1 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
4 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
5 allowed debug stmts before labels.
6 (expand_gimple_basic_block): Likewise.
7 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
8 * gimple-iterator.h (gsi_after_labels): Likewise.
9 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
10 rename reused variable, and simplify using gsi_move_before.
11 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
12 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
13 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
14 (gimple_verify_flow_info, gimple_block_label): Likewise.
15 (make_blocks): Move debug markers after adjacent labels.
16 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
17 allowed debug insns outside blocks.
18 * df-scan.c (df_insn_delete): Likewise.
19 * lra-constraints.c (update_ebb_live_info): Likewise.
20 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
21 (vt_initialize, delete_vta_debug_insns): Likewise.
22 (reemit_marker_as_note): Drop BB parm. Adjust callers.
24 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * poly-int-types.h (round_down_to_byte_boundary): New macro.
29 (round_up_to_byte_boundary): Likewise.
30 * expr.h (get_bit_range): Add temporary shim.
31 * gimple-ssa-store-merging.c (store_operand_info): Change the
32 bitsize, bitpos, bitregion_start and bitregion_end fields from
33 unsigned HOST_WIDE_INT to poly_uint64.
34 (merged_store_group): Likewise load_align_base.
35 (compatible_load_p, compatible_load_p): Update accordingly.
36 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
37 (split_group, imm_store_chain_info::output_merged_store): Likewise.
38 (mem_valid_for_store_merging): Return the bitsize, bitpos,
39 bitregion_start and bitregion_end as poly_uint64s rather than
40 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
41 (handled_load): Take the bitsize, bitpos,
42 bitregion_start and bitregion_end as poly_uint64s rather than
43 unsigned HOST_WIDE_INTs.
44 (pass_store_merging::process_store): Update call to
45 mem_valid_for_store_merging.
47 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
48 Alan Hayward <alan.hayward@arm.com>
49 David Sherwood <david.sherwood@arm.com>
51 * builtins.c (get_object_alignment_2): Track polynomial offsets
52 and sizes. Update the alignment handling.
54 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
55 Alan Hayward <alan.hayward@arm.com>
56 David Sherwood <david.sherwood@arm.com>
58 * tree.h (get_inner_reference): Add a version that returns the
59 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
60 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
61 the case in which bitpos is not associated with the first byte.
63 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
64 Alan Hayward <alan.hayward@arm.com>
65 David Sherwood <david.sherwood@arm.com>
67 * tree-affine.h (get_inner_reference_aff): Return the size as a
69 * tree-affine.c (get_inner_reference_aff): Likewise.
70 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
71 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
73 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
74 Alan Hayward <alan.hayward@arm.com>
75 David Sherwood <david.sherwood@arm.com>
77 * fold-const.c (pointer_may_wrap_p): Take the offset as a
78 HOST_WIDE_INT rather than a poly_int64.
80 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
81 Alan Hayward <alan.hayward@arm.com>
82 David Sherwood <david.sherwood@arm.com>
84 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
85 HOST_WIDE_INT to poly_int64_pod.
86 (perform_symbolic_merge): Update accordingly.
87 (bswap_replace): Likewise.
89 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
90 Alan Hayward <alan.hayward@arm.com>
91 David Sherwood <david.sherwood@arm.com>
93 * tree-affine.h (aff_tree::offset): Change from widest_int
95 (wide_int_ext_for_comb): Delete.
96 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
97 constants as poly_widest_int rather than widest_int.
98 (aff_combination_constant_multiple_p): Return the multiplier
100 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
102 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
103 version static and add an overload for poly_widest_int.
104 (aff_combination_const, aff_combination_add_cst)
105 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
106 the constants as poly_widest_int rather than widest_int.
107 (tree_to_aff_combination): Generalize INTEGER_CST case to
109 (aff_combination_to_tree): Track offsets as poly_widest_ints.
110 (aff_combination_add_product, aff_combination_mult): Handle
112 (aff_combination_constant_multiple_p): Return the multiplier
113 as a poly_widest_int.
114 * tree-predcom.c (determine_offset): Return the offset as a
116 (split_data_refs_to_components, suitable_component_p): Update
118 (valid_initializer_p): Update call to
119 aff_combination_constant_multiple_p.
120 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
121 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
122 as a poly_int64 rather than a HOST_WIDE_INT.
123 (get_address_cost): Handle polynomial offsets.
124 (iv_elimination_compare_lt): Likewise.
125 (rewrite_use_nonlinear_expr): Likewise.
127 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
128 Alan Hayward <alan.hayward@arm.com>
129 David Sherwood <david.sherwood@arm.com>
131 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
132 as a poly_int64_pod rather than a HOST_WIDE_INT.
133 (get_addr_base_and_unit_offset): Likewise.
134 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
135 (get_addr_base_and_unit_offset): Likewise.
136 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
137 to poly_int64 in example.
138 * fold-const.c (fold_binary_loc): Update call to
139 get_addr_base_and_unit_offset.
140 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
141 (maybe_canonicalize_mem_ref_addr): Likewise.
142 (gimple_fold_stmt_to_constant_1): Likewise.
143 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
145 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
146 * match.pd: Likewise.
147 * omp-low.c (lower_omp_target): Likewise.
148 * tree-sra.c (build_ref_for_offset): Likewise.
149 (build_debug_ref_for_model): Likewise.
150 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
151 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
152 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
153 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
154 (constant_pointer_difference): Likewise.
155 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
156 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
157 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
158 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
159 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
160 (set_ssa_val_to): Likewise.
161 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
162 (maybe_diag_stxncpy_trunc): Likewise.
163 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
164 * tree.c (build_simple_mem_ref_loc): Likewise.
165 (array_at_struct_end_p): Likewise.
167 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
168 Alan Hayward <alan.hayward@arm.com>
169 David Sherwood <david.sherwood@arm.com>
171 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
172 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
173 (get_ref_base_and_extent_hwi): Declare.
174 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
175 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
176 (get_ref_base_and_extent_hwi): New function.
177 * cfgexpand.c (expand_debug_expr): Update call to
178 get_ref_base_and_extent.
179 * dwarf2out.c (add_var_loc_to_decl): Likewise.
180 * gimple-fold.c (get_base_constructor): Return the offset as a
181 poly_int64_pod rather than a HOST_WIDE_INT.
182 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
183 * ipa-polymorphic-call.c
184 (ipa_polymorphic_call_context::set_by_invariant)
185 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
186 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
187 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
188 rather than get_ref_base_and_extent.
189 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
190 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
191 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
193 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
194 call to get_ref_base_and_extent.
195 * tree-sra.c (create_access, get_access_for_expr): Likewise.
196 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
197 (stmt_kills_ref_p): Likewise.
198 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
199 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
201 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
202 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
203 when calling native_encode_expr.
204 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
205 call to get_ref_base_and_extent.
206 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
207 get_ref_base_and_extent.
208 * var-tracking.c (track_expr_p): Likewise.
210 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
211 Alan Hayward <alan.hayward@arm.com>
212 David Sherwood <david.sherwood@arm.com>
214 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
215 HOST_WIDE_INT to poly_int64_pod.
216 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
217 polynomail parameter offsets.
219 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
220 Alan Hayward <alan.hayward@arm.com>
221 David Sherwood <david.sherwood@arm.com>
223 * gengtype.c (main): Handle poly_int64_pod.
224 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
225 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
226 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
227 from HOST_WIDE_INT to poly_int64_pod.
228 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
229 (copy_cfa): New function.
230 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
231 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
232 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
233 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
234 (initial_return_save): Treat offsets as poly_ints.
235 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
236 offset is nonconstant.
237 (reg_save): Take the offset as a poly_int64. Fall back to
238 DW_CFA_expression for nonconstant offsets.
239 (queue_reg_save): Take the offset as a poly_int64.
240 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
242 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
243 Alan Hayward <alan.hayward@arm.com>
244 David Sherwood <david.sherwood@arm.com>
246 * rtl.h (operand_subword, operand_subword_force): Take the offset
247 as a poly_uint64 an unsigned int.
248 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
250 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
251 Alan Hayward <alan.hayward@arm.com>
252 David Sherwood <david.sherwood@arm.com>
254 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
255 'p' format code. Use INT_LIST rather than SUBREG as the example of
256 a code with an XINT and an XEXP. Remove the implication that
257 accessing an rtx field using XINT is expected to work.
258 * rtl.def (SUBREG): Change format from "ei" to "ep".
259 * rtl.h (rtunion::rt_subreg): New field.
260 (XCSUBREG): New macro.
261 (SUBREG_BYTE): Use it.
262 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
263 Update constructor accordingly.
264 (subreg_shape::operator ==): Update accordingly.
265 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
266 than an unsigned int.
267 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
268 a poly_uint64 rather than an unsigned int.
269 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
270 than an unsigned int.
271 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
272 (subreg_size_highpart_offset): Return a poly_uint64 rather than
273 an unsigned int. Take the sizes as poly_uint64s.
274 (subreg_offset_from_lsb): Return a poly_uint64 rather than
275 an unsigned int. Take the shift as a poly_uint64 rather than
277 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
278 as a poly_uint64 rather than an unsigned int.
279 (simplify_subreg_regno): Likewise.
280 (byte_lowpart_offset): Return the memory offset as a poly_int64
282 (subreg_memory_offset): Likewise. Take the subreg offset as a
283 poly_uint64 rather than an unsigned int.
284 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
285 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
286 poly_uint64 rather than an unsigned int.
287 * rtl.c (rtx_format): Describe 'p' in comment.
288 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
289 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
290 offset as a poly_uint64 rather than an unsigned int.
291 (byte_lowpart_offset): Return the memory offset as a poly_int64
293 (subreg_memory_offset): Likewise. Take the subreg offset as a
294 poly_uint64 rather than an unsigned int.
295 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
296 mode sizes as poly_uint64s rather than unsigned ints. Return a
297 poly_uint64 rather than an unsigned int.
298 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
299 (copy_insn_1): Handle 'p'.
300 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
301 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
302 an unsigned int. Return the shift in the same way.
303 (subreg_lsb): Return the shift as a poly_uint64 rather than an
305 (subreg_size_offset_from_lsb): Take the sizes and shift as
306 poly_uint64s rather than unsigned ints. Return the offset as
308 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
309 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
311 * rtlhash.c (add_rtx): Handle 'p'.
312 * genemit.c (gen_exp): Likewise.
313 * gengenrtl.c (type_from_format, gendef): Likewise.
314 * gensupport.c (subst_pattern_match, get_alternatives_number)
315 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
316 (subst_dup): Likewise.
317 * gengtype.c (adjust_field_rtx_def): Likewise.
318 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
319 (match_pattern_2): Likewise.
320 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
321 (rtx_test::subreg_field): New function.
322 (operator ==, safe_to_hoist_p, transition_parameter_type)
323 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
324 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
325 * genpeep.c (match_rtx): Likewise.
326 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
327 (rtx_writer::print_rtx_operand): Handle 'p'.
328 (print_value): Handle SUBREG.
329 * read-rtl.c (apply_int_iterator): Likewise.
330 (rtx_reader::read_rtx_operand): Handle 'p'.
331 * alias.c (rtx_equal_for_memref_p): Likewise.
332 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
333 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
335 * calls.c (expand_call): Likewise.
336 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
337 (make_extraction, gen_lowpart_for_combine): Likewise.
338 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
340 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
341 rather than an unsigned int. Treat subreg offsets as poly_ints.
342 (exp_equiv_p): Handle 'p'.
343 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
344 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
345 * dse.c (find_shift_sequence): Likewise.
346 * dwarf2out.c (rtl_for_decl_location): Likewise.
347 * expmed.c (extract_low_bits): Likewise.
348 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
349 (expand_expr_real_2): Likewise.
350 * final.c (alter_subreg): Likewise.
351 (leaf_renumber_regs_insn): Handle 'p'.
352 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
353 Treat subreg offsets as poly_ints.
354 * fwprop.c (forward_propagate_and_simplify): Likewise.
355 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
356 * ira.c (get_subreg_tracking_sizes): Likewise.
357 * ira-conflicts.c (go_through_subreg): Likewise.
358 * ira-lives.c (process_single_reg_class_operands): Likewise.
359 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
360 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
361 as a poly_uint64 rather than an unsigned int.
362 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
363 subreg offsets as poly_ints.
364 * lra-constraints.c (operands_match_p): Handle 'p'.
365 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
366 * lra-spills.c (assign_mem_slot): Likewise.
367 * postreload.c (move2add_valid_value_p): Likewise.
368 * recog.c (general_operand, indirect_operand): Likewise.
369 * regcprop.c (copy_value, maybe_mode_change): Likewise.
370 (copyprop_hardreg_forward_1): Likewise.
371 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
372 (record_subregs_of_mode): Likewise.
373 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
374 * reload.c (operands_match_p): Handle 'p'.
375 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
376 * reload1.c (alter_reg, choose_reload_regs): Likewise.
377 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
378 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
379 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
380 (simplify_const_poly_int_tests<N>::run): Likewise.
381 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
382 a poly_uint64 rather than an unsigned int.
383 * valtrack.c (debug_lowpart_subreg): Likewise.
384 * var-tracking.c (var_lowpart): Likewise.
385 (loc_cmp): Handle 'p'.
387 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
388 Alan Hayward <alan.hayward@arm.com>
389 David Sherwood <david.sherwood@arm.com>
391 * ira.c (get_subreg_tracking_sizes): New function.
392 (init_live_subregs): Take an integer size rather than a register.
393 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
394 to init_live_subregs.
