz>
38806 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38807 that would prevent us to call alloca with -1 as argument.
38809 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38811 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38812 Avoid -Wformat-security warning.
38814 2017-01-23 Andrew Pinski <apinski@cavium.com>
38816 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38819 2017-01-23 Martin Sebor <msebor@redhat.com>
38821 PR middle-end/78703
38822 * gimple-ssa-sprintf.c (warn_level): New global.
38823 (format_integer): Use it here and throughout the rest of the file.
38824 Use the same switch to compute sign as base.
38825 (maybe_warn): New function.
38826 (format_directive): Factor out warnings into maybe_warn.
38827 Add debugging output. Use warn_level.
38828 (add_bytes): Use warn_level.
38829 (pass_sprintf_length::compute_format_length): Add debugging output.
38830 (try_substitute_return_value): Same.
38831 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38833 PR middle-end/78703
38834 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38835 (struct fmtresult, format_integer, format_floating): Adjust.
38836 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38837 (get_string_length, format_string,format_directive): Same.
38838 (pass_sprintf_length::compute_format_length): Same.
38839 (try_substitute_return_value): Simplify slightly.
38841 PR middle-end/78703
38842 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38843 (fmtresult::operator+=): Outlined.
38844 (struct fmtresult): Add ctors.
38845 (struct conversion_spec): Rename...
38846 (struct directive): ...to this. Add and remove data members.
38847 (directive::set_width, directive::set_precision): New functions.
38848 (format_percent): Use fmtresult ctor.
38849 (get_width_and_precision): Remove.
38850 (format_integer): Make naming changes. Avoid computing width and
38852 (format_floating): Same. Adjust indentation.
38853 (format_character, format_none): New functions.
38854 (format_string): Moved character handling to format_character.
38855 (format_directive): Remove arguments, change return type.
38856 (parse_directive): New function.
38857 (pass_sprintf_length::compute_format_length): Move directive
38858 parsing to parse_directive.
38860 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38862 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38863 (assign_assembler_name_if_needed): ... this.
38864 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38865 (assign_assembler_name_if_needed): ... this.
38866 (free_lang_data_in_cgraph): Adjust callers.
38867 * cgraphunit.c (cgraph_node::analyze): Likewise.
38868 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38871 2017-01-23 Richard Biener <rguenther@suse.de>
38873 PR tree-optimization/79088
38874 PR tree-optimization/79188
38875 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38876 resetting loop bounds after last path deletion. Reset loop
38877 bounds of the target loop, make code match the comments.
38878 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38879 Make sure loops need no fixups.
38881 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38883 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38884 exponent support with double type for first argument.
38885 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38886 type returned by __builtin_vec_extract_sig,
38887 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38888 functions from "vector int" to "vector unsigned int" or from
38889 "vector long long int" to "vector unsigned long long int".
38890 Changed type returned by __builtin_vec_extract_exp,
38891 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38892 functions from "vector int" to "vector unsigned int" or from
38893 "vector long long int" to "vector unsigned long long int".
38894 Changed return type of __builtin_vec_test_data_class,
38895 __builtin_vec_test_data_class_sp, and
38896 __builtin_vec_test_data_class_dp from "vector int" to
38897 "vector bool int" or from "vector long long int" to "vector bool
38898 long long int" and changed second argument type from "unsigned
38899 int" to "int". Added new overloaded function forms "vector float
38900 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38901 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38902 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38903 double, vector unsigned long long int)" and "vector double
38904 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38905 long int)". Changed return type of
38906 __builtin_scalar_test_data_class and
38907 __builtin_scalar_test_data_class_sp and
38908 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38909 int" and changed second argument from "unsigned int" to "int".
38910 Changed type returned by __builtin_scalar_test_neg,
38911 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38912 from "int" to "bool int". Added new overloaded function form
38913 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38914 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38915 exponent double-precision with floating point first argument.
38916 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38917 documentation of scalar_test_data_class, scalar_test_neg,
38918 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38919 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38920 vec_test_data_class built-in functions to reflect refinements in
38921 their type signatures.
38923 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38925 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38927 (aarch64_elf_asm_destructor): Likewise.
38929 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38931 PR rtl-optimization/78634
38932 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38933 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38934 * ifcvt.c (noce_try_cmove): Add missing cost check.
38936 PR rtl-optimization/71724
38937 * combine.c (if_then_else_cond): Look for situations where it is
38938 beneficial to undo the work of one of the recursive calls.
38940 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38942 PR tree-optimization/70754
38943 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38944 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38945 combined stmt before it if not NULL.
38946 (combine_chains): Process refs reversely and compute dominance point
38949 2017-01-23 Martin Liska <mliska@suse.cz>
38951 PR tree-optimization/79196
38952 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38953 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38955 (strlen_optimize_stmt): Call the renamed function.
38957 2017-01-23 Michael Matz <matz@suse.de>
38959 PR tree-optimization/78384
38960 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38962 2017-01-23 Richard Biener <rguenther@suse.de>
38964 PR tree-optimization/79186
38965 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38966 both incoming edges before moving an assert.
38968 2017-01-23 Martin Jambor <mjambor@suse.cz>
38970 * ipa-prop.c (load_from_param_1): Removed.
38971 (load_from_unmodified_param): Bits from load_from_param_1 put back
38973 (load_from_param): Removed.
38974 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38975 with stmt. Reverted back to use of load_from_unmodified_param.
38977 2017-01-23 Martin Jambor <mjambor@suse.cz>
38980 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38981 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38982 field a pointer to garbage collected vector, mark lattices and
38983 ipcp_orig_node with GTY((skip)).
38984 (ipa_get_param_count): Adjust to descriptors being a pointer.
38985 (ipa_get_param): Likewise.
38986 (ipa_get_type): Likewise.
38987 (ipa_get_param_move_cost): Likewise.
38988 (ipa_set_param_used): Likewise.
38989 (ipa_get_controlled_uses): Likewise.
38990 (ipa_set_controlled_uses): Likewise.
38991 (ipa_is_param_used): Likewise.
38992 (ipa_node_params_t): Move into garbage collector. New methods insert
38994 (ipa_node_params_sum): Annotate wth GTY(()).
38995 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38997 (ipa_load_from_parm_agg): Adjust declaration.
38998 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38999 * ipa-profile.c (ipa_profile): Likewise.
39000 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39001 (ipa_populate_param_decls): Make descriptors parameter garbage
39003 (ipa_dump_param): Adjust to descriptors being a pointer.
39004 (ipa_alloc_node_params): Likewise.
39005 (ipa_initialize_node_params): Likewise.
39006 (load_from_param_1): Make descriptors parameter garbage collected.
39007 (load_from_unmodified_param): Likewise.
39008 (load_from_param): Likewise.
39009 (ipa_load_from_parm_agg): Likewise.
39010 (ipa_node_params::~ipa_node_params): Removed.
39011 (ipa_free_all_node_params): Remove call to delete operator.
39012 (ipa_node_params_t::insert): New.
39013 (ipa_node_params_t::remove): Likewise.
39014 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39015 copy known_csts and known_contexts vectors.
39016 (ipa_read_node_info): Adjust to descriptors being a pointer.
39017 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39019 (ipcp_transform_function): Make descriptors variable garbage
39022 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39024 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39025 * config/i386/avx512dqintrin.h: Ditto.
39026 * config/i386/avx512fintrin.h: Ditto.
39027 * config/i386/i386.c: Handle new builtins.
39028 * config/i386/i386-builtin.def: Add new builtins.
39029 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39030 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39032 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39033 Martin Liska <mliska@suse.cz>
39035 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39036 * asan.c (asan_expand_poison_ifn): Support stores and use
39037 appropriate ASAN report function.
39038 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39039 * internal-fn.def (ASAN_POISON_USE): Declare.
39040 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39041 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39042 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39043 ASAN_POISON calls w/o LHS.
39044 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39045 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39046 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39047 * gimplify.c (asan_poison_variables): Add attribute
39048 use_after_scope_memory to variables that really needs to live
39050 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39051 having the attribute.
39053 2017-01-23 Martin Liska <mliska@suse.cz>
39055 * asan.c (create_asan_shadow_var): New function.
39056 (asan_expand_poison_ifn): Likewise.
39057 * asan.h (asan_expand_poison_ifn): New declaration.
39058 * internal-fn.c (expand_ASAN_POISON): Likewise.
39059 * internal-fn.def (ASAN_POISON): New builtin.
39060 * sanopt.c (pass_sanopt::execute): Expand
39061 asan_expand_poison_ifn.
39062 * tree-inline.c (copy_decl_for_dup_finish): Make function
39064 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39065 * tree-ssa.c (is_asan_mark_p): New function.
39066 (execute_update_addresses_taken): Rewrite local variables
39067 (identified just by use-after-scope as addressable) into SSA.