396 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
397 Alan Hayward <alan.hayward@arm.com>
398 David Sherwood <david.sherwood@arm.com>
400 * expr.c (store_constructor_field): Change bitsize from a
401 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
402 HOST_WIDE_INT to a poly_int64.
403 (store_constructor): Change size from a HOST_WIDE_INT to
405 (store_field): Likewise bitsize and bitpos.
407 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
408 Alan Hayward <alan.hayward@arm.com>
409 David Sherwood <david.sherwood@arm.com>
411 * expmed.h (store_bit_field): Change bitregion_start and
412 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
413 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
414 (store_bit_field_1, store_integral_bit_field, store_bit_field)
415 (store_fixed_bit_field, store_split_bit_field): Likewise.
416 * expr.c (store_constructor_field, store_field): Likewise.
417 (optimize_bitfield_assignment_op): Likewise. Make the same change
418 to bitsize and bitpos.
419 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
420 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
421 same change in the constructor arguments.
422 (get_best_mode): Change bitregion_start and bitregion_end from
423 unsigned HOST_WIDE_INT to poly_uint64.
424 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
425 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
427 (bit_field_mode_iterator::next_mode): Update for new types
428 of m_bitregion_start and m_bitregion_end.
429 (get_best_mode): Change bitregion_start and bitregion_end from
430 unsigned HOST_WIDE_INT to poly_uint64.
432 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
433 Alan Hayward <alan.hayward@arm.com>
434 David Sherwood <david.sherwood@arm.com>
436 * rtl.h (simplify_gen_subreg): Add a temporary overload that
437 accepts poly_uint64 offsets.
438 * expmed.h (extract_bit_field): Take bitsize and bitnum as
439 poly_uint64s rather than unsigned HOST_WIDE_INTs.
440 * expmed.c (lowpart_bit_field_p): Likewise.
441 (extract_bit_field_as_subreg): New function, split out from...
442 (extract_bit_field_1): ...here. Take bitsize and bitnum as
443 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
444 extractions, check that BITSIZE matches the size of the extracted
445 value and that BITNUM is an exact multiple of that size.
446 If all else fails, try forcing the value into memory if
447 BITNUM is variable, and adjusting the address so that the
448 offset is constant. Split the part that can only handle constant
449 bitsize and bitnum out into...
450 (extract_integral_bit_field): ...this new function.
451 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
452 rather than unsigned HOST_WIDE_INTs.
454 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
455 Alan Hayward <alan.hayward@arm.com>
456 David Sherwood <david.sherwood@arm.com>
458 * expmed.h (store_bit_field): Take bitsize and bitnum as
459 poly_uint64s rather than unsigned HOST_WIDE_INTs.
460 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
461 that returns the byte size.
462 (store_bit_field_1): Take bitsize and bitnum as
463 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
464 to simple_mem_bitfield_p. Split the part that can only handle
465 constant bitsize and bitnum out into...
466 (store_integral_bit_field): ...this new function.
467 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
468 than unsigned HOST_WIDE_INTs.
469 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
471 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
472 Alan Hayward <alan.hayward@arm.com>
473 David Sherwood <david.sherwood@arm.com>
475 * lra-int.h (lra_reg): Change offset from int to poly_int64.
476 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
478 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
479 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
480 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
481 offset as a poly_int64 rather than an int.
482 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
483 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
484 * lra-constraints.c (equiv_address_substitution): Track offsets
486 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
487 (curr_insn_transform): Handle the new form of sp_offset.
488 * lra-eliminations.c (lra_elim_table): Change previous_offset
489 and offset from HOST_WIDE_INT to poly_int64.
490 (print_elim_table, update_reg_eliminate): Update accordingly.
491 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
492 (get_elimination): Update accordingly.
493 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
494 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
495 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
496 poly_int64 offsets generally.
497 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
498 (mark_not_eliminable, init_elimination): Update accordingly.
499 (remove_reg_equal_offset_note): Return a bool and pass the new
500 offset back by pointer as a poly_int64.
501 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
502 rather than a HOST_WIDE_INT.
503 (do_remat): Track offsets poly_int64s.
504 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
506 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
507 Alan Hayward <alan.hayward@arm.com>
508 David Sherwood <david.sherwood@arm.com>
510 * rtl.h (mem_attrs): Add a default constructor. Change size and
511 offset from HOST_WIDE_INT to poly_int64.
512 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
513 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
514 (widen_memory_access): Take the sizes and offsets as poly_int64s
515 rather than HOST_WIDE_INTs.
516 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
517 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
519 (adjust_offset_for_component_ref): Change the offset from a
520 HOST_WIDE_INT to a poly_int64.
521 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
522 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
523 * dce.c (find_call_stack_args): Likewise.
524 * dse.c (record_store): Likewise.
525 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
526 * print-rtl.c (rtx_writer::print_rtx): Likewise.
527 * read-rtl-function.c (test_loading_mem): Likewise.
528 * rtlanal.c (may_trap_p_1): Likewise.
529 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
530 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
531 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
532 (mem_attrs::mem_attrs): New function.
533 (set_mem_attributes_minus_bitpos): Change bitpos from a
534 HOST_WIDE_INT to poly_int64.
535 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
536 (clear_mem_offset, clear_mem_size, change_address)
537 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
538 initialize mem_attrs.
539 (set_mem_offset, set_mem_size, adjust_address_1)
540 (adjust_automodify_address_1, offset_address, widen_memory_access):
541 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
543 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
544 Alan Hayward <alan.hayward@arm.com>
545 David Sherwood <david.sherwood@arm.com>
547 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
548 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
549 rather than 0 to represent an unknown size. Assert that the size
550 is known when the mode isn't BLKmode.
551 (may_trap_p_1): Use -1 for unknown sizes.
552 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
554 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
555 Alan Hayward <alan.hayward@arm.com>
556 David Sherwood <david.sherwood@arm.com>
558 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
559 to poly_int64. Update commentary for positions_needed.large.
560 (read_info_type): Change offset and width from HOST_WIDE_INT
562 (set_usage_bits): Likewise.
563 (canon_address): Return the offset as a poly_int64 rather than
564 a HOST_WIDE_INT. Use strip_offset_and_add.
565 (set_all_positions_unneeded, any_positions_needed_p): Use
566 positions_needed.large to track stores with non-constant widths.
567 (all_positions_needed_p): Likewise. Take the offset and width
568 as poly_int64s rather than ints. Assert that rhs is nonnull.
569 (record_store): Cope with non-constant offsets and widths.
570 Nullify the rhs of an earlier store if we can't tell which bytes
572 (find_shift_sequence): Take the access_size and shift as poly_int64s
574 (get_stored_val): Take the read_offset and read_width as poly_int64s
575 rather than HOST_WIDE_INTs.
576 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
577 non-constant offsets and widths.
579 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
580 Alan Hayward <alan.hayward@arm.com>
581 David Sherwood <david.sherwood@arm.com>
583 * inchash.h (inchash::hash::add_poly_int): New function.
584 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
585 Use poly_int64 rather than HOST_WIDE_INT.
586 (ao_ref::max_size_known_p): New function.
587 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
588 rather than HOST_WIDE_INT.
589 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
590 to temporaries until its interface is adjusted to match.
591 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
592 (aliasing_component_refs_p, decl_refs_may_alias_p)
593 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
594 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
595 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
597 * alias.c (ao_ref_from_mem): Likewise.
598 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
599 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
600 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
601 (maybe_trim_complex_store, maybe_trim_constructor_store)
602 (live_bytes_read, dse_classify_store): Likewise.
603 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
604 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
605 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
606 (vn_reference_lookup_3): Likewise.
607 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
609 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
610 Alan Hayward <alan.hayward@arm.com>
611 David Sherwood <david.sherwood@arm.com>
613 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
614 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
615 instead of ranges_overlap_p.
617 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
618 Alan Hayward <alan.hayward@arm.com>
619 David Sherwood <david.sherwood@arm.com>
621 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
622 sizes as poly_int64s rather than HOST_WIDE_INTs.
624 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
625 Alan Hayward <alan.hayward@arm.com>
626 David Sherwood <david.sherwood@arm.com>
628 * gimple-fold.h (fold_ctor_reference): Take the offset and size
629 as poly_uint64 rather than unsigned HOST_WIDE_INT.
630 * gimple-fold.c (fold_ctor_reference): Likewise.
632 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
633 Alan Hayward <alan.hayward@arm.com>
634 David Sherwood <david.sherwood@arm.com>
636 * target.def (dwarf_poly_indeterminate_value): New hook.
637 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
638 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
639 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
640 * doc/tm.texi: Regenerate.
641 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
642 offset as a poly_int64.
643 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
644 offset as a poly_int64.
645 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
646 Take the offset as a poly_int64.
647 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
648 (frame_pointer_fb_offset): Change to a poly_int64.
649 (int_loc_descriptor): Take the offset as a poly_int64. Use
650 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
651 (based_loc_descr): Take the offset as a poly_int64.
652 Use strip_offset_and_add to handle (plus X (const)).
653 Use new_reg_loc_descr instead of an open-coded version of the
654 previous implementation.
655 (mem_loc_descriptor): Handle CONST_POLY_INT.
656 (compute_frame_pointer_to_fb_displacement): Take the offset as a
657 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
659 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
660 Alan Hayward <alan.hayward@arm.com>
661 David Sherwood <david.sherwood@arm.com>
663 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
664 (gen_rtx_REG_offset): Take the offset as a poly_int64.
665 * inchash.h (inchash::hash::add_poly_hwi): New function.
666 * gengtype.c (main): Register poly_int64.
667 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
668 offset as a poly_int.
669 (reg_attr_hasher::equal): Use must_eq to compare offsets.
670 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
671 offset as a poly_int64.
672 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
673 * print-rtl.c (print_poly_int): New function.
674 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
676 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
678 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
679 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
680 (vt_get_decl_and_offset): Return the offset as a poly_int64.
681 Enforce track_offset_p for parts of a PARALLEL.
682 (vt_add_function_parameter): Use const_offset for the final
683 offset to track. Use get_tracked_reg_offset for the parts
686 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
687 Alan Hayward <alan.hayward@arm.com>
688 David Sherwood <david.sherwood@arm.com>
690 * target.def (truly_noop_truncation): Take poly_uint64s instead of
691 unsigned ints. Change default to hook_bool_puint64_puint64_true.
692 * doc/tm.texi: Regenerate.
693 * hooks.h (hook_bool_uint_uint_true): Delete.
694 (hook_bool_puint64_puint64_true): Declare.
695 * hooks.c (hook_bool_uint_uint_true): Delete.
696 (hook_bool_puint64_puint64_true): New function.
697 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
698 instead of unsigned ints.
699 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
700 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
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 * optabs.h (expand_operand): Add an int_value field.
707 (create_expand_operand): Add an int_value parameter and use it
708 to initialize the new expand_operand field.
709 (create_integer_operand): Replace with a declaration of a function
710 that accepts poly_int64s. Move the implementation to...
711 * optabs.c (create_integer_operand): ...here.
712 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
713 the mode preserves the value of int_value, instead of calling
714 const_int_operand on the rtx. Use gen_int_mode to generate
717 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
718 Alan Hayward <alan.hayward@arm.com>
719 David Sherwood <david.sherwood@arm.com>
721 * dumpfile.h (dump_dec): Declare.
722 * dumpfile.c (dump_dec): New function.
723 * pretty-print.h (pp_wide_integer): Turn into a function and
724 declare a poly_int version.
725 * pretty-print.c (pp_wide_integer): New function for poly_ints.
727 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
728 Alan Hayward <alan.hayward@arm.com>
729 David Sherwood <david.sherwood@arm.com>
731 * doc/generic.texi (POLY_INT_CST): Document.
732 * tree.def (POLY_INT_CST): New tree code.
733 * treestruct.def (TS_POLY_INT_CST): New tree layout.
734 * tree-core.h (tree_poly_int_cst): New struct.
735 (tree_node): Add a poly_int_cst field.
736 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
737 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
738 instead of a wide_int_ref.
739 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
741 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
742 instead of an unsigned HOST_WIDE_INT.
743 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
744 (ptrdiff_tree_p): Declare.
745 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
746 extern inline implementations if the target doesn't use POLY_INT_CST.
747 (poly_int_tree_p): New function.
748 (wi::unextended_tree): New class.
749 (wi::int_traits <unextended_tree>): New override.
750 (wi::extended_tree): Add a default constructor.
751 (wi::extended_tree::get_tree): New function.
752 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
753 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
754 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
755 (wi::tree_to_poly_wide_ref): New typedefs.
756 (wi::ints_for): Provide overloads for extended_tree and
758 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
759 (wi::to_wide): New functions.
760 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
761 * tree.c (poly_int_cst_hasher): New struct.
762 (poly_int_cst_hash_table): New variable.
763 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
764 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
766 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
767 (init_ttree): Initialize poly_int_cst_hash_table.
768 (build_int_cst, build_int_cst_type, build_invariant_address): Take
769 a poly_int64 instead of a HOST_WIDE_INT.
770 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
771 instead of an unsigned HOST_WIDE_INT.
772 (wide_int_to_tree): Rename to...
773 (wide_int_to_tree_1): ...this.
774 (build_new_poly_int_cst, build_poly_int_cst): New functions.
775 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
776 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
777 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
778 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
779 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
781 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
782 (streamer_read_tree_body): Likewise.
783 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
784 (streamer_write_tree_body): Likewise.
785 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
786 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
787 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
788 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
789 * gimple-expr.h (is_gimple_constant): Likewise.
790 * gimplify.c (maybe_with_size_expr): Likewise.
791 * print-tree.c (print_node): Likewise.
792 * tree-data-ref.c (data_ref_compare_tree): Likewise.