39069 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39071 * doc/install.texi (Specific): opensource.apple.com uses https
39072 now. Remove trailing slash.
39074 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39076 * README.Portability: Remove note on an Irix compatibility issue.
39078 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39080 * gcov.c (INCLUDE_ALGORITHM): Define.
39081 (INCLUDE_VECTOR): Define.
39082 No longer include <vector> and <algorithm> directly.
39084 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39086 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39088 * doc/invoke.texi (Code Gen Options): Ditto.
39090 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39093 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39095 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39097 rtl-optimization/79125
39098 * cprop.c (local_cprop_pass): Handle cases where we make an
39099 unconditional trap.
39101 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39105 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39106 read from, for big endian.
39108 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39110 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39111 register pauth builtins for LP64 only.
39113 2017-01-20 Marek Polacek <polacek@redhat.com>
39116 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39119 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39121 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39123 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39124 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39125 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39127 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39130 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39131 in big-endian mode when they are not a single duplicated value.
39133 2017-01-20 Richard Biener <rguenther@suse.de>
39135 * BASE-VER: Bump to 7.0.1.
39137 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39139 * omp-low.c (omplow_simd_context): New struct. Use it...
39140 (lower_rec_simd_input_clauses): ...here and...
39141 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39142 references to idx, lane, max_vf, is_simt.
39144 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39146 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39149 2017-01-20 Martin Jambor <mjambor@suse.cz>
39151 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39152 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39154 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39155 (GTFILES): Rename hsa.c to hsa-common.c.
39156 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39157 * hsa-dump.c: Likewise.
39158 * hsa-gen.c: Likewise.
39159 * hsa-regalloc.c: Likewise.
39160 * ipa-hsa.c: Likewise.
39161 * omp-expand.c: Likewise.
39162 * omp-low.c: Likewise.
39163 * toplev.c: Likewise.
39165 2017-01-20 Marek Polacek <polacek@redhat.com>
39168 * doc/invoke.texi: Document -Wduplicated-branches.
39169 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39170 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39171 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39172 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39173 return 0 only when not OEP_LEXICOGRAPHIC.
39174 (fold_build_cleanup_point_expr): Use the expression
39175 location when building CLEANUP_POINT_EXPR.
39176 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39177 * tree.c (add_expr): Handle error_mark_node.
39179 2017-01-20 Martin Liska <mliska@suse.cz>
39182 * tree-profile.c (init_ic_make_global_vars): Do not call
39184 (gimple_init_gcov_profiler): Likewise.
39186 2017-01-20 Martin Liska <mliska@suse.cz>
39189 * cgraph.h (maybe_create_reference): Remove argument and
39191 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39193 * ipa-cp.c (create_specialized_node): Likewise.
39194 * symtab.c (symtab_node::maybe_create_reference): Handle
39195 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39197 2017-01-20 Martin Liska <mliska@suse.cz>
39199 * read-rtl-function.c (function_reader::create_function): Use
39200 build_decl instread of build_decl_stat.
39202 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39204 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39205 * config/i386/avx512dqintrin.h: Ditto.
39206 * config/i386/avx512fintrin.h: Ditto.
39207 * config/i386/i386-builtin-types.def: Add new types.
39208 * config/i386/i386.c: Handle new types.
39209 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39210 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39211 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39212 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39213 (__builtin_ia32_kshiftridi): New.
39214 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39216 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39220 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39221 define to rs6000_init_stack_protect_guard.
39222 (rs6000_init_stack_protect_guard): New function.
39224 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39225 Yunqiang Su <yunqiang.su@imgtec.com>
39227 * config.gcc (supported_defaults): Add madd4.
39228 (with_madd4): Add validation.
39229 (all_defaults): Add madd4.
39230 * config/mips/mips.opt (mmadd4): New option.
39231 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39233 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39235 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39236 (ISA_HAS_FUSED_MADD4): Likewise.
39237 * doc/invoke.texi (-mmadd4): Document the new option.
39238 * doc/install.texi (--with-madd4): Document the new option.
39240 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39242 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39243 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39244 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39245 (aarch64_init_pauth_hint_builtins): New.
39246 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39247 (aarch64_expand_builtin): Expand new builtins.
39249 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39251 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39252 * combine-stack-adj.c (no_unhandled_cfa): Handle
39253 REG_CFA_TOGGLE_RA_MANGLE.
39254 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39255 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39256 info for return address signing.
39257 (aarch64_expand_epilogue): Likewise.
39259 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39261 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39262 * config/aarch64/aarch64-protos.h
39263 (aarch64_return_address_signing_enabled): New declaration.
39264 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39266 (aarch64_expand_prologue): Sign return address before it's pushed onto
39268 (aarch64_expand_epilogue): Authenticate return address fetched from
39270 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39271 (aarch64_attributes): New function attributes for "sign-return-address".
39272 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39273 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39274 ("*do_return"): Generate combined instructions according to key index.
39275 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39276 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39278 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39279 * config/aarch64/aarch64.opt (msign-return-address=): New.
39280 * doc/extend.texi (AArch64 Function Attributes): Documents
39281 "sign-return-address=".
39282 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39284 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39286 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39287 overall option summary.
39289 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39291 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39292 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39293 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39294 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39296 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39298 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39299 -mpower9-minmax by default for -mcpu=power9.
39300 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39301 128-bit floating point.
39303 2017-01-20 Alan Modra <amodra@gmail.com>
39305 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39306 optimizing for size.
39308 2017-01-20 Alan Modra <amodra@gmail.com>
39311 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39312 for strcmp and strncmp from corresponding builtin decl.
39314 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39316 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39317 instead of i386/rtems-64.h.
39318 * config/i386/rtems-64.h: Remove.
39320 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39324 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39326 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39328 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39330 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39331 Change int to HOST_WIDE_INT.
39332 * config/aarch64/aarch64-protos.h
39333 (aarch64_simd_gen_const_vector_dup): Likewise.
39334 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39336 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39338 * langhooks-def.h (lhd_type_for_size): New decl.
39339 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39340 * langhooks.c (lhd_type_for_size): New function, taken from
39343 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39345 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39346 define_bypass for CR latency.
39347 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39348 (power9-alu2): Add define_bypass for CR latency.
39350 (power9-mul): Update insn latency.
39351 (power9-mul-compare): Update insn latency, bypass latency and remove
39354 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39356 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39358 * config/aarch64/aarch64.md
39359 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39360 aarch64_nopcrelative_literal_loads.
39361 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39363 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39365 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39366 TARGET_LOONGSON_3A.
39367 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39369 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39372 * config.gcc (supported_defaults): Add lxc1-sxc1.
39373 (with_lxc1_sxc1): Add validation.
39374 (all_defaults): Add lxc1-sxc1.
39375 * config/mips/mips.opt (mlxc1-sxc1): New option.
39376 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39378 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39379 __mips_no_lxc1_sxc1.
39380 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39381 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39382 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39384 2017-01-19 Richard Biener <rguenther@suse.de>
39386 PR tree-optimization/72488
39387 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39388 sure to restore SSA info.
39389 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39391 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39393 PR rtl-optimization/79121
39394 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39395 of the inner type when shifting an extended value.
39397 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39400 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39401 interposable aliases.
39403 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39406 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39407 Use the evmergelohi instruction.
39408 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39409 (mov_sitf_e500_subreg8_2_be): Likewise.
39410 (mov_sitf_e500_subreg12_2_le): Likewise.
39411 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39412 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39413 (mov_sitf_e500_subreg8_2_le): Likewise.
39414 (mov_sitf_e500_subreg12_2_be): Likewise.
39416 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39418 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39419 attribute from vecsimple to vecperm.
39420 (altivec_vbpermq2): Likewise.
39422 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39425 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39427 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39428 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39429 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39430 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39431 case where N arg is SIZE_MAX.
39432 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39433 (cmpstrsi): Add pattern.
39435 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39437 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39438 __builtin_vec_revb builtins.
39439 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39440 built-in functions to support generation of the ISA 3.0 XXBR<x>
39441 vector byte reverse instructions.
39442 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39443 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39444 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39445 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39446 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39447 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39448 (P9V_BUILTIN_VEC_REVB): Likewise.
39449 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39450 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39451 (p9_xxbrq_v16qi): Likewise.
39452 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39453 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39454 (p9_xxbrh_v8hi): Likewise.
39455 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39456 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39457 vec_revb built-in functions.
39459 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39461 PR rtl-optimization/78952
39462 * config/i386/i386.md (any_extract): New code iterator.
39463 (*insvqi_2): Use any_extract for source operand.
39464 (*insvqi_3): Use any_shiftrt for source operand.