793 * tree-pretty-print.c (dump_generic_node): Likewise.
794 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
795 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
796 * tree-vrp.c (compare_values_warnv): Likewise.
797 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
798 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
799 (force_expr_to_var_cost): Likewise.
800 * tree-ssa-loop.c (for_each_index): Likewise.
801 * fold-const.h (build_invariant_address, size_int_kind): Take a
802 poly_int64 instead of a HOST_WIDE_INT.
803 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
804 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
806 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
807 (int_const_binop_2): New function, split out from...
808 (int_const_binop_1): ...here. Handle POLY_INT_CST.
809 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
810 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
811 * gimple-ssa-strength-reduction.c (slsr_process_add)
812 (slsr_process_mul): Check for INTEGER_CSTs before using them
814 * stor-layout.c (bits_from_bytes): New function.
815 (bit_from_pos): Use it.
816 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
817 by BITS_PER_UNIT to get the TYPE_SIZE.
818 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
819 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
821 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
822 Alan Hayward <alan.hayward@arm.com>
823 David Sherwood <david.sherwood@arm.com>
825 * doc/rtl.texi (const_poly_int): Document. Also document the
826 rtl sharing behavior.
827 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
828 * rtl.h (const_poly_int_def): New struct.
829 (rtx_def::u): Add a cpi field.
830 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
831 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
832 (wi::rtx_to_poly_wide_ref): New typedef
833 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
834 (poly_int_rtx_p): New functions.
835 (trunc_int_for_mode): Declare a poly_int64 version.
836 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
837 (immed_wide_int_const): Take a poly_wide_int_ref rather than
839 (strip_offset): Declare.
840 (strip_offset_and_add): New function.
841 * rtl.def (CONST_POLY_INT): New rtx code.
842 * rtl.c (rtx_size): Handle CONST_POLY_INT.
843 (shared_const_p): Use poly_int_rtx_p.
844 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
846 (gen_int_shift_amount): Likewise.
847 * emit-rtl.c (const_poly_int_hasher): New class.
848 (const_poly_int_htab): New variable.
849 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
850 (const_poly_int_hasher::hash): New function.
851 (const_poly_int_hasher::equal): Likewise.
852 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
853 (immed_wide_int_const): Rename to...
854 (immed_wide_int_const_1): ...this and make static.
855 (immed_wide_int_const): New function, taking a poly_wide_int_ref
856 instead of a wide_int_ref.
857 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
858 (gen_lowpart_common): Handle CONST_POLY_INT.
859 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
860 * cselib.c (cselib_hash_rtx): Likewise.
861 * dwarf2out.c (const_ok_for_output_1): Likewise.
862 * expr.c (convert_modes): Likewise.
863 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
864 * rtlhash.c (add_rtx): Likewise.
865 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
866 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
867 Handle existing CONST_POLY_INT rtxes.
868 * expmed.h (expand_shift): Take a poly_int64 instead of a
870 * expmed.c (expand_shift): Likewise.
871 * rtlanal.c (strip_offset): New function.
872 (commutative_operand_precedence): Give CONST_POLY_INT the same
873 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
875 * rtl-tests.c (const_poly_int_tests): New struct.
876 (rtl_tests_c_tests): Use it.
877 * simplify-rtx.c (simplify_const_unary_operation): Handle
879 (simplify_const_binary_operation): Likewise.
880 (simplify_binary_operation_1): Fold additions of symbolic constants
882 (simplify_subreg): Handle extensions and truncations of
884 (simplify_const_poly_int_tests): New struct.
885 (simplify_rtx_c_tests): Use it.
886 * wide-int.h (storage_ref): Add default constructor.
887 (wide_int_ref_storage): Likewise.
888 (trailing_wide_ints): Use GTY((user)).
889 (trailing_wide_ints::operator[]): Add a const version.
890 (trailing_wide_ints::get_precision): New function.
891 (trailing_wide_ints::extra_size): Likewise.
893 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
894 Alan Hayward <alan.hayward@arm.com>
895 David Sherwood <david.sherwood@arm.com>
897 * emit-rtl.h (gen_int_shift_amount): Declare.
898 * emit-rtl.c (gen_int_shift_amount): New function.
899 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
901 * calls.c (shift_return_value): Likewise.
902 * cse.c (fold_rtx): Likewise.
903 * dse.c (find_shift_sequence): Likewise.
904 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
905 (expand_shift, expand_smod_pow2): Likewise.
906 * lower-subreg.c (shift_cost): Likewise.
907 * optabs.c (expand_superword_shift, expand_doubleword_mult)
908 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
909 (expand_vec_perm_var): Likewise.
910 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
911 (simplify_binary_operation_1): Likewise.
912 * combine.c (try_combine, find_split_point, force_int_to_mode)
913 (simplify_shift_const_1, simplify_shift_const): Likewise.
914 (change_zero_ext): Likewise. Use simplify_gen_binary.
916 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
918 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
920 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
922 * doc/invoke.texi (ARM Options): Document accepted extension options
923 for -march=armv8.3-a.
925 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
928 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
929 or --with-float=softfp, set the default CPU to arm10e.
931 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
933 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
934 * config/visium/predicates.md (const_shift_operand): Likewise.
935 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
936 (visium_legitimize_reload_address): Likewise.
938 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
941 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
943 2017-12-20 Jakub Jelinek <jakub@redhat.com>
946 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
948 2017-12-20 Julia Koval <julia.koval@intel.com>
950 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
951 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
952 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
953 * config.gcc: Include vpclmulqdqintrin.h.
954 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
955 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
957 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
958 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
959 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
960 * config/i386/i386.c (isa2_opts): Add -mcx16.
961 (isa_opts): Add -mpclmulqdq, remove -mcx16.
962 (ix86_option_override_internal): Move mcx16 to flags2.
963 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
964 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
965 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
966 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
967 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
968 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
969 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
970 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
971 * doc/invoke.texi: Add -mvpclmulqdq.
973 2017-12-20 Tom de Vries <tom@codesourcery.com>
976 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
977 * calls.c (rtx_for_static_chain): ... here. New function.
978 * calls.h (rtx_for_static_chain): Declare.
979 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
980 instead of targetm.calls.static_chain.
981 * df-scan.c (df_get_entry_block_def_set): Same.
983 2017-12-19 Tom de Vries <tom@codesourcery.com>
985 PR tree-optimization/83493
986 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
989 2017-12-19 Jakub Jelinek <jakub@redhat.com>
991 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
992 inform with hardcoded english plural handling.
994 2017-12-18 Jeff Law <law@redhat.com>
996 PR tree-optimization/83477
997 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
998 a non-virtual PHI, always push a new range.
1000 2017-12-19 Martin Sebor <msebor@redhat.com>
1003 * builtins.c (compute_objsize): Handle non-constant offsets.
1005 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1007 PR tree-optimization/83444
1008 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1009 character load case, if get_stridx on MEM_REF's operand doesn't
1010 look usable, retry with get_addr_stridx.
1012 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1015 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1016 (variable_tracking_main_1): Keep markers even when VTA fails.
1019 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1020 even if there are markers before it.
1021 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1023 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1025 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1026 typical order conditions.
1027 * sel-sched.c (extract_new_fences_from): Likewise.
1028 * config/visium/constraints.md (J, K, L): Likewise.
1029 * config/visium/predicates.md (const_shift_operand): Likewise.
1030 * config/visium/visium.c (visium_legitimize_address,
1031 visium_legitimize_reload_address): Likewise.
1032 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1033 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1034 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1035 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1036 SET_DUMP_DETAIL): Likewise.
1037 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1038 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1039 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1040 avr_set_core_architecture, avr_set_current_function,
1041 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1042 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1043 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1044 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1045 avr_map_decompose, avr_fold_builtin): Likewise.
1046 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1047 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1048 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1049 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1050 * config/m32c/m32c.c (m32c_conditional_register_usage,
1051 m32c_address_cost): Likewise.
1052 * config/m32c/predicates.md (shiftcount_operand,
1053 longshiftcount_operand): Likewise.
1054 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1055 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1056 can_use_cdx_ldstw): Likewise.
1057 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1058 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1060 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1061 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1062 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1063 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1064 * config/vax/vax.c (adjacent_operands_p): Likewise.
1065 * config/ft32/constraints.md (L, b, KA): Likewise.
1066 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1068 * cfgexpand.c (expand_stack_alignment): Likewise.
1069 * gcse.c (insert_expr_in_table): Likewise.
1070 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1071 * cgraphunit.c (cgraph_node::expand): Likewise.
1072 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1073 * emit-rtl.c (add_insn): Likewise.
1074 * input.c (dump_location_info): Likewise.
1075 * passes.c (NEXT_PASS): Likewise.
1076 * read-rtl-function.c (parse_note_insn_name,
1077 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1079 * sched-rgn.c (sched_rgn_init): Likewise.
1080 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1081 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1082 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1083 * explow.c (eliminate_constant_term): Likewise.
1084 * final.c (leaf_renumber_regs_insn): Likewise.
1085 * cfgrtl.c (print_rtl_with_bb): Likewise.
1086 * genhooks.c (emit_init_macros): Likewise.
1087 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1088 * tree-data-ref.c (conflict_fn): Likewise.
1089 * selftest.c (assert_streq): Likewise.
1090 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1091 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1092 fold_binary_loc, multiple_of_p): Likewise.
1093 * reload.c (push_reload, find_equiv_reg): Likewise.
1094 * et-forest.c (et_nca, et_below): Likewise.
1095 * dbxout.c (dbxout_symbol_location): Likewise.
1096 * reorg.c (relax_delay_slots): Likewise.
1097 * dojump.c (do_compare_rtx_and_jump): Likewise.
1098 * gengtype-parse.c (type): Likewise.
1099 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1100 simplify_const_relational_operation): Likewise.
1101 * reload1.c (do_output_reload): Likewise.
1102 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1103 * gengtype.c (type_for_name): Likewise.
1104 * gimple-ssa-sprintf.c (format_directive): Likewise.
1106 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1109 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1110 accessing it. Adjust comment.
1112 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1115 * predict.c (zero_one_minusone): New function.
1116 (apply_return_prediction): Avoid return prediction for functions
1117 returning only -1, 0 and 1 values, unless they only return -1 and 0
1120 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1122 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1123 fall-through warning.
1124 (arc_compute_frame_size): Remove unused variables.
1125 (arc_print_operand): Fix fprintif format.
1126 (arc_can_follow_jump): Clean fall-through warning.
1128 2017-12-19 Marek Polacek <polacek@redhat.com>
1131 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1134 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1136 * config/arc/arc.c (overriderregs): New variable.
1137 (arc_override_options): Track fixed/call saved/call options.
1138 (arc_conditional_register_usage): Check against overrideregs
1139 variable whenever we change register properties.
1141 2017-12-19 Nathan Sidwell <nathan@acm.org>
1143 * opts.c (finish_options): Don't prefix dump_base_name if it
1144 already contains directories.
1146 2017-12-19 Martin Liska <mliska@suse.cz>
1148 PR rtl-optimization/82675
1149 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1150 more element in sbitmap.
1152 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1154 * gimplify.c (gimplify_expr): Use error_operand_p.
1156 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1159 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1162 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1164 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1165 the jump is to a label.
1167 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1169 PR tree-optimization/83444
1170 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1175 * ipa-inline.c (flatten_remove_node_hook): New function.
1176 (ipa_inline): Keep only nodes with flatten attribute at the end of
1177 the array in the order from ipa_reverse_postorder, only walk that
1178 portion of array for flattening, if there is more than one such
1179 node, temporarily register a removal hook and ignore removed nodes.
1181 PR tree-optimization/80631
1182 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1183 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1184 IFN_REDUC_MAX or IFN_REDUC_MIN.
1186 2017-12-18 Jeff Law <law@redhat.com>
1188 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1189 record symbolic equivalences from backedges in the CFG.
1192 2017-11-19 Jeff Law <law@redhat.com>
1194 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1195 of degenerates resulting from ignoring an edge.
1197 2017-12-18 Martin Sebor <msebor@redhat.com>
1200 PR tree-optimization/78450
1201 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1202 (handle_builtin_strlen): Call it.
1204 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1206 PR rtl-optimization/83424
1207 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1209 2017-12-18 Marek Polacek <polacek@redhat.com>
1212 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1213 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1216 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1218 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1219 (umaddsidi4, umaddsidi_split): Likewise.
1221 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1223 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1226 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1229 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1231 2017-12-18 Richard Biener <rguenther@suse.de>
1233 PR tree-optimization/81877
1234 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1235 (outermost_indep_loop): Adjust.
1236 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1237 (can_sm_ref_p): Adjust.
1239 2017-12-18 Richard Biener <rguenther@suse.de>
1242 * tree.c (array_at_struct_end_p): Return true if the underlying
1243 object has space for at least one element in excess of what
1244 the array domain specifies.
1246 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1248 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1249 example to avoid overfull hbox.
1250 * doc/invoke.texi (Option Summary): Add missing @gol.
1251 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1254 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1255 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1257 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1258 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1259 (AArch64 Options): Document -mverbose-cost-dump.
1260 (ARM Options): Likewise, plus -mflip-thumb.
1262 2017-12-17 Martin Sebor <msebor@redhat.com>
1265 * gimple-ssa-warn-restrict.c
1266 (builtin_memref::offset_out_of_bounds): Correct the handling of
1269 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1271 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1273 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1275 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1277 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1279 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1282 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1284 * doc/invoke.texi: Fix some typos.
1286 2017-12-16 Martin Sebor <msebor@redhat.com>
1288 PR tree-optimization/78918
1289 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1290 * builtins.c (check_sizes): Rename...