39466 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39468 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39470 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39472 2017-01-18 Matthias Klose <doko@ubuntu.com>
39474 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39476 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39478 * config/rs6000/altivec.h (vec_bperm): Change #define.
39479 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39480 (altivec_vbpermq2): New define_insn.
39481 (altivec_vbpermd): Likewise.
39482 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39483 function interface.
39484 (VBPERMD): Likewise.
39485 (VBPERM): New polymorphic function interface.
39486 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39487 Add entries for P9V_BUILTIN_VEC_VBPERM.
39488 * doc/extend.texi: Add interfaces for vec_bperm.
39490 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39492 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39493 first letter of error messages.
39494 (s390_resolve_overloaded_builtin): Likewise.
39495 * config/s390/s390.c (s390_expand_builtin): Likewise.
39496 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39497 (s390_valid_target_attribute_inner_p): Likewise.
39498 * config/s390/s390.md ("tabort"): Likewise.
39500 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39502 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39503 (ISA_AVOID_DIV_HILO): New macro.
39504 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39505 (ISA_HAS_DDIV): Likewise.
39507 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39509 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39511 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39513 * doc/invoke.texi (fabi-version): Spelling fix.
39515 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39518 * doc/invoke.texi (fabi-version): Mention mangling fix for
39521 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39524 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39526 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39529 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39530 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39532 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39534 (stack_protect_test): Ditto.
39535 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39536 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39538 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39539 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39540 -mstack-protector-guard-offset=.
39541 (RS/6000 and PowerPC Options): Ditto.
39543 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39545 * config/i386/i386.h (MASK_CLASS_P): New define.
39546 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39547 there are no registers from different register sets also when
39548 mask registers are used. Update function comment.
39549 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39550 to (*k/*r) and (*k/*km) alternatives.
39552 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39554 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39555 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39556 (EH_RETURN_HANDLER_RTX): New define.
39557 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39558 Force frame pointer in EH return functions.
39559 (aarch64_expand_epilogue): Add barrier for eh_return.
39560 (aarch64_final_eh_return_addr): Remove.
39561 (aarch64_eh_return_handler_rtx): New function.
39562 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39564 (aarch64_eh_return_handler_rtx): New prototype.
39566 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39568 * config/rs6000/altivec.h (vec_rlmi): New #define.
39569 (vec_vrlnm): Likewise.
39570 (vec_rlnm): Likewise.
39571 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39572 (UNSPEC_VRLNM): Likewise.
39573 (VIlong): New mode iterator.
39574 (altivec_vrl<VI_char>mi): New define_insn.
39575 (altivec_vrl<VI_char>nm): Likewise.
39576 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39578 (VRLDNM): Likewise.
39579 (RLNM): New polymorphic function entry.
39580 (VRLWMI): New monomorphic function entry.
39581 (VRLDMI): Likewise.
39582 (RLMI): New polymorphic function entry.
39583 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39584 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39585 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39588 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39591 * dwarf2out.c (field_byte_offset): Restore the
39592 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39593 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39594 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39597 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39600 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39602 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39605 * dwarf2out.c (add_data_member_location_attribute): For constant
39606 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39607 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39608 DW_AT_byte_size attributes.
39610 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39612 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39613 after forcing to constant memory when the code model is medium.
39615 2017-01-17 Julia Koval <julia.koval@intel.com>
39618 * config/i386/avx512fintrin.h
39619 (_mm512_i32gather_ps): Change __addr type to void const*.
39620 (_mm512_mask_i32gather_ps): Ditto.
39621 (_mm512_i32gather_pd): Ditto.
39622 (_mm512_mask_i32gather_pd): Ditto.
39623 (_mm512_i64gather_ps): Ditto.
39624 (_mm512_mask_i64gather_ps): Ditto.
39625 (_mm512_i64gather_pd): Ditto.
39626 (_mm512_mask_i64gather_pd): Ditto.
39627 (_mm512_i32gather_epi32): Ditto.
39628 (_mm512_mask_i32gather_epi32): Ditto.
39629 (_mm512_i32gather_epi64): Ditto.
39630 (_mm512_mask_i32gather_epi64): Ditto.
39631 (_mm512_i64gather_epi32): Ditto.
39632 (_mm512_mask_i64gather_epi32): Ditto.
39633 (_mm512_i64gather_epi64): Ditto.
39634 (_mm512_mask_i64gather_epi64): Ditto.
39635 (_mm512_i32scatter_ps): Change __addr type to void*.
39636 (_mm512_mask_i32scatter_ps): Ditto.
39637 (_mm512_i32scatter_pd): Ditto.
39638 (_mm512_mask_i32scatter_pd): Ditto.
39639 (_mm512_i64scatter_ps): Ditto.
39640 (_mm512_mask_i64scatter_ps): Ditto.
39641 (_mm512_i64scatter_pd): Ditto.
39642 (_mm512_mask_i64scatter_pd): Ditto.
39643 (_mm512_i32scatter_epi32): Ditto.
39644 (_mm512_mask_i32scatter_epi32): Ditto.
39645 (_mm512_i32scatter_epi64): Ditto.
39646 (_mm512_mask_i32scatter_epi64): Ditto.
39647 (_mm512_i64scatter_epi32): Ditto.
39648 (_mm512_mask_i64scatter_epi32): Ditto.
39649 (_mm512_i64scatter_epi64): Ditto.
39650 (_mm512_mask_i64scatter_epi64): Ditto.
39651 * config/i386/avx512pfintrin.h
39652 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39653 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39654 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39655 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39656 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39657 (_mm512_prefetch_i32scatter_ps): Ditto.
39658 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39659 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39660 (_mm512_prefetch_i64scatter_pd): Ditto.
39661 (_mm512_prefetch_i64scatter_ps): Ditto.
39662 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39663 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39664 * config/i386/avx512vlintrin.h
39665 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39666 (_mm_mmask_i32gather_ps): Ditto.
39667 (_mm256_mmask_i32gather_pd): Ditto.
39668 (_mm_mmask_i32gather_pd): Ditto.
39669 (_mm256_mmask_i64gather_ps): Ditto.
39670 (_mm_mmask_i64gather_ps): Ditto.
39671 (_mm256_mmask_i64gather_pd): Ditto.
39672 (_mm_mmask_i64gather_pd): Ditto.
39673 (_mm256_mmask_i32gather_epi32): Ditto.
39674 (_mm_mmask_i32gather_epi32): Ditto.
39675 (_mm256_mmask_i32gather_epi64): Ditto.
39676 (_mm_mmask_i32gather_epi64): Ditto.
39677 (_mm256_mmask_i64gather_epi32): Ditto.
39678 (_mm_mmask_i64gather_epi32): Ditto.
39679 (_mm256_mmask_i64gather_epi64): Ditto.
39680 (_mm_mmask_i64gather_epi64): Ditto.
39681 (_mm256_i32scatter_ps): Change __addr type to void*.
39682 (_mm256_mask_i32scatter_ps): Ditto.
39683 (_mm_i32scatter_ps): Ditto.
39684 (_mm_mask_i32scatter_ps): Ditto.
39685 (_mm256_i32scatter_pd): Ditto.
39686 (_mm256_mask_i32scatter_pd): Ditto.
39687 (_mm_i32scatter_pd): Ditto.
39688 (_mm_mask_i32scatter_pd): Ditto.
39689 (_mm256_i64scatter_ps): Ditto.
39690 (_mm256_mask_i64scatter_ps): Ditto.
39691 (_mm_i64scatter_ps): Ditto.
39692 (_mm_mask_i64scatter_ps): Ditto.
39693 (_mm256_i64scatter_pd): Ditto.
39694 (_mm256_mask_i64scatter_pd): Ditto.
39695 (_mm_i64scatter_pd): Ditto.
39696 (_mm_mask_i64scatter_pd): Ditto.
39697 (_mm256_i32scatter_epi32): Ditto.
39698 (_mm256_mask_i32scatter_epi32): Ditto.
39699 (_mm_i32scatter_epi32): Ditto.
39700 (_mm_mask_i32scatter_epi32): Ditto.
39701 (_mm256_i32scatter_epi64): Ditto.
39702 (_mm256_mask_i32scatter_epi64): Ditto.
39703 (_mm_i32scatter_epi64): Ditto.
39704 (_mm_mask_i32scatter_epi64): Ditto.
39705 (_mm256_i64scatter_epi32): Ditto.
39706 (_mm256_mask_i64scatter_epi32): Ditto.
39707 (_mm_i64scatter_epi32): Ditto.
39708 (_mm_mask_i64scatter_epi32): Ditto.
39709 (_mm256_i64scatter_epi64): Ditto.
39710 (_mm256_mask_i64scatter_epi64): Ditto.
39711 (_mm_i64scatter_epi64): Ditto.
39712 (_mm_mask_i64scatter_epi64): Ditto.
39713 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39714 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39715 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39716 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39717 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39718 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39719 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39720 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39721 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39722 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39723 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39724 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39725 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39726 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39727 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39728 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39729 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39730 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39731 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39732 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39733 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39734 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39735 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39736 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39737 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39738 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39739 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39740 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39741 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39742 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39743 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39744 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39745 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39746 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39747 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39748 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39749 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39750 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39751 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39752 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39753 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39754 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39755 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39756 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39757 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39758 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39759 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39760 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39761 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39762 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39763 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39764 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39765 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39766 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39767 definitions accordingly.
39769 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39770 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39773 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39776 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39779 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39780 hard regs for inner regno.
39782 2017-01-17 Martin Liska <mliska@suse.cz>
39785 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39786 assumption and add comment.
39788 2017-01-17 Nathan Sidwell <nathan@acm.org>
39790 * ipa-visibility.c (localize_node): New function, broken out of ...
39791 (function_and_variable_visibility): ... here. Call it.
39793 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39795 PR middle-end/77445
39796 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39797 correctly set frequency of oudgoing edge.
39798 (duplicate_thread_path): Fix profile updating.
39800 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39803 * configure.ac: Add GCC_BASE_VER.
39804 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39805 version from BASE-VER file.
39806 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39807 (gcc.o): Depend on $(BASEVER).
39808 * common.opt (dumpfullversion): New option.
39809 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39810 * doc/invoke.texi: Document -dumpfullversion.
39811 * doc/install.texi: Document --with-gcc-major-version-only.
39812 * configure: Regenerated.
39814 2017-01-17 Richard Biener <rguenther@suse.de>
39816 PR tree-optimization/71433
39817 * tree-vrp.c (register_new_assert_for): Merge same asserts
39818 on all incoming edges.
39819 (process_assert_insertions_for): Handle insertions at the
39822 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39824 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39825 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39827 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39830 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39833 2017-01-17 Alan Modra <amodra@gmail.com>
39836 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39837 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39838 symbolic stack limit when pic.
39840 2017-01-16 Martin Sebor <msebor@redhat.com>
39842 PR tree-optimization/78608
39843 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39845 2017-01-16 Jeff Law <law@redhat.com>
39848 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39849 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39850 for several include directories that may be relative to sysroot.
39851 * config/i386/x-mingw32 (gplus_includedir): Define.
39852 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39853 (native_system_includedir): Likewise.
39854 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39855 override if TARGET_SYSTEM_ROOT is defined.
39856 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39858 PR tree-optimization/79090
39859 PR tree-optimization/33562
39860 PR tree-optimization/61912
39861 PR tree-optimization/77485
39862 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39863 and computed trims into the dump file.
39865 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39867 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39869 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39872 * gimplify.c (gimplify_init_constructor): If want_value and
39873 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39877 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39878 sequence. Formatting fixes.
39879 (doloop_optimize): Formatting fixes.
39882 * gcc.c (debug_level_greater_than_spec_func): New function.
39883 (static_spec_functions): Add debug-level-gt spec function.
39884 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39886 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39887 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39888 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39889 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39890 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39891 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39893 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39895 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39896 QImode fixups to general and mask registers only.
39898 2017-01-16 Carl Love <cel@us.ibm.com>
39900 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39901 for built-in functions
39902 vector signed char vec_nabs (vector signed char)
39903 vector signed short vec_nabs (vector signed short)
39904 vector signed int vec_nabs (vector signed int)
39905 vector signed long long vec_nabs (vector signed long long)
39906 vector float vec_nabs (vector float)
39907 vector double vec_nabs (vector double)
39908 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39910 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39911 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39912 * doc/extend.texi: Update the documentation file for the new built-in
39915 2017-01-16 Martin Sebor <msebor@redhat.com>
39917 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39920 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39922 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39923 UNSPEC_VSX__XXSPLTD to require special splat handling.
39925 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39928 * system.h: Poison strndup.
39930 2017-01-16 Alan Modra <amodra@gmail.com>
39933 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39936 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39938 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39940 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39942 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39943 call recog here. Assert that INSN_CODE (insn) is non-negative.
39945 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39948 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39950 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39951 in the currently scheduled RTL fragment.
39953 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39955 PR rtl-optimization/78751
39956 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39959 2017-01-14 Jeff Law <law@redhat.com>
39961 PR tree-optimization/79090
39962 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39963 variable length stores.
39964 (compute_trims): Delete dead assignment to *trim_tail.
39965 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39968 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39970 PR rtl-optimization/78626
39971 PR rtl-optimization/78727
39972 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39973 of a block, and split such blocks after everything else is finished.
39975 2017-01-14 Alan Modra <amodra@gmail.com>
39978 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39979 target legitimate_combined_insn.
39980 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39981 (rs6000_legitimate_combined_insn): New function.
39982 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39984 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39985 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39986 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39988 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39990 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39992 2017-01-13 Jeff Law <law@redhat.com>
39994 PR tree-optimization/33562
39995 PR tree-optimization/61912
39996 PR tree-optimization/77485
39997 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39999 (delete_dead_assignment): Likewise.
40000 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40001 statement to delete_dead_call and delete_dead_assignment.
40003 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40006 * substring-locations.c (format_warning_va): Strengthen case 1 so
40007 that both endpoints of the substring must be within the format
40008 range for just the substring to be printed.
40010 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40012 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40013 * config/i386/i386.c (ix86_target_string): Add missing options
40014 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40015 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40016 flags_other and ix86_target_other to flags2_other. Display unknown
40018 (ix86_valid_target_attribute_inner_p): Add missing options and
40019 reorder options by implied ISAs, as in ix86_target_string.
40021 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40023 * hash-table.h (hash_table::too_empty_p): New function.
40024 (hash_table::expand): Use it.
40025 (hash_table::traverse): Likewise.
40026 (hash_table::empty_slot): Use sizeof (value_type) instead of
40027 sizeof (PTR) to convert bytes to elements. Shrink the table
40028 if the current size is excessive for the current number of
40031 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40033 * ira-costs.c (record_reg_classes): Break from the inner loop
40034 early once alt_fail is known to be true. Update outer loop
40035 handling accordingly.
40037 2017-01-13 Jeff Law <law@redhat.com>
40039 * tree-ssa-dse.c (decrement_count): New function.
40040 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40041 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40042 when we know the partially dead statement is a mem* function.
40044 PR tree-optimization/61912
40045 PR tree-optimization/77485
40046 * tree-ssa-dse.c: Include expr.h.
40047 (maybe_trim_constructor_store): New function.
40048 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40050 PR tree-optimization/33562
40051 PR tree-optimization/61912
40052 PR tree-optimization/77485
40053 * doc/invoke.texi: Document new dse-max-object-size param.
40054 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40055 * tree-ssa-dse.c: Include params.h.
40056 (dse_store_status): New enum.
40057 (initialize_ao_ref_for_dse): New, partially extracted from
40059 (valid_ao_ref_for_dse, normalize_ref): New.
40060 (setup_live_bytes_from_ref, compute_trims): Likewise.
40061 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40062 (maybe_trim_partially_dead_store): Likewise.
40063 (maybe_trim_complex_store): Likewise.
40064 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40065 Track what bytes live from the original store. Return tri-state
40066 for dead, partially dead or live.
40067 (dse_dom_walker): Add constructor, destructor and new private members.
40068 (delete_dead_call, delete_dead_assignment): New extracted from
40070 (dse_optimize_stmt): Make a member of dse_dom_walker.
40071 Use initialize_ao_ref_for_dse.
40073 PR tree-optimization/33562
40074 PR tree-optimization/61912
40075 PR tree-optimization/77485
40076 * sbitmap.h (bitmap_count_bits): Prototype.
40077 (bitmap_clear_range, bitmap_set_range): Likewise.
40078 * sbitmap.c (bitmap_clear_range): New function.
40079 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40081 2017-01-13 Martin Liska <mliska@suse.cz>
40084 * function.c (set_cfun): Add new argument force.
40085 * function.h (set_cfun): Likewise.
40086 * ipa-inline-transform.c (inline_call): Use the function when
40087 strict alising from is dropped for function we inline to.
40089 2017-01-13 Richard Biener <rguenther@suse.de>
40091 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40092 for dumping GIMPLE INTEGER_CSTs.
40094 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40096 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40097 to 201112L since C++17.
40099 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40102 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40103 if -fsanitize=kernel-address is present.
40105 2017-01-13 Richard Biener <rguenther@suse.de>
40107 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40108 as _Literal ( type ) number in case usual suffixes do not
40109 preserve all information.
40111 2017-01-13 Richard Biener <rguenther@suse.de>
40113 PR tree-optimization/77283
40114 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40115 and ssa-iterators.h.