1291 (check_access): ...to this. Rename function arguments for clarity.
1292 (check_memop_sizes): Adjust names.
1293 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1294 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1295 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1296 (check_strncat_sizes, expand_builtin_strncat): Same.
1297 (expand_builtin_strncpy, expand_builtin_memset): Same.
1298 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1299 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1300 (maybe_emit_sprintf_chk_warning): Same.
1301 (expand_builtin_strcpy): Adjust.
1302 (expand_builtin_stpcpy): Same.
1303 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1304 in pointer-checking forms of memcpy, memmove, and mempcpy.
1305 (gcall_to_tree_minimal, max_object_size): Define new functions.
1306 * builtins.h (max_object_size): Declare.
1307 * calls.c (alloc_max_size): Call max_object_size instead of
1308 hardcoding ssizetype limit.
1309 (get_size_range): Handle new argument.
1310 * calls.h (get_size_range): Add a new argument.
1311 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1312 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1313 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1315 (gimple_fold_builtin_memory_chk): Same.
1316 (gimple_fold_builtin_stxcpy_chk): New function.
1317 * gimple-ssa-warn-restrict.c: New source.
1318 * gimple-ssa-warn-restrict.h: New header.
1319 * gimple.c (gimple_build_call_from_tree): Propagate location.
1320 * passes.def (pass_warn_restrict): Add new pass.
1321 * tree-pass.h (make_pass_warn_restrict): Declare.
1322 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1324 (handle_builtin_strcat): Same.
1325 (strlen_optimize_stmt): Rename...
1326 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1327 stpncpy, strncpy, and their checking forms.
1329 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1331 PR rtl-optimization/82849
1332 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1333 and get_max_loop_iterations_int.
1335 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1340 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1341 size as a poly_uint64.
1342 (mode_for_vector, mode_for_int_vector): Take the number of vector
1343 elements as a poly_uint64.
1344 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1345 size as a poly_uint64.
1346 (mode_for_vector, mode_for_int_vector): Take the number of vector
1347 elements as a poly_uint64.
1349 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1350 Alan Hayward <alan.hayward@arm.com>
1351 David Sherwood <david.sherwood@arm.com>
1353 * machmode.h (MACRO_MODE): New macro.
1354 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1355 * targhooks.c (default_libcall_value, default_secondary_reload)
1356 (default_memory_move_cost, default_register_move_cost)
1357 (default_class_max_nregs): Likewise.
1359 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1364 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1366 * genautomata.c (main): Likewise.
1367 * genconditions.c (write_header): Likewise.
1368 * genemit.c (main): Likewise.
1369 * genextract.c (print_header): Likewise.
1370 * genopinit.c (main): Likewise.
1371 * genoutput.c (output_prologue): Likewise.
1372 * genpeep.c (main): Likewise.
1373 * genpreds.c (write_insn_preds_c): Likewise.
1374 * genrecog.c (writer_header): Likewise.
1375 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1376 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1377 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1378 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1379 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1380 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1381 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1382 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1383 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1384 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1385 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1386 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1387 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1388 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1389 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1390 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1391 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1392 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1393 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1394 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1395 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1396 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1397 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1398 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1399 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1400 * config/darwin.c (IN_TARGET_CODE): Likewise.
1401 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1402 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1403 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1404 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1405 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1406 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1407 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1408 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1409 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1410 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1411 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1412 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1413 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1414 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1415 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1416 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1417 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1418 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1419 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1420 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1421 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1422 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1423 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1424 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1425 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1426 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1427 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1428 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1429 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1430 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1431 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1432 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1433 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1434 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1435 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1436 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1437 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1438 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1439 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1440 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1441 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1442 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1443 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1444 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1445 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1446 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1447 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1448 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1449 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1450 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1451 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1452 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1453 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1454 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1455 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1456 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1457 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1458 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1459 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1460 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1461 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1462 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1463 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1464 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1465 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1466 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1467 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1468 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1469 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1470 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1471 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1472 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1473 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1474 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1475 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1476 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1477 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1478 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1479 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1480 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1481 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1482 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1483 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1484 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1485 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1486 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1487 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1488 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1489 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1490 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1491 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1492 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1493 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1494 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1495 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1496 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1497 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1498 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1499 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1500 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1501 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1502 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1503 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1504 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1505 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1506 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1508 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1509 Alan Hayward <alan.hayward@arm.com>
1510 David Sherwood <david.sherwood@arm.com>
1512 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1515 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1520 * doc/md.texi (vec_series@var{m}): Document.
1521 * tree.def (VEC_SERIES_EXPR): New tree code.
1522 * tree.h (build_vec_series): Declare.
1523 * tree.c (build_vec_series): New function.
1524 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1525 * tree-pretty-print.c (dump_generic_node): Likewise.
1526 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1527 * tree-inline.c (estimate_operator_cost): Likewise.
1528 * expr.c (expand_expr_real_2): Likewise.
1529 * optabs-tree.c (optab_for_tree_code): Likewise.
1530 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1531 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1532 * expmed.c (make_tree): Handle VEC_SERIES.
1533 * optabs.def (vec_series_optab): New optab.
1534 * optabs.h (expand_vec_series_expr): Declare.
1535 * optabs.c (expand_vec_series_expr): New function.
1536 * tree-vect-generic.c (expand_vector_operations_1): Check that
1537 the operands also have vector type.
1539 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1540 Alan Hayward <alan.hawyard@arm.com>
1541 David Sherwood <david.sherwood@arm.com>
1543 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1544 (VEC_COND_EXPR): Add missing @tindex.
1545 * doc/md.texi (vec_duplicate@var{m}): Document.
1546 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1547 * tree.c (build_vector_from_val): Add stubbed-out handling of
1548 variable-length vectors, using VEC_DUPLICATE_EXPR.
1549 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1550 * cfgexpand.c (expand_debug_expr): Likewise.
1551 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1552 * tree-inline.c (estimate_operator_cost): Likewise.
1553 * tree-pretty-print.c (dump_generic_node): Likewise.
1554 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1555 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1556 (test_vec_duplicate_folding): New function.
1557 (fold_const_c_tests): Call it.
1558 * optabs.def (vec_duplicate_optab): New optab.
1559 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1560 * optabs.h (expand_vector_broadcast): Declare.
1561 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1562 vec_duplicate_optab.
1563 * expr.c (store_constructor): Try using vec_duplicate_optab for
1565 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1567 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1570 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1571 div/mod latencies a bit.
1573 2017-12-15 Jeff Law <law@redhat.com>
1575 PR tree-optimization/36550
1576 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1577 (mark_threaded_blocks): Rewrite code to avoid block copying when
1578 optimizing for size. Don't pessimize blocks which will be
1579 copied, but all the statements will be dead.
1581 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1583 PR tree-optimization/81165
1584 * tree-ssa-threadupdate.c (uses_in_bb): New.
1585 (estimate_threading_killed_stmts): New.
1586 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1587 * tree-ssa-threadedge.c
1588 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1591 2017-12-15 Jeff Law <law@redhat.com>
1593 PR tree-optimization/83410
1594 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1595 threads when parallelizing loops.
1597 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1599 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1601 * config/alpha/alpha.c (vms_attribute_table): Swap
1602 affects_type_identity and handler fields, adjust comments.
1603 * config/mips/mips.c (mips_attribute_table): Likewise.
1604 * config/visium/visium.c (visium_attribute_table): Likewise.
1605 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1606 * config/microblaze/microblaze.c (microblaze_attribute_table):
1608 * config/spu/spu.c (spu_attribute_table): Likewise.
1609 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1610 * config/arc/arc.c (arc_attribute_table): Likewise.
1611 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1612 * config/v850/v850.c (v850_handle_interrupt_attribute,
1613 v850_handle_data_area_attribute): Formatting fixes.
1614 (v850_attribute_table): Swap affects_type_identity and handler
1615 fields, adjust comments.
1616 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1617 * config/arm/arm.c (arm_attribute_table): Likewise.
1618 * config/avr/avr.c (avr_attribute_table): Likewise.
1619 * config/s390/s390.c (s390_attribute_table): Likewise.
1620 * config/sh/sh.c (sh_attribute_table): Likewise.
1621 * config/i386/i386.c (ix86_handle_cconv_attribute,
1622 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1623 (ix86_attribute_table): Swap affects_type_identity and handler
1624 fields, adjust comments.
1625 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1626 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1627 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1628 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1629 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1630 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1631 * config/rx/rx.c (rx_attribute_table): Likewise.
1632 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1633 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1634 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1635 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1636 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1637 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1638 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1639 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1640 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1641 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1642 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1644 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1646 (register_scoped_attributes, decl_attributes): Formatting fixes.
1648 PR tree-optimization/83269
1649 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1650 subtraction in arg0's type if type is signed and arg0 is unsigned.
1654 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1655 convert? on @0 instead of convert. Check type of @1, not @0.
1656 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1657 convert? on @0 instead of convert. Check type of @1, not @0.
1658 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1659 only optimize if either both @1 and @2 types are narrower
1660 precision, or both are wider or equal precision, and in the former
1661 case only if both have undefined overflow.
1663 2017-12-15 Richard Biener <rguenther@suse.de>
1666 * internal-fn.def (IFN_NOP): Add.
1667 * internal-fn.c (expand_NOP): Do nothing.
1668 * lto-streamer-in.c (input_function): Instead of removing
1669 sanitizer calls replace them with IFN_NOP calls.
1671 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1672 Alan Hayward <alan.hayward@arm.com>
1673 David Sherwood <david.sherwood@arm.com>
1675 * dse.c (store_info, read_info_type): Replace begin and end with
1677 (print_range): New function.
1678 (set_all_positions_unneeded, any_positions_needed_p)
1679 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1681 (record_store): Likewise. Optimize the case in which all positions
1683 (get_stored_val): Replace read_begin and read_end with read_offset
1685 (replace_read): Update call accordingly.
1687 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1689 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1690 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1691 (loop_cand::loop_cand): Initialize above members.
1692 (loop_cand::supported_operations): Delete.
1693 (loop_cand::can_interchange_p): Inline above function.
1694 (loop_cand::classify_simple_reduction): Record number of constant
1695 initialized simple reductions.
1696 (should_interchange_loops): New parameters. Check stmt cost of loops
1698 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1699 Update call to should_interchange_loops.
1700 (should_interchange_loop_nest): Update call to
1701 should_interchange_loops.
1703 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1706 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1707 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1708 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1710 2017-12-15 Julia Koval <julia.koval@intel.com>
1712 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1713 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1714 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1715 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1716 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1718 2017-12-15 Julia Koval <julia.koval@intel.com>
1720 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1721 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1722 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1723 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1724 _mm_aesenc_epi128): New intrinsics.
1726 2017-12-15 Julia Koval <julia.koval@intel.com>
1728 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1729 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1730 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1731 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1732 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1734 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1736 * gimple-ssa-strength-reduction.c (analyze_increments):
1737 Distinguish replacement costs for constant strides from those for
1740 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1742 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1744 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1746 * doc/invoke.texi: Document -Wcast-function-type.
1747 * recog.h (stored_funcptr): Change signature.
1748 * tree-dump.c (dump_node): Avoid warning.
1749 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1751 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1754 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1757 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1759 PR tree-optimization/83312
1760 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1761 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1762 "val" param, and to cope with arbitrary basic blocks.
1763 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1764 handle NULL_TREE for "val", dropping "bb" param.
1765 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1766 drop "bb" param. Handle NULL_TREE for "val".
1767 (find_case_label_for_value): Make "switch_stmt" param const.
1768 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1770 (vrp_prop::check_all_array_refs): Reimplement as...
1771 (check_array_bounds_dom_walker::before_dom_children): ...this new
1772 vfunc. Replace linear search through BB block list, excluding
1773 those with non-executable in-edges via dominator walk.
1775 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1777 * config/arm/arm.opt (mverbose-cost-dump): New option.
1778 * config/arm/arm.c (arm_rtx_costs): Use it.
1780 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1783 * reload1.c (emit_input_reload_insns): Skip debug markers.
1785 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1787 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1791 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1792 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1793 * config/sh/sh.c: Skip debug insns besides notes.
1794 * config/sh/sh.md: Likewise.
1795 * config/sh/sh_treg_combine.cc: Likewise.
1796 * config/sh/sync.md: Likewise.
1798 2017-12-14 Tom de Vries <tom@codesourcery.com>
1800 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1801 Add item for weak_undefined.
1803 2017-12-14 Richard Biener <rguenther@suse.de>
1805 PR tree-optimization/67842
1806 * sese.h (bb_in_region): Remove #if 0'ed code.
1808 2017-12-14 Richard Biener <rguenther@suse.de>
1810 PR tree-optimization/83326
1811 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1812 may_be_zero parameter and handle it by not marking the first
1813 peeled copy as not exiting the loop.
1814 (try_peel_loop): Likewise.
1815 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1816 to handle the case of constant or zero iterations and perform
1817 loop header copying on-the-fly.
1819 2017-12-14 Richard Biener <rguenther@suse.de>
1821 PR tree-optimization/83418
1822 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1823 Instead of asserting we don't get unfolded comparisons deal with
1826 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1829 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1830 basic blocks. Assert debug bind insns don't appear outside of bbs,
1831 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1834 PR tree-optimization/83198
1835 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1836 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1837 value if arg is a REAL_CST with incompatible type.
1839 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1840 Bin Cheng <bin.cheng@arm.com>
1843 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1845 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1848 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1850 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1851 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1853 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1854 Alan Hayward <alan.hayward@arm.com>
1855 David Sherwood <david.sherwood@arm.com>
1857 * poly-int.h: New file.