40116 (is_feasible_trace): Implement a cost model based on joiner
40119 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40122 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40123 char or short to __float128/_Float128 directly.
40125 2017-01-12 Martin Sebor <msebor@redhat.com>
40127 to -Wformat-overflow.
40128 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40129 (min_bytes_remaining): Same.
40130 (get_string_length): Same.
40131 (format_string): Same.
40132 (format_directive): Same.
40134 (pass_sprintf_length::handle_gimple_call): Same.
40136 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40138 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40139 info.nowrite calls with no lhs that can't throw. Return bool
40140 whether gsi_remove has been called or not.
40141 (pass_sprintf_length::handle_gimple_call): Return bool whether
40142 try_substitute_return_value called gsi_remove. Formatting fix.
40143 (pass_sprintf_length::execute): Don't use gsi_remove if
40144 handle_gimple_call returned true.
40147 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40148 be removed due to side-effects, don't remove following barrier nor
40149 turn the successor edge into fallthru edge.
40151 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40154 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40155 element-reversing loads and stores as not swappable.
40157 2017-01-12 Nathan Sidwell <nathan@acm.org>
40158 Nicolai Stange <nicstange@gmail.com>
40160 * combine.c (try_combine): Don't ignore result of overlap checking
40161 loop. Combine overlap & asm check into single loop.
40163 2017-01-12 Richard Biener <rguenther@suse.de>
40165 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40166 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40168 2017-01-12 Richard Biener <rguenther@suse.de>
40170 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40171 and TS_TARGET_OPTION directly derive from TS_BASE.
40172 * tree-core.h (tree_optimization_option): Derive from tree_base.
40173 (tree_target_option): Likewise.
40175 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40177 * config/i386/i386.c (memory_address_length): Increase len
40178 only when rip_relative_addr_p returns false.
40180 2017-01-11 Julia Koval <julia.koval@intel.com>
40182 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40183 (OPTION_MASK_ISA_SGX_SET): New.
40184 (ix86_handle_option): Handle OPT_msgx.
40185 * config.gcc: Added sgxintrin.h.
40186 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40187 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40188 * config/i386/i386.c (ix86_target_string): Add -msgx.
40190 (ix86_option_override_internal): Handle new options.
40191 (ix86_valid_target_attribute_inner_p): Add sgx.
40192 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40193 * config/i386/i386.opt: Add msgx.
40194 * config/i386/sgxintrin.h: New file.
40195 * config/i386/x86intrin.h: Add sgxintrin.h.
40197 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40200 * fold-const.c (maybe_nonzero_address): Return 1 for function
40202 (tree_single_nonzero_warnv_p): Don't handle function local objects
40206 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40209 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40212 * opts.c: Include "spellcheck.h"
40213 (struct string_fragment): New struct.
40214 (struct edit_distance_traits<const string_fragment &>): New
40216 (get_closest_sanitizer_option): New function.
40217 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40219 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40221 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40223 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40224 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40225 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40226 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40227 for initial die_offset if dwarf_split_debug_info.
40228 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40229 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40231 (output_skeleton_debug_sections): Formatting fix. Use
40232 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40233 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40235 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40237 * config/arm/cortex-a53.md: Add bypasses for
40238 cortex_a53_r2f_cvt.
40239 (cortex_a53_r2f): Only use for transfers.
40240 (cortex_a53_f2r): Likewise.
40241 (cortex_a53_r2f_cvt): Add reservation for conversions.
40242 (cortex_a53_f2r_cvt): Likewise.
40244 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40246 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40247 to all inlined functions, change static to extern.
40249 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40252 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40254 (arm_assemble_integer): Likewise.
40256 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40258 * config.gcc: Use new awk script to check CPU, FPU and architecture
40259 parameters for --with-... options.
40260 * config/arm/parsecpu.awk: New file
40261 * config/arm/arm-cpus.in: New file.
40262 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40264 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40266 * config/arm/t-arm: Update dependency rules.
40267 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40268 of processing .def files.
40269 * config/arm/genopt.sh: Deleted.
40270 * config/arm/gentune.sh: Deleted.
40271 * config/arm/arm-cores.def: Deleted.
40272 * config/arm/arm-arches.def: Deleted.
40273 * config/arm/arm-fpus.def: Deleted.
40274 * config/arm/arm-tune.md: Regenerated.
40275 * config/arm/arm-tables.opt: Regenerated.
40276 * config/arm/arm-cpu.h: New generated file.
40277 * config/arm/arm-cpu-data.h: New generated file.
40278 * config/arm/arm-cpu-cdata.h: New generated file.
40280 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40283 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40285 (input_varpool_node): Unpack dynamically_initialized bit.
40287 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40289 PR rtl-optimization/79032
40290 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40291 the alignment of the adjusted memory reference against that of MODE,
40292 instead of the alignment of the original memory reference.
40294 2017-01-11 Martin Jambor <mjambor@suse.cz>
40296 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40298 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40299 decorated functions.
40301 2017-01-11 Richard Biener <rguenther@suse.de>
40303 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40304 set range/nonnull info for PHI results. Do not set it on
40305 stmts marked for removal.
40307 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40309 * expr.c (store_field): In the bitfield case, fetch the return value
40310 from the registers before applying a single big-endian adjustment.
40311 Always do a final load for a BLKmode value not larger than a word.
40313 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40316 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40317 that we correctly handle column numbers greater than
40318 LINE_MAP_MAX_COLUMN_NUMBER.
40320 2017-01-10 Martin Sebor <msebor@redhat.com>
40322 PR middle-end/78245
40323 * gimple-ssa-sprintf.c (get_destination_size): Call
40324 {init,fini}object_sizes.
40325 * tree-object-size.c (addr_object_size): Adjust.
40326 (pass_through_call): Adjust.
40327 (pass_object_sizes::execute): Adjust.
40328 * tree-object-size.h (fini_object_sizes): Declare.
40330 2017-01-10 Martin Sebor <msebor@redhat.com>
40332 PR tree-optimization/78775
40333 * builtins.c (get_size_range): Move...
40334 * calls.c: ...to here.
40335 (alloc_max_size): Accept zero argument.
40336 (operand_signed_p): Remove.
40337 (maybe_warn_alloc_args_overflow): Call get_size_range.
40338 * calls.h (get_size_range): Declare.
40340 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40342 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40343 from TI's devices.csv file as of September 2016.
40344 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40346 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40348 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40349 * doc/invoke.texi: Likewise.
40350 * doc/md.texi: Likewise.
40351 * doc/objc.texi: Likewise.
40353 2017-01-10 Joshua Conner <joshconner@google.com>
40355 * config/arm/fuchsia-elf.h: New file.
40356 * config/fuchsia.h: New file.
40357 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40358 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40360 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40362 2016-01-10 Richard Biener <rguenther@suse.de>
40364 PR tree-optimization/79034
40365 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40366 Propagate out degenerate PHIs in the joiner.
40368 2017-01-10 Martin Liska <mliska@suse.cz>
40370 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40371 (sort_congruence_classes_by_decl_uid): Likewise.
40372 (sort_congruence_class_groups_by_decl_uid): Likewise.
40373 (sem_item_optimizer::merge_classes): Sort class, groups in these
40374 classes and members in the groups by DECL_UID of declarations.
40375 This would make merge operations stable.
40377 2017-01-10 Martin Liska <mliska@suse.cz>
40379 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40380 usage of m_classes_vec.
40381 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40382 (sem_item_optimizer::get_group_by_hash): Likewise.
40383 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40384 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40385 (sem_item_optimizer::verify_classes): Likewise.
40386 (sem_item_optimizer::process_cong_reduction): Likewise.
40387 (sem_item_optimizer::dump_cong_classes): Likewise.
40388 (sem_item_optimizer::merge_classes): Likewise.
40389 * ipa-icf.h (congruence_class_hash): Rename from
40390 congruence_class_group_hash. Remove declaration of m_classes_vec.
40392 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40394 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40395 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40396 * config.gcc: Add avx512vpopcntdqintrin.h.
40397 * config/i386/avx512vpopcntdqintrin.h: New.
40398 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40399 * config/i386/i386-builtin-types.def: Add new types.
40400 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40401 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40402 __builtin_ia32_vpopcountq_v8di_mask): New.
40403 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40404 __AVX512VPOPCNTDQ__.
40405 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40406 (PTA_AVX512VPOPCNTDQ): Define.
40407 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40408 TARGET_AVX512VPOPCNTDQ_P): Define.
40409 * config/i386/i386.opt: Add mavx512vpopcntdq.
40410 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40411 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40413 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40415 PR middle-end/77484
40416 * predict.def (PRED_CALL): Set to 67.