1858 * poly-int-types.h: Likewise.
1859 * coretypes.h: Include them.
1860 (POLY_INT_CONVERSION): Define.
1861 * target.def (estimated_poly_value): New hook.
1862 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1863 * doc/tm.texi: Regenerate.
1864 * doc/poly-int.texi: New file.
1865 * doc/gccint.texi: Include it.
1866 * doc/rtl.texi: Describe restrictions on subreg modes.
1867 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1868 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1869 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1870 * targhooks.h (default_estimated_poly_value): Declare.
1871 * targhooks.c (default_estimated_poly_value): New function.
1872 * target.h (estimated_poly_value): Likewise.
1873 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1874 (wi::unary_traits): Delete.
1875 (wi::binary_traits::signed_shift_result_type): Define for
1876 offset_int << HOST_WIDE_INT, etc.
1877 (generic_wide_int::operator <<=): Define for all types and use
1878 wi::lshift instead of <<.
1879 (wi::hwi_with_prec): Add a default constructor.
1880 (wi::ints_for): New class.
1881 (operator <<): Define for all wide-int types.
1882 (operator /): New function.
1883 (operator %): Likewise.
1884 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1885 (ASSERT_MAYBE_NE_AT): New macros.
1887 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1888 Dominik Vogt <vogt@linux.vnet.ibm.com>
1891 * emit-rtl.c (init_emit): Remove ??? comment.
1892 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1893 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1894 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1895 alignment of 3 virtual registers to BITS_PER_WORD.
1897 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1899 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1901 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1902 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1904 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1907 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1909 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1911 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1912 with the exception of debug begin stmt markers appear before
1916 * final.c (rest_of_handle_final): Call variable_tracking_main only
1917 if !flag_var_tracking.
1919 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1920 Jakub Jelinek <jakub@redhat.com>
1924 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1925 labels debug stmts that can only appear after labels.
1927 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1929 PR rtl-optimization/82398
1930 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1931 EXPR_USEFULNESS in priority comparison.
1933 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1935 PR rtl-optimization/83393
1936 * combine.c (move_deaths): If reg_stat points to a too new insn in
1937 last_death, do not use it: find the proper insn instead.
1939 2017-12-12 Jeff Law <law@redhat.com>
1941 PR tree-optimization/83298
1942 PR tree-optimization/83362
1943 PR tree-optimization/83383
1944 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1945 push_value_range a public interface. Add new argument to
1946 record_ranges_from_stmt.
1947 * gimple-ssa-evrp-analyze.c
1948 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1949 Update comments. Handle recording temporary equivalences.
1950 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1951 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1952 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1953 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1954 gimple-ssa-evrp-analyze.h.
1955 (record_temporary_equivalences_from_phis): Add new argument. When
1956 the PHI arg is an SSA_NAME, set the result's range to the range
1958 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1959 from statements too.
1960 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1961 Pass it down to children as needed.
1962 (thread_outgoing_edges): Likewise.
1963 (thread_across_edge): Likewise. Push/pop range state as needed.
1964 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1966 2017-12-12 Julia Koval <julia.koval@intel.com>
1968 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1969 (PTA_CANNONLAKE): Remove PTA_CLWB.
1971 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1974 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1975 (vcondv2div2df): Likewise.
1976 (vconduv2dfv2di): Likewise.
1978 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1980 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1981 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1982 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1983 ATTR_MATHFN_FPROUNDING_ERRNO.
1985 2017-12-12 Richard Biener <rguenther@suse.de>
1987 PR tree-optimization/83385
1988 * graphite-scop-detection.c (get_order, order): Remove.
1989 (bb_to_rpo): New global.
1991 (build_scops): Sort pbbs in RPO order.
1993 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1995 * combine.c (simplify_set): Do not transform subregs to zero_extends
1996 if the destination mode is a vector mode.
1998 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2000 PR tree-optimization/83359
2001 * tree-cfg.h (fold_loop_internal_call): Declare.
2002 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2003 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2004 (find_loop_dist_alias): New function.
2005 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2006 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2009 PR tree-optimization/80631
2010 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2011 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2012 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2013 hardcoding zero as the value if COND_EXPR is never true. For
2014 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2015 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2016 hardcoding MAX_EXPR as the reduction operation.
2017 (is_nonwrapping_integer_induction): Allow negative step.
2018 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2019 vect_create_epilog_for_reduction, if no value is suitable, don't
2020 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2022 2017-12-12 Richard Biener <rguenther@suse.de>
2024 PR tree-optimization/81889
2025 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2026 range info from the non-wrapping IV instead of just the range
2029 2017-12-12 Julia Koval <julia.koval@intel.com>
2031 * config.gcc: Add vaesintrin.h.
2032 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2034 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2035 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2037 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2038 * config/i386/immintrin.h: Include vaesintrin.h.
2039 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2040 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2041 _mm_aesdec_epi128): New intrinsics.
2043 2017-12-12 Julia Koval <julia.koval@intel.com>
2045 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2046 OPTION_MASK_ISA_VAES_UNSET): New.
2047 (ix86_handle_option): Handle -mvaes.
2048 * config/i386/cpuid.h: Define bit_VAES.
2049 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2050 * config/i386/i386-c.c (__VAES__): New.
2051 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2052 (ix86_valid_target_attribute_inner_p): Ditto.
2053 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2054 * config/i386/i386.opt: Add -mvaes.
2055 * doc/invoke.texi: Ditto.
2057 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2059 * debug.h (gcc_debug_hooks): Add inline_entry.
2060 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2061 * debug.c (do_nothing_debug_hooks): Likewise.
2062 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2063 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2064 (dwarf2_lineno_debug_hooks): Likewise.
2066 * common.opt (gstatement-frontiers): New, setting
2067 debug_nonbind_markers_p.
2068 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2069 * toplev.c (process_options): Autodetect value for debug statement
2071 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2072 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2074 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2075 markers. Integrate source bind into debug stmt expand loop.
2076 (pass_expand::execute): Check debug marker limit. Avoid deep
2077 TER and expand debug locations for debug bind insns only.
2078 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2080 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2081 * final.c (reemit_insn_block_notes): Take current block from
2082 nonbind markers. Declare note where it's first set.
2083 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2084 begin stmt markers if enabled.
2085 (notice_source_line): Handle nonbind markers. Fail if their
2086 location is unknown or that of builtins.
2087 (rest_of_handle_final): Convert begin stmt markers to notes if
2088 var-tracking didn't run.
2089 (rest_of_clean_state): Skip begin stmt markers.
2090 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2092 * function.c (allocate_struct_function): Set begin_stmt_markers.
2093 * function.h (struct function): Add debug_marker_count counter
2094 and debug_nonbind_markers flag.
2095 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2096 * gimple-low.c (lower_function_body): Adjust
2097 debug_nonbind_markers.
2098 (lower_stmt): Drop or skip gimple debug stmts.
2099 (lower_try_catch): Skip debug stmts.
2100 * gimple.c (gimple_build_debug_begin_stmt): New.
2101 (gimple_copy): Increment debug_marker_count if copying one.
2102 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2103 * gimplify.c (rexpr_location): New.
2104 (rexpr_has_location): New.
2105 (warn_switch_unreachable_r): Handle gimple debug stmts.
2106 (shortcut_cond_r): Call expr_location.
2108 (find_goto_label): New.
2109 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2111 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2113 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2114 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2115 (LANG_HOOKS_INITIALIZER): ... this.
2116 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2117 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2119 (update_ebb_live_info): Skip debug insn markers.
2120 * lra.c (debug_insn_static_data): Rename to...
2121 (debug_bind_static_data): ... this.
2122 (debug_marker_static_data): New.
2123 (lra_set_insn_recog_data): Select one of the above depending
2125 (lra_update_isn_regno_info): Don't assume debug insns have
2127 (push_insns): Skip debug insns.
2128 * lto-streamer-in.c (input_function): Drop debug stmts
2129 depending on active options. Adjust debug_nonbind_markers.
2130 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2131 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2132 begin stmt marker notes.
2133 (print_insn): Likewise.
2134 * recog.c (extract_insn): Recognize rtl for debug markers.
2135 * rtl.def (DEBUG_MARKER): New.
2136 * tree-inline.c: Include params.h.
2137 (remap_gimple_stmt): Handle nonbind markers.
2138 (maybe_move_debug_stmts_to_successors): Likewise.
2139 (copy_debug_stmt): Likewise.
2140 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2141 markers regardless of no side effects.
2142 (tsi_link_before): Don't update container's side effects when adding
2143 a begin stmt marker.
2144 (tsi_link_after): Likewise.
2145 (expr_first): Skip begin stmt markers.
2146 (expr_last): Likewise.
2147 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2148 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2149 Disregard nonbind markers.
2150 * tree.c (make_node_stat): Don't set side effects for begin stmt
2152 (build1_stat): Likewise.
2153 * tree.def (DEBUG_BEGIN_STMT): New.
2154 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2155 * var-tracking.c (delete_debug_insns): Renamed to...
2156 (delete_vta_debug_insns): ... this.
2157 (reemit_marker_as_note): New.
2158 (vt_initialize): Reemit markers.
2159 (delete_vta_debug_insns): Likewise.
2160 (vt_debug_insns_local): Reemit or delete markers.
2161 (variable_tracking_main_1): Likewise.
2162 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2163 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2164 (gimple_debug_nonbind_marker_p): New.
2165 (gimple_build_debug_bind): Adjust.
2166 (gimple_build_debug_begin_stmt): New.
2167 * doc/invoke.texi (max-debug-marker-count): New param.
2168 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2169 (debug_parameter_ref, debug_marker): New.
2170 (NOTE_INSN_BEGIN_STMT): New.
2171 (DEBUG_INSN): Describe begin stmt markers.
2173 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2174 without debug insns, we wouldn't, but clean up debug insns
2175 after a control flow insn nevertheless.
2177 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2178 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2179 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2180 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2181 (rtl_tidy_fallthru_edge): Likewise.
2182 (rtl_verify_fallthru): Likewise.
2183 (rtl_verify_bb_layout): Likewise.
2184 (skip_insns_after_block): Likewise.
2185 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2186 * dwarf2out.c: Include print-rtl.h.
2187 (dwarf2out_next_real_insn): New.
2188 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2189 Dump debug binds in asm comments.
2190 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2191 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2192 callers to use gsi_start_nondebug_bb instead.
2193 (gsi_after_labels): Skip gimple debug stmts.
2194 (gsi_start_nondebug): New.
2195 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2196 (proper_loop_form_for_interchange): Adjust.
2197 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2198 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2199 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2200 (collect_fallthrough_labels): Likewise.
2201 (should_warn_for_implicit_fallthrough): Likewise.
2202 (warn_implicit_fallthrough_r): Likewise.
2203 (expand_FALLTHROUGH_r): Likewise.
2204 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2205 (graphite_copy_stmts_from_block): Skip nonbind markers.
2206 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2207 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2208 * jump.c (clean_barriers): Skip debug insns.
2209 * omp-expand.c (expand_parallel_call): Skip debug insns.
2210 (expand_task_call): Likewise.
2211 (remove_exit_barrier): Likewise.
2212 (expand_omp_taskreg): Likewise.
2213 (expand_omp_for_init_counts): Likewise.
2214 (expand_omp_for_generic): Likewise.
2215 (expand_omp_for_static_nochunk): Likewise.
2216 (expand_omp_for_static_chunk): Likewise.
2217 (expand_omp_simd): Likewise.
2218 (expand_omp_taskloop_for_outer): Likewise.
2219 (expand_omp_taskloop_for_inner): Likewise.
2220 (expand_oacc_for): Likewise.
2221 (expand_omp_sections): Likewise.
2222 (expand_omp_single): Likewise.
2223 (expand_omp_synch): Likewise.
2224 (expand_omp_atomic_load): Likewise.
2225 (expand_omp_atomic_store): Likewise.
2226 (expand_omp_atomic_fetch_op): Likewise.
2227 (expand_omp_atomic_pipeline): Likewise.
2228 (expand_omp_atomic_mutex): Likewise.
2229 (expand_omp_target): Likewise.
2230 (grid_expand_omp_for_loop): Likewise.
2231 (grid_expand_target_grid_body): Likewise.
2232 (build_omp_regions_1): Likewise.
2233 * omp-low.c (check_combined_parallel): Skip debug stmts.
2234 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2235 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2236 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2238 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2239 (make_edges): Likewise.
2240 (cleanup_dead_labels): Likewise.
2241 (gimple_can_merge_blocks_p): Likewise.
2242 (stmt_starts_bb_p): Likewise.
2243 (gimple_block_label): Likewise.
2244 (gimple_redirect_edge_and_branch): Likewise.
2245 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2247 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2249 * tree-pretty-print (print_declaration): Omit initializer in slim
2251 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2253 (eliminate_unnecessary_stmts): Stabilize block removal order.
2254 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2255 * var-tracking.c (get_first_insn): New.
2256 (vt_emit_notes): Call it.
2257 (vt_initialize): Walk any insns before the first BB.
2258 (delete_debug_insns): Likewise.
2260 * gimple.h (enum gimple_debug_subcode): Add
2261 GIMPLE_DEBUG_BEGIN_STMT.
2262 (gimple_debug_begin_stmt_p): New.
2263 (gimple_debug_nonbind_marker_p): New.
2264 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2265 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2266 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2267 * insn-notes.def (BEGIN_STMT): New.
2268 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2269 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2270 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2271 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2272 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2273 (INSN_DEBUG_MARKER_KIND): New.
2274 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2275 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2276 (INSN_VAR_LOCATION_PTR): New.
2277 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2279 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2280 deps only if debug bind insns are enabled.
2281 (pass_expand::execute): Avoid deep TER and expand
2282 debug locations for debug bind insns only.
2283 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2284 debug stmts special handling down to debug bind stmts.
2285 * combine.c (try_combine): Narrow debug insns special handling
2286 down to debug bind insns.
2287 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2288 Narrow debug insns preexisting special handling down to debug
2290 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2291 handling down to debug bind insns.
2292 * function.c (instantiate_virtual_regs): Skip debug markers,
2293 adjust handling of debug binds.
2294 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2295 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2296 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2297 insns to debug bind insns.
2298 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2299 special handling of debug stmts to debug bind stmts.
2300 * ipa-split.c (split_function): Likewise.
2301 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2302 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2304 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2305 * regrename.c (build_def_use): Likewise.
2306 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2307 (pass_cprop_hardreg): Narrow special casing of debug insns to
2309 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2310 * reload1.c (reload): Likewise.
2311 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2312 casing of debug stmts to debug bind stmts.
2313 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2314 * ssa-iterators.h (num_imm_uses): Likewise.
2315 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2316 debug stmts to debug bind stmts.
2317 * tree-inline.c (tree_function_versioning): Narrow special casing
2318 of debug stmts to debug bind stmts.
2319 * tree-loop-distribution.c (generate_loops_for_partition):
2320 Narrow special casing of debug stmts to debug bind stmts.
2321 * tree-sra.c (analyze_access_subtree): Narrow special casing
2322 of debug stmts to debug bind stmts.
2323 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2324 stmts to debug bind stmts.
2325 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2326 casing of debug stmts to debug bind stmts.
2327 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2328 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2329 casing of debug stmts to debug bind stmts.
2330 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2332 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2333 debug stmts to debug bind stmts.
2334 (gimple_replace_ssa_lhs): Likewise.
2335 (insert_debug_temp_for_var_def): Likewise.
2336 (insert_debug_temps_for_defs): Likewise.
2337 (reset_debug_uses): Likewise.
2338 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2339 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2340 (adjust_debug_stmts): Likewise.
2341 (adjust_phi_and_debug_stmts): Likewise.
2342 (vect_do_peeling): Likewise.
2343 * tree-vect-loop.c (vect_transform_loop): Likewise.
2344 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2345 * var-tracking.c (adjust_mems): Narrow special casing of debug
2346 insns to debug bind insns.
2347 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2348 (compute_bb_dataflow, vt_find_locations): Likewise.
2349 (vt_expand_loc, emit_notes_for_changes): Likewise.
2350 (vt_init_cfa_base): Likewise.
2351 (vt_emit_notes): Likewise.
2352 (vt_initialize): Likewise.
2353 (vt_finalize): Likewise.
2355 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2356 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2357 (next_nonnote_nondebug_insn_bb): New.
2358 (prev_nonnote_nondebug_insn_bb): New.
2359 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2360 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2361 (prev_nonnote_nondebug_insn_bb): Declare.
2362 (next_nonnote_nondebug_insn_bb): Declare.
2363 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2364 * cfgrtl.c (get_last_bb_insn): Likewise.
2365 * lra.c (push_insns): Likewise.
2367 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2370 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2372 * selftest.c (selftest::test_runner::test_runner): New ctor.
2373 (selftest::test_runner::~test_runner): New dtor.
2374 * selftest.h (class selftest::test_runner): New class.
2376 2017-12-11 Carl Love <cel@us.ibm.com>
2378 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2379 vec_extract_fp32_from_shortl]): Add #defines.
2380 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2381 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2382 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2383 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2384 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2385 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2386 * doc/extend.texi: Add documentation for the added builtins.
2388 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2390 PR rtl-optimization/80693
2391 PR rtl-optimization/81019
2392 PR rtl-optimization/81020
2393 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2394 are not mentioned in i3. Place the REG_UNUSED note on i2,
2395 possibly modified to REG_DEAD, if it did not originate in i3.
2397 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2399 * recog.c (store_data_bypass_p_1): New function.
2400 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2401 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2403 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2405 PR rtl-optimization/83361
2406 * ifcvt.c (if_convert): Call fixup_partitions.
2408 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2410 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2411 early folding of splat_u{8,16,32}.
2413 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2415 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2416 output_operand_lossage first argument with capital letter.
2417 (aarch64_override_options): Don't start error and sorry first argument
2418 with capital letter.
2420 2017-12-11 Andi Kleen <ak@linux.intel.com>
2422 PR gcov-profile/83355
2423 * auto-profile.c (string_table::get_index_by_decl): Don't
2424 recurse when abstract origin points to itself.
2426 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2428 PR tree-optimization/83320
2429 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2430 (prune_datarefs_not_in_loop): Ditto.
2432 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2434 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2436 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2438 PR tree-optimization/83337
2439 * gimple-loop-interchange.cc (compute_access_stride): Handle
2440 bitfield DRs properly.
2442 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2444 PR tree-optimization/83338
2445 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2446 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2447 vector of unsigned integers to vector of signed integers.
2449 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2451 PR rtl-optimization/83317
2452 * lra-constraints.c (process_address_1): Add insn code check.
2454 2017-12-08 Michael Matz <matz@suse.de>
2456 Fix PR tree-optimization/83323
2457 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2458 head-controlled loops and loop BBs.
2459 * common.opt (funroll-and-jam): Remove, instead ...
2460 (floop-unroll-and-jam): ... reuse this option.
2461 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2462 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2463 (-floop-unroll-and-jam): ... this option.
2465 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2467 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2468 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2469 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2471 2017-12-08 Richard Biener <rguenther@suse.de>
2474 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2475 handle accesses outside of zero-sized vars.
2477 2017-12-08 Martin Jambor <mjambor@suse.cz>
2479 PR tree-optimization/83141
2480 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2481 test for MEM_REFs implicitely changing types with padding. Remove
2483 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2485 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2487 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2489 * config/avr/avr.c (avr_attribute_table): Likewise.
2490 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2491 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2492 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2493 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2494 of struct attribute_spec.
2495 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2497 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2500 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2502 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2505 * profile-count.c (profile_count::from_gcov_type): Move from
2506 profile-count.h; handle overflow.
2507 * profile-count.h (profile_count::from_gcov_type): Move offline.
2509 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2511 PR rtl-optimization/83304
2512 * combine.c (move_deaths): If we do not know where a register died,
2515 2017-12-08 Richard Biener <rguenther@suse.de>
2517 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2518 Provide -fopt-info-loop feedback when we interchange in a nest.
2520 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2522 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2523 for armv6 ARM CPU IDs.
2525 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2527 * common/config/arm/arm-common.c: Include <algorithm>.
2528 (INCLUDE_VECTOR): Define.
2529 (compare_opt_names): New function.
2530 (arm_rewrite_selected_arch): Only strip out extensions that can be
2531 expressed through -mfpu. Sort the remaining extensions
2534 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2536 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2537 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2538 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2539 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2541 2017-06-08 Tristan Gingold <gindold@adacore.com>
2544 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2545 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2547 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2549 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2550 of struct attribute_spec.
2552 2017-12-08 Julia Koval <julia.koval@intel.com>
2554 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2555 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2556 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2557 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2558 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2559 _mm_maskz_dpwssds_epi32): Ditto.
2561 2017-12-08 Richard Biener <rguenther@suse.de>
2563 PR tree-optimization/81303
2564 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2565 conditions try to create a comparison vector type matching
2566 the data vector type.
2567 (vectorizable_condition): Adjust.
2568 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2569 Leave invariant conditions alone in case we can vectorize those.
2571 2017-12-08 Julia Koval <julia.koval@intel.com>
2573 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2574 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2575 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2576 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2577 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2579 2017-12-08 Julia Koval <julia.koval@intel.com>
2581 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2582 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2583 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2584 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2585 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2586 _mm_maskz_dpbusds_epi32): New intrinsics.
2588 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2590 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2592 (ld<bh>io_signed): Likewise.
2593 (st<bhw>io): Likewise.
2594 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2595 SMALL_INT12 constant integer operand.
2597 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2599 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2601 (nios2_large_constant_allowed): ...this. Adjust uses.
2602 (nios2_plus_symbolic_constant_p): Rename to...
2603 (nios2_plus_large_constant_p): ...this. Adjust uses.
2604 (nios2_legitimate_address_p): Correct CONST_INT handling.
2605 (nios2_symbolic_memory_operand_p): Rename to...
2606 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2607 (nios2_large_constant_p): Check for large constant integers too.
2608 (nios2_split_large_constant): Handle constant integers.
2609 (nios2_split_symbolic_memory_operand): Rename to...
2610 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2611 (nios2_legitimize_constant_address): Handle constant integers.
2612 (r0rel_constant_p): Handle small constant integers.
2613 (nios2_print_operand_address): Handle r0-relative integer addresses.
2614 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2615 * config/nios2/nios2.md: Adjust for renamed functions.
2617 2017-12-07 Andrew Waterman <andrew@sifive.com>
2619 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2620 (TARGET_HAVE_SRODATA_SECTION): New define.
2621 (riscv_select_section): New function.
2623 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2624 Alexander Monakov <amonakov@ispras.ru>
2625 Jakub Jelinek <jakub@redhat.com>
2628 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2630 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2632 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2633 a bool success value. Don't call output_operand_lossage here.
2634 (aarch64_print_ldpstp_address): Return a bool success value.
2635 (aarch64_print_operand_address): Call output_addr_const if
2636 aarch64_print_address_internal fails.
2637 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2638 'y'; call output_operand_lossage instead. Call output_operand_lossage
2639 if aarch64_print_ldpstp_address fails.
2641 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2643 * tree-vector-builder.h
2644 (tree_vector_builder::binary_encoded_nelts): Declare.
2645 * tree-vector-builder.c
2646 (tree_vector_builder::binary_encoded_nelts): New function.
2647 * fold-const.c (negate_expr_p): Likewise.
2648 (operand_equal_p, fold_checksum_tree): Likewise.
2649 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2650 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2651 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2652 (uniform_vector_p): Likewise.
2653 * varasm.c (const_hash_1, compare_constant): Likewise.
2654 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2655 (valid_lattice_transition): Operate directly on the VECTOR_CST
2657 * ipa-icf.c: Include tree-vector-builder.h.
2658 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2659 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2661 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2663 * tree.c (build_vector): Delete.
2664 * tree.h (build_vector): Make static and move into the self-testing
2667 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2669 * vector-builder.h (vector_builder::derived): New const overload.
2670 (vector_builder::elt): New function.
2671 * tree-vector-builder.h (tree_vector_builder::type): New function.
2672 (tree_vector_builder::apply_step): Declare.
2673 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2675 * gimple-fold.h (tree_vector_builder): Declare.
2676 (gimple_build_vector): Take a tree_vector_builder instead of a
2677 type and vector of elements.
2678 * gimple-fold.c (gimple_build_vector): Likewise.
2679 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2681 (get_initial_defs_for_reduction): Likewise.
2682 (vectorizable_induction): Likewise.
2684 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2686 * tree-vector-builder.h
2687 (tree_vector_builder::new_binary_operation): Declare.
2688 * tree-vector-builder.c
2689 (tree_vector_builder::new_binary_operation): New function.
2690 * fold-const.c (fold_relational_const): Use it.
2691 (const_binop): Likewise. Check that both input vectors have
2692 the same number of elements, thus excluding things like WIDEN_SUM.
2693 Check whether it is possible to operate directly on the encodings
2696 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2698 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2699 new_unary_operation, operating only on the encoded elements.
2700 (const_unop): Likewise.
2701 (exact_inverse): Likewise.
2702 (distributes_over_addition_p): New function.
2703 (const_binop): Use tree_vector_builder and new_unary_operation
2704 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2705 on the encoded elements unless the encoding is strided and the
2706 operation does not distribute over addition.
2707 (fold_convert_const): Use tree_vector_builder and
2708 new_unary_operation. Operate only on the encoded elements
2709 for truncating integer conversions, or for non-stepped encodings.
2711 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2713 * config/sparc/sparc.c: Include tree-vector-builder.h.
2714 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2715 * expmed.c: Include tree-vector-builder.h.
2716 (make_tree): Use tree_vector_builder instead of build_vector.
2717 * fold-const.c: Include tree-vector-builder.h.
2718 (const_binop): Use tree_vector_builder instead of build_vector.
2719 (const_unop): Likewise.
2720 (native_interpret_vector): Likewise.
2721 (fold_vec_perm): Likewise.
2722 (fold_ternary_loc): Likewise.
2723 * gimple-fold.c: Include tree-vector-builder.h.
2724 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2726 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2727 (simplify_vector_constructor): Use tree_vector_builder instead
2729 * tree-vect-generic.c: Include tree-vector-builder.h.
2730 (add_rshift): Use tree_vector_builder instead of build_vector.
2731 (expand_vector_divmod): Likewise.
2732 (optimize_vector_constructor): Likewise.
2733 * tree-vect-loop.c: Include tree-vector-builder.h.
2734 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2735 of build_vector. Explicitly use a stepped encoding for
2737 * tree-vect-slp.c: Include tree-vector-builder.h.
2738 (vect_get_constant_vectors): Use tree_vector_builder instead
2740 (vect_transform_slp_perm_load): Likewise.
2741 (vect_schedule_slp_instance): Likewise.
2742 * tree-vect-stmts.c: Include tree-vector-builder.h.
2743 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2744 (vect_gen_perm_mask_any): Likewise.
2745 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2746 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2748 (build_vector_from_val): Likewise. Explicitly use a duplicate
2751 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2753 * doc/generic.texi (VECTOR_CST): Describe new representation of
2755 * vector-builder.h: New file.