40418 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40420 * expr.c (store_field): In the bitfield case, if the value comes from
40421 a function call and is of an aggregate type returned in registers, do
40422 not modify the field mode; extract the value in all cases if the mode
40423 is BLKmode and the size is not larger than a word.
40425 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40428 * config/i386/cpuid.h: Fix undefined behavior.
40430 2017-01-04 Jeff Law <law@redhat.com>
40432 PR tree-optimization/79007
40433 PR tree-optimization/67955
40434 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40435 conservative for pt.null when flag_non_call_exceptions is on.
40437 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40439 PR translation/79019
40440 PR translation/79020
40441 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40442 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40443 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40445 * config/avr/avr.opt (maccumulate-args): Likewise.
40446 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40447 * common.opt (freport-bug): Likewise.
40448 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40449 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40450 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40451 translatable string.
40452 * config/i386/i386.c (function_value_32): Likewise.
40453 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40454 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40456 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40457 * common/config/msp430/msp430-common.c (msp430_handle_option):
40459 * symtab.c (symtab_node::verify_base): Likewise.
40460 * opts.c (set_debug_level): Likewise.
40461 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40462 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40463 missing whitespace to translatable strings.
40464 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40465 * config/sh/superh.h: Likewise.
40466 * config/i386/xopintrin.h: Likewise.
40467 * config/i386/znver1.md: Likewise.
40468 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40469 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40470 * double-int.h (struct double_int): Likewise.
40471 * double-int.c (div_and_round_double): Likewise.
40472 * wide-int.cc: Likewise.
40473 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40474 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40475 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40476 (crossjumps_occurred): ... this.
40477 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40480 PR tree-optimization/78899
40481 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40482 returning bool return struct loop *, NULL for failure and the new
40484 (versionable_outer_loop_p): Don't version outer loop if it has
40485 dont_vectorized bit set.
40486 (tree_if_conversion): When versioning outer loop, ensure
40487 tree_if_conversion is performed also on the inner loop of the
40488 non-vectorizable outer loop copy.
40489 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40490 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40491 prevent vectorization of it.
40492 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40493 the outer loop vectorization of the non-scalar version is attempted
40494 before vectorization of the inner loop in scalar version. If
40495 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40496 vectorization of its inner loop.
40497 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40498 has 2 inner loops, rename also on edges from bb whose single pred
40499 is outer_loop->header. Fix typo in function comment.
40501 2017-01-09 Martin Sebor <msebor@redhat.com>
40504 * asan.c (asan_emit_stack_protection): Increase local buffer size
40505 to avoid snprintf truncation warning.
40507 2017-01-09 Andrew Pinski <apinski@cavium.com>
40509 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40510 to reference thunderx2t99 for the tuning structure
40511 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40513 (thunderx2t99_extra_costs): This.
40514 * config/aarch64/aarch64-tune.md: Regenerate.
40515 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40516 (vulcan_addrcost_table): This.
40517 (vulcan_regmove_cost): Rename to ...
40518 (thunderx2t99_regmove_cost): This.
40519 (vulcan_vector_cost): Rename to ...
40520 (thunderx2t99_vector_cost): this.
40521 (vulcan_branch_cost): Rename to ...
40522 (thunderx2t99_branch_cost): This.
40523 (vulcan_tunings): Rename to ...
40524 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40525 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40527 2017-01-09 Martin Jambor <mjambor@suse.cz>
40531 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40532 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40533 (propagate_vr_accross_jump_function): Use the above function for all
40534 value range computations for pass-through jump functions and type
40535 converasion from explicit value range values.
40536 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40537 parameters from TYPE_ARG_TYPES.
40538 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40539 (ipa_write_node_info): Stream type of the actual argument.
40540 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40542 2017-01-09 Martin Liska <mliska@suse.cz>
40545 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40546 (lookup_compiler): Do not show error message with have_E.
40548 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40550 PR tree-optimization/78938
40551 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40552 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40553 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40554 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40557 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40559 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40562 2017-01-09 Richard Biener <rguenther@suse.de>
40564 PR tree-optimization/78997
40565 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40566 name condition properly.
40568 2017-01-09 Richard Biener <rguenther@suse.de>
40571 * dwarf2out.c (is_cxx): New overload with context.
40572 (is_naming_typedef_decl): Use it.
40574 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40576 * invoke.texi (Option Summary): Correct spacing in option lists
40577 and add line breaks to fix over-long lines.
40579 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40581 PR middle-end/17660
40583 * extend.texi (Common Variable Attributes): Add xref to GCC
40584 Internals manual to explain mode attribute keywords.
40586 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40589 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40590 and Preprocessor Options.
40591 (Options for Linking): Document -pthread here....
40592 (RS/6000 and PowerPC Options): ...not here.
40593 (Solaris 2 Options): ...or here.
40594 * doc/cppopts.texi: Document -pthread.
40596 2017-01-08 Martin Sebor <msebor@redhat.com>
40598 PR middle-end/77708
40599 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40600 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40601 New member functions.
40602 (format_directive): Used them.
40604 (pass_sprintf_length::handle_gimple_call): Same.
40605 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40606 to avoid truncation for any argument.
40607 (extract_affine_mul): Same.
40608 * tree.c (get_file_function_name): Same.
40610 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40612 PR middle-end/77484
40613 * predict.def (PRED_INDIR_CALL): Set to 86.
40615 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40617 PR preprocessor/54124
40618 * doc/cppopts.texi: Reformat -d subtable to list the full name
40619 of the options. Add cross-reference to the docs for the general
40620 compiler -d options.
40621 * doc/invoke.texi (Developer Options): Add cross-reference to the
40622 preprocessor-specific -d option documentation.
40624 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40626 PR preprocessor/13498
40627 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40628 redudant material, and reflect new command-line options.
40629 (System Headers): Likewise.
40631 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40633 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40634 -isystem, and -idirafter. Copy-edit.
40635 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40636 default for -ftrack-macro-expansion. Delete obsolete and
40637 badly-formatted implementation details about -fdebug-cpp output.
40638 * doc/cppwarnopts.texi: Copy-edit.
40640 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40643 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40644 that the transition from a max line width >= 1<<10 to narrower
40645 lines works correctly.
40647 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40649 * doc/options.texi (PerFunction): New.
40650 * opt-functions.awk (switch_flags): Map both Optimization and
40651 PerFunction to CL_OPTIMIZATION.
40652 * opth-gen.awk: Test for PerFunction flag along with
40654 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40655 it only when the latter is present. Skip those that don't in
40656 the hash function generator.
40657 * common.opt (fvar-tracking): Mark as PerFunction instead of
40659 (fvar-tracking-assignments): Likewise.
40660 (fvar-tracking-assignments-toggle): Likewise.
40661 (fvar-tracking-uninit): Likewise.
40663 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40665 PR translation/79018
40666 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40669 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40672 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40673 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40674 TARGET_LONG_JUMP_TABLE_OFFSETS.
40675 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40676 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40677 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40678 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40679 * config/m68k/m68k.md (tablejump expander): Likewise.
40680 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40681 TARGET_LONG_JUMP_TABLE_OFFSETS.
40682 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40683 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40685 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40686 David Holsgrove <david.holsgrove@xilinx.com>
40688 * common/config/microblaze/microblaze-common.c
40689 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40690 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40692 * config/microblaze/microblaze.c (microblaze_must_save_register)
40693 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40695 (microblaze_eh_return): New function.
40696 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40697 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40698 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40699 * config/microblaze/microblaze.md (eh_return): New pattern.
40701 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40703 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40704 GCC_DIAGNOSTIC_STRINGIFY): Define.
40706 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40708 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40710 * config/arm/arm.md (<mcrr>): New.
40712 * config/arm/arm.c (arm_arch5te): New.
40713 (arm_option_override): Set arm_arch5te.
40714 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40716 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40717 (arm_mcrr_qualifiers): ... this. New.
40718 (MRRC_QUALIFIERS): Define to...
40719 (arm_mrrc_qualifiers): ... this. New.
40720 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40722 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40723 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40724 (MRRCI, mrrc, MRRC): New.
40725 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40726 VUNSPEC_MRRC2): New.
40728 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40730 * config/arm/arm.md (<mcr>): New.
40732 * config/arm/arm.c (arm_coproc_builtin_available): Add
40733 support for mcr, mrc, mcr2 and mrc2.
40734 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40735 (arm_mcr_qualifiers): ... this. New.
40736 (MRC_QUALIFIERS): Define to ...
40737 (arm_mrc_qualifiers): ... this. New.
40738 (MCR_QUALIFIERS): Define to ...
40739 (arm_mcr_qualifiers): ... this. New.
40740 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40742 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40743 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40744 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40745 VUNSPEC_MRC2): New.
40747 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40749 * config/arm/arm.md (*ldc): New.
40753 * config/arm/arm.c (arm_coproc_builtin_available): Add
40754 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40755 (arm_coproc_ldc_stc_legitimate_address): New.
40756 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40757 'qualifier_const_pointer'.
40758 (LDC_QUALIFIERS): Define to...
40759 (arm_ldc_qualifiers): ... this. New.
40760 (STC_QUALIFIERS): Define to...
40761 (arm_stc_qualifiers): ... this. New.
40762 * config/arm/arm-protos.h
40763 (arm_coproc_ldc_stc_legitimate_address): New.
40764 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40765 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40766 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40767 stc2, stcl, stc2l): New.
40768 * config/arm/constraints.md (Uz): New.
40769 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40770 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40771 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40772 VUNSPEC_STC2L): New.
40774 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40776 * config/arm/arm.md (<cdp>): New.
40777 * config/arm/arm.c (neon_const_bounds): Rename this ...
40778 (arm_const_bounds): ... this.
40779 (arm_coproc_builtin_available): New.
40780 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40781 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40782 (CDP_QUALIFIERS): Define to...
40783 (arm_cdp_qualifiers): ... this. New.
40785 (arm_expand_builtin_args): Add case for 6 arguments.
40786 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40787 (arm_const_bounds): ... this.
40788 (arm_coproc_builtin_available): New.
40789 * config/arm/arm_acle.h (__arm_cdp): New.
40791 * config/arm/arm_acle_builtins.def (cdp): New.
40793 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40794 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40795 'arm_const_bounds'.
40796 * config/arm/types.md (coproc): New.
40797 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40798 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40799 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40800 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40802 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40804 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40805 (UBINOP_QUALIFIERS): New.
40807 (acle_builtin_data): New. Change comment.
40808 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40809 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40810 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40811 arm_acle_builtins.def.
40812 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40813 (arm_init_acle_builtins): New.
40814 (CRC32_BUILTIN): Remove.
40815 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40816 crc32cb, crc32ch and crc32cw.
40817 (arm_init_crc32_builtins): Remove.
40818 (arm_init_builtins): Use arm_init_acle_builtins rather
40819 than arm_init_crc32_builtins.
40820 (arm_expand_acle_builtin): New.
40821 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40822 * config/arm/arm_acle_builtins.def: New.
40824 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40826 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40827 (arm_builtin_datum): ... this.
40828 (arm_init_neon_builtin): Rename to ...
40829 (arm_init_builtin): ... this. Add a new parameters PREFIX
40830 and USE_SIG_IN_NAME.
40831 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40832 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40833 'arm_builtin_datum'.
40834 (arm_init_vfp_builtins): Likewise.
40835 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40836 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40837 (arm_expand_neon_args): Rename to ...
40838 (arm_expand_builtin_args): ... this. Rename builtin_arg
40839 enum values and differentiate between ARG_BUILTIN_MEMORY
40840 and ARG_BUILTIN_NEON_MEMORY.
40841 (arm_expand_neon_builtin_1): Rename to ...
40842 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40843 values, arm_expand_builtin_args and add bool parameter NEON.
40844 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40845 (arm_expand_vfp_builtin): Likewise.
40846 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40848 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40850 PR middle-end/77484
40851 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40852 * predict.c (tree_estimate_probability_bb): Reverse direction of
40853 polymorphic call predictor.
40855 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40857 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40858 (determine_pass_name_match): ...this new function and...
40859 (should_skip_pass_p): ...this new function.
40861 2017-01-06 Nathan Sidwell <nathan@acm.org>
40863 * ipa-visibility.c (function_and_variable_visibility): Reformat
40864 comments and long lines. Remove extrneous if.
40865 * symtab.c (symtab_node::make_decl_local): Fix code format.
40866 (symtab_node::set_section_for_node): Fix comment typo.
40868 2017-01-06 Martin Liska <mliska@suse.cz>
40871 * lra-constraints.c: Rename invariant to lra_invariant.
40872 * predict.c (set_even_probabilities): Initialize e to NULL.
40874 2017-01-05 Martin Sebor <msebor@redhat.com>
40876 PR tree-optimization/78910
40877 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40878 (format_integer): Correct off-by-one error in the handling
40879 of precision with negative numbers in signed conversions..
40881 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40883 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40885 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40887 PR tree-optimization/71016
40888 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40889 factor_out_conditional_conversion. Formatting fix.
40890 (factor_out_conditional_conversion): Add cond_stmt argument.
40891 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40892 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40895 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40897 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40898 read-rtl-function.o, and selftest-rtl.o.
40899 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40900 (selftest::aarch64_test_loading_full_dump): New function.
40901 (selftest::aarch64_run_selftests): New function.
40902 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40903 selftest::aarch64_run_selftests.
40904 * config/i386/i386.c
40905 (selftest::ix86_test_loading_dump_fragment_1): New function.
40906 (selftest::ix86_test_loading_call_insn): New function.
40907 (selftest::ix86_test_loading_full_dump): New function.
40908 (selftest::ix86_test_loading_unspec): New function.
40909 (selftest::ix86_run_selftests): Call the new functions.
40910 * emit-rtl.c (maybe_set_max_label_num): New function.
40911 * emit-rtl.h (maybe_set_max_label_num): New decl.
40912 * function.c (instantiate_decls): Guard call to
40913 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40914 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40916 * gensupport.c (gen_reader::gen_reader): Pass "false"
40917 for new "compact" param of rtx_reader.
40918 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40919 rather than an empty string for NULL strings.
40920 * read-md.c: Potentially include config.h rather than bconfig.h.
40921 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40922 (have_error): New global, copied from errors.c.
40923 (md_reader::read_name): Rename to...
40924 (md_reader::read_name_1): ...this, adding "out_loc" param,
40925 and converting "missing name or number" to returning false, rather
40927 (md_reader::read_name): Reimplement in terms of read_name_1.
40928 (md_reader::read_name_or_nil): New function.
40929 (md_reader::read_string): Handle "(nil)" by returning NULL.
40930 (md_reader::md_reader): Add new param "compact".
40931 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40932 (md_reader::read_file): New method.
40933 * read-md.h (md_reader::md_reader): Add new param "compact".
40934 (md_reader::read_file): New method.
40935 (md_reader::is_compact): New accessor.
40936 (md_reader::read_name): Convert return type from void to file_location.
40937 (md_reader::read_name_or_nil): New decl.
40938 (md_reader::read_name_1): New decl.
40939 (md_reader::m_compact): New field.
40940 (noop_reader::noop_reader): Pass "false" for new "compact" param
40942 (rtx_reader::rtx_reader): Add new "compact" param.
40943 (rtx_reader::read_rtx_operand): Make virtual and convert return
40944 type from void to rtx.
40945 (rtx_reader::read_until): New decl.
40946 (rtx_reader::handle_any_trailing_information): New virtual function.
40947 (rtx_reader::postprocess): New virtual function.
40948 (rtx_reader::finalize_string): New virtual function.
40949 (rtx_reader::m_in_call_function_usage): New field.
40950 (rtx_reader::m_reuse_rtx_by_id): New field.
40951 * read-rtl-function.c: New file.
40952 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40953 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40954 (selftest::verify_three_block_rtl_cfg): New decl.
40955 * read-rtl-function.h: New file.
40956 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40957 For host, include function.h, memmodel.h, and emit-rtl.h.
40958 (one_time_initialization): New function.
40959 (struct compact_insn_name): New struct.
40960 (compact_insn_names): New array.
40961 (find_code): Handle insn codes in compact dumps.
40962 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40963 (bind_subst_iter_and_attr): Likewise.
40964 (add_condition_to_string): Likewise.
40965 (add_condition_to_rtx): Likewise.
40966 (apply_attribute_uses): Likewise.
40967 (add_current_iterators): Likewise.
40968 (apply_iterators): Likewise.
40969 (initialize_iterators): Guard usage of apply_subst_iterator with
40970 #ifdef GENERATOR_FILE.
40971 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40972 (md_reader::read_mapping): Likewise.
40973 (add_define_attr_for_define_subst): Likewise.
40974 (add_define_subst_attr): Likewise.
40975 (read_subst_mapping): Likewise.
40976 (check_code_iterator): Likewise.
40977 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40979 (one_time_initialization): New function.
40980 (rtx_reader::read_until): New method.
40981 (read_flags): New function.
40982 (parse_reg_note_name): New function.
40983 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40984 Handle reuse_rtx ids.
40985 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40986 Add parsing support for RTL dumps, mirroring the special-cases in
40987 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40988 values, and calling handle_any_trailing_information.
40989 (rtx_reader::read_rtx_operand): Convert return type from void
40990 to rtx, returning return_rtx. Handle case 'e'. Call
40991 finalize_string on XSTR and XTMPL fields.