2756 * tree-vector-builder.h: Likewise.
2757 * tree-vector-builder.c: Likewise.
2758 * Makefile.in (OBJS): Add tree-vector-builder.o.
2759 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2760 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2761 (tree_vector): Change the number of elements to
2762 vector_cst_encoded_nelts.
2763 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2764 (VECTOR_CST_ELTS): Delete.
2765 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2766 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2767 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2768 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2769 (VECTOR_CST_ENCODED_ELT): Likewise.
2770 (vector_cst_encoded_nelts): New function.
2771 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2772 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2773 (vector_cst_int_elt, vector_cst_elt): Declare.
2774 * tree.c: Include tree-vector-builder.h.
2775 (tree_code_size): Abort if passed VECTOR_CST.
2776 (tree_size): Update for new VECTOR_CST layout.
2777 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2778 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2779 (build_vector): Use tree_vector_builder.
2780 (vector_cst_int_elt, vector_cst_elt): New functions.
2781 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2782 encoded elements and then create the vector in the canonical form.
2783 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2784 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2785 (tree_c_tests): Call test_vector_cst_patterns.
2786 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2788 (hash_tree): Likewise.
2789 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2790 (streamer_write_tree_header): Likewise.
2791 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2792 (streamer_alloc_tree): Likewise. Update call to make_vector.
2793 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2795 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2797 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2798 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2799 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2801 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2802 Richard Biener <rguenther@suse.de>
2804 PR tree-optimization/81303
2805 * Makefile.in (gimple-loop-interchange.o): New object file.
2806 * common.opt (floop-interchange): Reuse the option from graphite.
2807 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2808 -floop-interchange and mention it for -O3.
2809 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2810 * gimple-loop-interchange.cc: New file.
2811 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2812 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2813 * passes.def (pass_linterchange): New pass.
2814 * timevar.def (TV_LINTERCHANGE): New time var.
2815 * tree-pass.h (make_pass_linterchange): New declaration.
2816 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2817 interchange. Record IV before/after increment in new parameters.
2818 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2819 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2821 (check_reduction_path): ...New function here.
2822 * tree-vectorizer.h (check_reduction_path): New declaration.
2824 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2827 PR rtl-optimization/80818
2828 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2829 always early clobbered.
2830 * lra-lives.c (process_bb_lives): Check input hard regs for early
2831 clobbered non-operand hard reg.
2833 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2836 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2837 types_compatible_p, just that TYPE_MODE is the same.
2839 2017-12-07 Martin Sebor <msebor@redhat.com>
2842 * attribs.c (empty_attribute_table): Initialize new member of
2843 struct attribute_spec.
2844 (decl_attributes): Add argument. Handle mutually exclusive
2845 combinations of attributes.
2846 (selftests::test_attribute_exclusions): New function.
2847 (selftests::attribute_c_tests): Ditto.
2848 * attribs.h (decl_attributes): Add default argument.
2849 * selftest.h (attribute_c_tests): Declare.
2850 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2851 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2853 * doc/extend.texi (Common Function Attributes): Update const and pure.
2854 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2855 of struct attribute_spec.
2856 * config/arc/arc.c (arc_attribute_table): Same.
2857 * config/arm/arm.c (arm_attribute_table): Same.
2858 * config/avr/avr.c ( avr_attribute_table): Same.
2859 * config/bfin/bfin.c (bfin_attribute_table): Same.
2860 * config/cr16/cr16.c (cr16_attribute_table): Same.
2861 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2862 * config/h8300/h8300.c (h8300_attribute_table): Same.
2863 * config/i386/i386.c (ix86_attribute_table): Same.
2864 * config/ia64/ia64.c (ia64_attribute_table): Same.
2865 * config/m32c/m32c.c (m32c_attribute_table): Same.
2866 * config/m32r/m32r.c (m32r_attribute_table): Same.
2867 * config/m68k/m68k.c (m68k_attribute_table): Same.
2868 * config/mcore/mcore.c (mcore_attribute_table): Same.
2869 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2870 * config/mips/mips.c (mips_attribute_table): Same.
2871 * config/msp430/msp430.c (msp430_attribute_table): Same.
2872 * config/nds32/nds32.c (nds32_attribute_table): Same.
2873 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2874 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2875 * config/rl78/rl78.c (rl78__attribute_table): Same.
2876 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2877 * onfig/rx/rx.c (rx_attribute_table): Same.
2878 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2879 * config/sh/sh.c (sh_attribute_table): Same.
2880 * config/sparc/sparc.c (sparc_attribute_table): Same.
2881 * config/spu/spu.c (spu_attribute_table): Same.
2882 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2883 * config/v850/v850.c (v850_attribute_table): Same.
2884 * config/visium/visium.c (visium_attribute_table): Same.
2886 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2889 * config/arm/arm.c (INCLUDE_STRING): Define.
2890 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2891 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2894 2017-12-07 Michael Matz <matz@suse.de>
2896 Add unroll and jam pass
2898 * gimple-loop-jam.c: New file.
2899 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2900 * common.opt (funroll-and-jam): New option.
2901 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2902 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2903 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2904 * passes.def: Add pass_loop_jam.
2905 * timevar.def (TV_LOOP_JAM): Add.
2906 * tree-pass.h (make_pass_loop_jam): Declare.
2907 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2908 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2909 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2910 to flow_loop_tree_node_add.
2911 (duplicate_subloops, copy_loops_to): Append to sibling list.
2912 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2913 * doc/invoke.texi (-funroll-and-jam): Document new option.
2914 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2916 2017-12-07 Richard Biener <rguenther@suse.de>
2918 PR tree-optimization/83296
2919 PR tree-optimization/67769
2920 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2921 flow sensitive info in an unrelated BB.
2922 (value_replacement): Use reset_flow_sensitive_info.
2923 (minmax_replacement): Reset flow sensitive info on the def
2924 we move. Do not reset flow sensitive info in the whole BB
2925 we move the stmt to.
2926 (abs_replacement): Likewise.
2928 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2931 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2932 rs6000_cpu to the given -mcpu=, or to the default processor.
2934 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2936 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2937 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2938 instead of rs6000_cpu_attr.
2939 (rs6000_adjust_cost): Ditto.
2940 (is_microcoded_insn): Ditto.
2941 (rs6000_adjust_priority): Ditto.
2942 (rs6000_issue_rate): Ditto.
2943 (rs6000_use_sched_lookahead): Ditto.
2944 (rs6000_use_sched_lookahead_guard): Ditto.
2945 (rs6000_sched_reorder): Ditto.
2946 (force_new_group): Ditto.
2947 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2948 (group_ending_nop): Ditto.
2950 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2952 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2953 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2954 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2955 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2957 (rs6000_reassociation_width): Ditto.
2958 (rs6000_emit_epilogue): Ditto.
2959 (rs6000_adjust_cost): Ditto.
2960 (is_microcoded_insn): Ditto.
2961 (is_cracked_insn): Ditto.
2962 (rs6000_adjust_priority): Ditto.
2963 (rs6000_sched_reorder): Ditto.
2964 (rs6000_sched_reorder2): Ditto.
2965 (insn_must_be_first_in_group): Ditto.
2966 (insn_must_be_last_in_group): Ditto.
2967 (rs6000_register_move_cost): Ditto.
2968 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2971 2017-12-07 Julia Koval <julia.koval@intel.com>
2973 * config.gcc: Add vaesintrin.h.
2974 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2976 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2977 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2979 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2980 * config/i386/immintrin.h: Include vaesintrin.h.
2981 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2982 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2983 _mm_aesdec_epi128): New intrinsics.
2985 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2987 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2988 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2989 c-family/c-spellcheck.cc.
2990 (best_macro_match::best_macro_match): Likewise.
2992 (struct edit_distance_traits<cpp_hashnode *>): Move to
2993 c-family/c-spellcheck.h.
2994 (class best_macro_match): Likewise.
2996 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2998 PR tree-optimization/83293
2999 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3000 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3001 might insert into empty bb.
3004 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3005 simplify for plus with :c added, and pointer_plus without that.
3006 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3007 with undefined overflow and the conversion is not widening,
3008 perform negation in utype and only convert to type afterwards.
3009 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3010 simplify for plus with :c added, and pointer_plus without that.
3011 If type is integral with undefined overflow and the conversion is
3012 not widening, perform minus in utype and only convert to type
3013 afterwards. Move the last pointer_diff_expr simplify into the
3016 2017-12-06 Martin Sebor <msebor@redhat.com>
3018 PR tree-optimization/82646
3019 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3020 strncpy, not maxlen.
3022 2017-12-06 Martin Sebor <msebor@redhat.com>
3024 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3027 PR tree-optimization/83075
3028 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3029 strncat/strncpy don't change length of source string.
3031 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3034 2017-11-29 Martin Aberg <maberg@gaisler.com>
3036 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3037 to prevent b2bst errata sequence.
3038 (sqrtdf2_fix): Likewise.
3040 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3042 PR tree-optimization/81945
3043 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3044 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3045 to dest_cfun has orig_loop_num set, either remap it to the new
3046 loop number if the loop got moved too, or clear it.
3048 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3050 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3052 (thunderx2t99_nothing): New insn reservation.
3053 (thunderx2t99_mrs): New insn reservation.
3054 (thunderx2t99_multiple): New insn reservation.
3055 (thunderx2t99_alu_basi): Add bfx to reservation.
3056 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3058 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3061 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3063 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3065 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3066 tree-ssa-pre.c::remove_dead_inserted_code.
3067 * tree-ssa-dce.h: New file.
3068 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3069 (remove_dead_inserted_code): Move and rename to function
3070 tree-ssa-dce.c::simple_dce_from_worklist.
3071 (pass_pre::execute): Update use.
3073 2017-12-05 Richard Biener <rguenther@suse.de>
3075 PR tree-optimization/83277
3076 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3077 to code-gen liveout vars.
3079 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3081 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3082 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3083 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3086 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3088 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3090 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3091 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3094 2017-12-05 Richard Biener <rguenther@suse.de>
3096 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3097 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3098 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3099 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3101 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3103 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3105 2017-12-05 Martin Liska <mliska@suse.cz>
3106 Jakub Jelinek <jakub@redhat.com>
3108 * doc/invoke.texi: Document the options.
3109 * flag-types.h (enum sanitize_code): Add
3110 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3111 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3112 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3113 * opts.c: Define new sanitizer options.
3114 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3115 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3117 2017-12-05 Julia Koval <julia.koval@intel.com>
3119 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3120 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3121 (ix86_handle_option): Handle -mavx512vnni.
3122 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3123 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3124 * config/i386/i386-c (__AVX512VNNI__): New.
3125 * config/i386/i386.c (ix86_target_string): Handle new option.
3126 (ix86_valid_target_attribute_inner_p): Handle new option.
3127 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3128 * config/i386/i386.opt (mavx512vnni): New option.
3130 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3133 * athlon.md: Disable for generic.
3134 * haswell.md: Enable for generic.
3135 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3136 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3138 (ix86_adjust_cost): Move generic to haswell path.
3140 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3142 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3143 instead of MEM_P in a couple more places. Fix formatting issues.
3145 2017-12-04 Jim Wilson <jimw@sifive.com>
3147 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3148 instead of GP_REG_LAST-1.
3149 (riscv_adjust_libcall_cfi_prologue): Likewise.
3150 (riscv_adjust_libcall_cri_epilogue): Likewise.
3151 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3154 2017-12-04 Luis Machado <luis.machado@linaro.org>
3156 * ipa-pure-const.c (check_decl): Add missing newline.
3157 (state_from_flags): Likewise.
3159 2017-12-04 Jeff Law <law@redhat.com>
3161 PR tree-optimizatin/78496
3162 * gimple-ssa-evrp-analyze.h
3163 (evrp_range_analyzer::get_vr_values): Simplify.
3164 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3165 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3166 and gimple-ssa-evrp-analyze.h.
3167 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3168 (simplify_stmt_for_jump_threading): Copy a blob of code from
3169 tree-vrp.c to use ranges to simplify statements.
3170 (dom_opt_dom_walker::before_dom_children): Call
3171 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3172 (dom_opt_dom_walker::after_dom_children): Similarly for
3173 evrp_range_analyzer::leave.
3174 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3177 * gimple-ssa-evrp-analyze.c
3178 (evrp_range_analyzer::extract_range_from_stmt): Always use
3179 vr_values::update_value_range so preexisting range info is
3180 medged with new range info, even if the new range is VR_VARYING.
3182 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3184 * combine.c: Adjust comment.
3185 (use_crosses_set_p): Delete.
3186 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3187 (try_combine): Ditto.
3189 2017-12-04 Richard Biener <rguenther@suse.de>
3191 PR tree-optimization/83255
3192 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3193 Re-add zero-iteration check.
3195 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3197 PR rtl-optimization/83245
3198 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3199 hard registers as earlyclobber, also if not in an asm.
3201 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3205 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3209 rs6000_cpu based on cpu_index, not tune_index.
3211 2017-12-04 Richard Biener <rguenther@suse.de>
3213 PR tree-optimization/83238
3214 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3215 code match comment, rejecting invalid SESE regions.
3217 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3219 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3220 require base operand is a REG_POINTER prior to reload on targets
3221 with non-equivalent space registers.
3223 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3225 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3226 (update_specialized_profile): Fix updating of counts.
3227 (perhaps_add_new_callers): Likewise.
3229 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3232 * x86-tune.def: Remove obsolette FIXMEs.
3233 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3234 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3235 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3237 (X86_TUNE_PAD_RETURNS): Disable for generic.