40992 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40993 "(nil)" values were omitted. Call the postprocess vfunc on the
40995 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40996 class ctor. Initialize m_in_call_function_usage. Call
40997 one_time_initialization.
40998 * rtl-tests.c (selftest::test_uncond_jump): Call
40999 set_new_first_and_last_insn.
41000 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41001 * selftest-rtl.c: New file.
41002 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41003 (selftest::get_insn_by_uid): New decl.
41004 * selftest-run-tests.c (selftest::run_tests): Call
41005 read_rtl_function_c_tests.
41006 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41007 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41010 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41012 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41013 operands in a special way. Assert that pos+len <= mode precision.
41015 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41017 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41018 3 argument Alias with unlimited for the negative form.
41019 (fno-vect-cost-model): Removed.
41021 2017-01-05 Martin Liska <mliska@suse.cz>
41023 * hsa-gen.c (gen_hsa_divmod): New function.
41024 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41026 2017-01-05 Martin Liska <mliska@suse.cz>
41029 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41032 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41034 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41035 small constant length operands.
41037 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41039 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41040 between loop iterations.
41042 2017-01-05 Martin Liska <mliska@suse.cz>
41045 * gimplify.c (gimplify_decl_expr): Compare to
41046 asan_poisoned_variables instread of checking flags.
41047 (gimplify_target_expr): Likewise.
41048 (gimplify_expr): Likewise.
41049 (gimplify_function_tree): Conditionally initialize
41050 asan_poisoned_variables.
41052 2017-01-04 Jeff Law <law@redhat.com>
41054 PR tree-optimizatin/78812
41055 * rtl.h (contains_mem_rtx_p): Prototype.
41056 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41057 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41058 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41059 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41060 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41062 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41064 * input.c (assert_char_at_range): Default-initialize actual_range.
41066 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41068 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41069 to match the caller.
41071 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41073 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41074 insns after final jump in test to emit dummy move.
41076 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41078 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41079 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41081 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41083 * multiple_target.c (create_dispatcher_calls): Init e_next.
41084 * tree-ssa-loop-split.c (split_loop): Init border.
41085 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41088 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41093 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41094 (altivec_register_operand): Do not return true if the operand
41095 contains a SUBREG mixing SImode and SFmode.
41096 (vsx_register_operand): Likewise.
41097 (vsx_reg_sfsubreg_ok): New predicate.
41098 (vfloat_operand): Do not return true if the operand contains a
41099 SUBREG mixing SImode and SFmode.
41100 (vint_operand): Likewise.
41101 (vlogical_operand): Likewise.
41102 (gpc_reg_operand): Likewise.
41103 (int_reg_operand): Likewise.
41104 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41105 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41106 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41108 (rs6000_emit_move_si_sf_subreg): New helper function.
41109 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41110 fixup SUBREGs involving SImode and SFmode.
41111 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41112 numbers for the new peephole2 optimization.
41113 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41114 the GLIBC math library that do AND/IOR/XOR operations on single
41115 precision floating point.
41116 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41117 target macros to say whether we need to avoid SUBREGs mixing
41119 (TARGET_ALLOW_SF_SUBREG): Likewise.
41120 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41121 (UNSPEC_SI_FROM_SF): Likewise.
41122 (iorxor): Change spacing.
41123 (and_ior_xor): New iterator for AND, IOR, and XOR.
41124 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41125 (movdi_from_sf_zero_ext): Likewise.
41126 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41127 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41128 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41129 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41130 (fms<mode>4): Likewise.
41131 (fnma<mode>4): Likewise.
41132 (fnms<mode>4): Likewise.
41133 (nfma<mode>4): Likewise.
41134 (nfms<mode>4): Likewise.
41136 2017-01-04 Marek Polacek <polacek@redhat.com>
41139 * doc/invoke.texi: Document -Wpointer-compare.
41141 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41143 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41146 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41147 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41150 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41153 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41154 documentation of the powerpc_popcntb_ok attribute.
41155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41156 code to issue warning messages if a requested CPU configuration is
41157 not supported by the binary (assembler and loader) toolchain.
41158 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41159 made to define a built-in function that has been disabled.
41160 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41161 made to define a built-in function that has been disabled.
41162 (altivec_init_builtins): Add comment explaining why definition
41163 of the DST built-in functions is not preceded by an assertion
41164 check. Add assertions to prevent ICE if attempts are made to
41165 define an altivec predicate or an abs* built-in function that has
41167 (htm_init_builtins): Add comment explaining why definition of the
41168 htm built-in functions is not preceded by an assertion check.
41170 2017-01-04 Jeff Law <law@redhat.com>
41172 PR tree-optimizatin/67955
41173 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41174 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41175 the points-to solution does not include pt_null. Use DECL_PT_UID
41178 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41180 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41181 Use gen_int_mode instead of gen_lopwart for const_int operands.
41183 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41185 PR tree-optimization/71563
41186 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41187 out of range value - has low bits known to be zero.
41189 2017-01-04 Alan Modra <amodra@gmail.com>
41191 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41192 * configure: Regenerate.
41193 * config.in: Regenerate.
41195 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41198 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41199 a substring of the message, but strcmp with the whole message. Ifdef
41200 ENABLE_NLS, translate the message first using dgettext.
41202 2017-01-03 Jeff Law <law@redhat.com>
41204 PR tree-optimizatin/78856
41205 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41206 (mark_threaded_blocks): Remove code to truncate thread paths that
41207 cross multiple loop headers. Instead invalidate the cached loop
41208 iteration information and handle case of a thread path walking
41209 into an irreducible region.
41211 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41214 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41215 assertions. Add support for doing the signbit if the IEEE 128-bit
41216 floating point value is in a GPR.
41217 * config/rs6000/rs6000.md (Fsignbit): Delete.
41218 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41219 Update the length attribute if the value is in a GPR.
41220 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41221 the sign or zero extension instruction, since the value is always 0/1.
41222 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41225 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41226 extracting SImode to a GPR register so that we can generate a
41227 store, limit the vector to be in a traditional Altivec register
41228 for the vextuwrx instruction.
41230 2017-01-03 Ian Lance Taylor <iant@google.com>
41232 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41234 2017-01-03 Martin Sebor <msebor@redhat.com>
41236 PR tree-optimization/78696
41237 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41238 precision. Use MPFR for %f for greater fidelity. Correct handling
41240 (pass_sprintf_length::compute_format_length): Set width and precision
41241 specified by asrerisk to void_node for vararg functions.
41242 (try_substitute_return_value): Adjust dump output.
41244 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41246 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41248 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41250 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41251 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41252 -mlra/-mno-lra was passed to the compiler.
41254 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41256 PR rtl-optimization/65618
41257 * emit-rtl.c (try_split): Move initialization of "before" and
41258 "after" to just before the call to emit_insn_after_setloc.
41260 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41262 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41264 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41266 * dwarf2out.c (gen_enumeration_type_die): When
41267 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41269 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41271 PR tree-optimization/78965
41272 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41273 Change first argument from const call_info & to call_info &. For %n
41274 set info.nowrite to false.
41276 PR middle-end/78901
41277 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41278 possibly throwing calls.
41280 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41281 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41282 and fns handling, rather than in a separate case SSA_NAME.
41284 2017-01-02 Jeff Law <law@redhat.com>
41286 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41287 fixes for first_period and second_period variables.
41289 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41292 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41293 (*insvqi_1): New insn pattern.
41294 (*insvqi_1_mem_rex64): Ditto.
41295 (*insvqi_2): Ditto.
41296 (*insvqi_3): Rename from *insvqi.
41298 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41300 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41302 * doc/cfg.texi (Edges): Remove reference to Java.
41303 (Maintaining the CFG): Ditto.
41305 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41307 PR middle-end/77674
41308 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41309 transparent aliases.
41311 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41313 PR middle-end/77484
41314 * predict.def (PRED_CALL): Update hitrate.
41315 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41316 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41317 into direct/indirect/polymorphic variants.
41319 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41321 Update copyright years.
41323 * gcc.c (process_command): Update copyright notice dates.
41324 * gcov-dump.c (print_version): Ditto.
41325 * gcov.c (print_version): Ditto.
41326 * gcov-tool.c (print_version): Ditto.
41327 * gengtype.c (create_file): Ditto.
41328 * doc/cpp.texi: Bump @copying's copyright year.
41329 * doc/cppinternals.texi: Ditto.
41330 * doc/gcc.texi: Ditto.
41331 * doc/gccint.texi: Ditto.
41332 * doc/gcov.texi: Ditto.
41333 * doc/install.texi: Ditto.
41334 * doc/invoke.texi: Ditto.
41336 Copyright (C) 2017 Free Software Foundation, Inc.
41338 Copying and distribution of this file, with or without modification,
41339 are permitted in any medium without royalty provided the copyright
41340 notice and this notice are preserved.