3239 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3241 PR tree-optimization/83170
3242 PR tree-optimization/83241
3243 * gimple-ssa-store-merging.c
3244 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3245 gimple_vuse (ins_stmt) in case it has changed.
3246 (imm_store_chain_info::output_merged_store): Likewise.
3248 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3252 * tree-cfg.c (pass_warn_function_return::execute): Handle
3253 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3254 with BUILTINS_LOCATION.
3258 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3259 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3260 instead of DECL_MODE.
3262 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3263 last argument to rtx pointer.
3264 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3265 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3266 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3267 depending on the chosen ISAs.
3268 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3269 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3270 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3272 * config/i386/sse.md (mov<mode>_internal): Likewise.
3273 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3275 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3277 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3279 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3281 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3282 parameters from prototype.
3283 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3284 parameters. Don't print a comment.
3285 (emit_fusion_gpr_load): Adjust.
3286 (emit_fusion_load_store): Adjust.
3287 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3288 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3289 comment on the second line.
3291 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3294 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3295 rs6000_cpu based on cpu_index, not tune_index.
3297 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3299 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3300 which_alternative instead of which_alternative + 1.
3301 (output_asm_insn): Print an extra tab if the template is short.
3303 2017-12-01 Jim Wilson <jimw@sifive.com>
3305 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3307 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3308 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3310 * doc/tm.texi: Regenerate.
3312 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3315 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3316 whether we can allocate pseudos before trying to fix an address.
3317 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3318 memory address is indexed or indirect.
3319 (floatuns_<mode>si2_hw2): Likewise.
3321 2017-12-01 Jason Merrill <jason@redhat.com>
3323 * Makefile.in (TAGS): Add c-family/*.cc.
3325 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3327 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3328 (call_value_insn): Likewise.
3329 (sibcall_insn): Likewise.
3330 (sibcall_value_insn): Likewise.
3331 (movsi_aarch64): Likewise.
3332 (movdi_aarch64): Likewise.
3333 (add_losym_): Likewise.
3334 (ldr_got_small_): Likewise.
3335 (ldr_got_small_sidi): Likewise.
3336 (ldr_got_small_28k_): Likewise.
3337 (ldr_got_small_28k_sidi): Likewise.
3338 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3339 Move output_addr_const to symbolic case. Add error check.
3341 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3343 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3344 * config/s390/s390.md: Disable MVC merging peephole if it would
3345 disable operand forwarding.
3346 (new peephole2): Split MVCs if it would turn them into up to 2
3349 2017-12-01 Richard Biener <rguenther@suse.de>
3351 PR tree-optimization/83232
3352 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3353 detection of same access. Instead of breaking the group here
3354 do not consider the duplicate. Add comment explaining real fix.
3356 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3358 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3360 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3362 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3364 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3366 * function.h (struct function): Remove cilk_frame_decl,
3367 is_cilk_function and calls_cilk_spawn fields.
3368 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3370 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3371 cilk_elemental field.
3372 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3373 * target.def: Adjust comment.
3374 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3375 Don't test cilk_elemental.
3377 PR tree-optimization/83233
3378 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3379 bswap_stat name for the struct.
3382 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3383 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3384 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3385 SWITCH_BREAK_LABEL_P set on the label.
3386 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3387 added for default case if it was missing and not all cases covered.
3388 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3389 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3391 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3392 to avoid -Wimplicit-fallthrough warning.
3393 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3394 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3396 PR tree-optimization/83221
3397 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3399 (init_reassoc): Formatting fix.
3402 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3403 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3407 * tree-cfg.c: Include asan.h.
3408 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3410 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3412 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3414 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3416 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3417 vpcmpeqd instruction.
3419 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3421 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3422 registers type for 128bit mode.
3424 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3426 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3428 * gengtype-state.c (read_state_pair): Likewise.
3429 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3430 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3431 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3432 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3433 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3434 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3435 * builtins.c (fold_builtin_3): Likewise.
3436 * graphite-scop-detection.c
3437 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3438 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3441 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3444 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3445 parameter. Return true if ignore_decl_rtl_set_p is true and other
3446 conditions are satisfied.
3447 * asan.h (asan_protect_global): Add new parameter.
3448 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3449 to asan_protect_global calls.
3451 2017-11-30 Jim Wilson <jimw@sifive.com>
3453 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3454 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3455 -mstrict-align, add info on default value. Delete redundant lines for
3456 -mabi. Add missing -mexplicit-relocs docs.
3458 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3460 * config/arc/arc.md (trap): New pattern.
3462 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3464 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3465 instruction to end into a delay slot.
3466 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3467 can be placed into a delay slot against reg_note.
3469 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3471 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3472 labels number of usages.
3474 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3476 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3478 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3480 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3483 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3484 multiplication by power of 2 constant into two shifts + comparison.
3486 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3489 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3491 2017-11-30 Richard Biener <rguenther@suse.de>
3493 PR tree-optimization/83202
3494 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3495 allow_peel argument and guard peeling.
3496 (canonicalize_loop_induction_variables): Likewise.
3497 (canonicalize_induction_variables): Pass false.
3498 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3499 peeling from cunrolli.
3501 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3503 * combine.c (try_combine): Print a message to dump file whenever
3504 I0, I1, or I2 cannot be combined into I3.
3506 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3508 PR rtl-optimization/83156
3509 PR rtl-optimization/82621
3510 * combine.c (try_combine): Don't split an I2 if one of the dests is
3511 set again before I3. Allow unused dests.
3513 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3515 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3517 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3519 PR rtl-optimization/80818
3520 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3521 recursively. Use insn code for clobber.
3522 (lra_set_insn_recog_data): Pass the new arg to
3523 collect_non_operand_hard_regs.
3524 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3526 (lra_update_insn_regno_info): Pass insn to
3527 add_regs_to_insn_regno_info.
3529 2017-11-29 Jim Wilson <jimw@sifive.com>
3530 Andrew Waterman <andrew@sifive.com>
3532 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3533 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3534 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3535 use SINGLE_SHIFT_COST.
3536 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3537 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3539 2017-11-29 Julia Koval <julia.koval@intel.com>
3541 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3542 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3543 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3544 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3545 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3546 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3547 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3548 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3549 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3550 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3551 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3552 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3553 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3554 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3555 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3556 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3557 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3558 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3559 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3560 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3561 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3562 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3563 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3564 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3565 __builtin_ia32_vpshldv_v2di_mask,
3566 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3567 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3568 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3570 2017-11-29 Julia Koval <julia.koval@intel.com>
3572 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3573 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3574 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3575 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3576 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3577 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3578 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3579 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3580 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3581 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3582 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3583 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3584 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3585 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3586 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3587 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3588 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3589 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3590 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3591 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3592 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3594 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3596 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3597 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3600 2017-11-29 Martin Aberg <maberg@gaisler.com>
3602 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3603 to prevent b2bst errata sequence.
3604 (sqrtdf2_fix): Likewise.
3606 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3608 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3609 (div_sqrt_insn_p): New function.
3610 (sparc_do_work_around_errata): Insert NOP instructions to
3611 prevent sequences that could trigger the TN-0013 errata for
3612 certain LEON3 processors.
3613 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3614 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3615 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3616 (in_branch_delay): Prevent div and sqrt in delay slot if
3618 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3620 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3622 * config/sparc/sparc.c (atomic_insn_p): New function.
3623 (sparc_do_work_around_errata): Insert NOP instructions to
3624 prevent sequences that could trigger the TN-0010 errata for
3626 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3627 instruction referable in atomic_insns_p.
3629 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3631 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3632 (atomic_compare_and_swap_leon3_1): Likewise.
3635 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3637 * config/sparc/sparc.c (fpop_insn_p): New function.
3638 (sparc_do_work_around_errata): Insert NOP instructions to
3639 prevent sequences that could trigger the TN-0012 errata for
3641 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3642 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3643 (in_branch_annul_delay): Prevent floating-point instructions
3644 in delay slot of annulled integer branch.
3646 2017-11-29 Richard Biener <rguenther@suse.de>
3648 PR tree-optimization/83202
3649 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3651 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3652 nodes vectorized to the same stmts multiple times.
3653 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3654 (vect_analyze_slp_instance): Adjust.
3655 (scalar_stmts_to_slp_tree_map_t): New typedef.
3656 (vect_schedule_slp_instance): Add a map recording the SLP node
3657 representing the vectorized stmts for a set of scalar stmts.
3658 Avoid code-generating redundancies.
3659 (vect_schedule_slp): Allocate map and pass it down.
3661 2017-11-29 Nathan Sidwell <nathan@acm.org>
3664 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3667 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3669 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3670 cases for printing LDP/STP memory addresses.
3671 (aarch64_print_address_internal): Renamed from
3672 aarch64_print_operand_address, added parameter, add Pmode check.
3673 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3674 (aarch64_print_operand_address): Indirect to
3675 aarch64_print_address_internal.
3676 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3677 'y' operand output specifier.
3679 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3682 * tree.c (build_simple_mem_ref_loc): Handle
3683 get_addr_base_and_unit_offset returning a MEM_REF.
3686 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3689 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3692 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3694 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3696 (layout::move_to_column): Likewise.
3698 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3701 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3702 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3704 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3705 Martin Jambor <mjambor@suse.cz>
3708 * tree.h (expr_type_first_operand_type_p): Declare
3709 * tree.c (expr_type_first_operand_type_p): New function.
3710 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3711 (ipa_value_from_jfunc): Adjust declaration.
3712 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3713 Use it as result type for arithmetics, unless it is NULL in which case
3714 be more conservative.
3715 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3716 ipa_get_jf_pass_through_result.
3717 (propagate_vals_across_pass_through): Likewise.
3718 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3719 is to propagate_vals_across_pass_through.
3720 (propagate_constants_across_call): Pass PARM_TYPE to
3721 propagate_scalar_across_jump_function.
3722 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3723 ipa_value_from_jfunc.
3724 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3725 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3726 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3727 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3728 target_type, pass it to ipa_value_from_jfunc.
3729 (update_indirect_edges_after_inlining): Pass parameter type to
3730 try_make_edge_direct_simple_call.
3732 2017-11-28 Jeff Law <law@redhat.com>
3734 * gimple-ssa-evrp-analyze.c
3735 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3736 refine ranges if scev_initialized_p returns true.
3737 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3739 2017-11-28 Julia Koval <julia.koval@intel.com>
3741 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3742 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3743 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3744 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3745 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3746 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3747 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3748 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3749 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3750 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3751 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3752 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3753 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3754 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3755 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3756 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3757 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3758 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3759 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3760 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3761 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3762 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3764 2017-11-28 Julia Koval <julia.koval@intel.com>
3766 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3767 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3768 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3769 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3770 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3771 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3772 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3773 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3774 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3775 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3776 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3777 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3778 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3779 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3780 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3781 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3782 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3783 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3784 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3785 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3786 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3787 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3788 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3789 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3790 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3791 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3792 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3793 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3794 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3795 __builtin_ia32_vpshld_v2di_mask): New builtins.
3796 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3797 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3799 2017-11-28 Richard Biener <rguenther@suse.de>
3801 PR tree-optimization/80776
3802 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3804 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3806 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3807 If the incoming edge is an effective fallthru because the other
3808 edge only reaches a __builtin_unreachable () then record ranges
3809 derived from the controlling condition in SSA info.
3810 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3811 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3813 2017-11-28 Olivier Hainque <hainque@adacore.com>
3815 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3816 on mingw build hosts.
3818 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3821 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3823 2017-11-28 Richard Biener <rguenther@suse.de>
3826 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3827 copies generated from memcpy use a character array as reference
3830 2017-11-28 Julia Koval <julia.koval@intel.com>
3831 Sebastian Peryt <sebastian.peryt@intel.com>
3833 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3834 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3835 cilk-common.o, cilk.h, cilk-common.c): Remove.
3837 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3838 * builtins.c (is_builtin_name): Remove cilkplus condition.
3839 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3840 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3841 cilk-builtins.def, cilkplus.def): Remove.
3842 * cif-code.def (CILK_SPAWN): Remove.
3843 * cilk-builtins.def: Delete.
3844 * cilk-common.c: Ditto.
3846 * cilkplus.def: Ditto.
3847 * config/darwin.h (fcilkplus): Delete.
3848 * cppbuiltin.c: Ditto.
3849 * doc/extend.texi: Remove cilkplus doc.
3850 * doc/generic.texi: Ditto.
3851 * doc/invoke.texi: Ditto.
3852 * doc/passes.texi: Ditto.
3853 * gcc.c (fcilkplus): Remove.
3854 * gengtype.c (cilk.h): Remove.
3855 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3857 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3859 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3861 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3862 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3863 cilkplus conditions.
3864 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3865 inline_read_section): Ditto.
3866 * ipa-inline-analysis.c (cilk.h): Remove.
3867 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3868 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3869 append_linker_options): Remove condition for fcilkplus.
3870 * lto/lto-lang.c (cilk.h): Remove.
3871 (lto_init): Remove condition for fcilkplus.
3872 * omp-expand.c (expand_cilk_for_call): Delete.
3873 (expand_omp_taskreg, expand_omp_for_static_chunk,
3874 expand_omp_for): Remove cilkplus
3876 (expand_cilk_for): Delete.
3877 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3878 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3879 execute_lower_omp, diagnose_sb_0): Ditto.
3880 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3881 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3882 * tree-nested.c: Ditto.
3883 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3884 (dump_generic_node): Ditto.
3885 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3886 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3888 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3890 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3892 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3893 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3894 for vector mode and !TARGET_SIMD.
3896 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3898 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3900 * tree.h (SWITCH_LABELS): Remove.
3901 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3902 assert SWITCH_BODY is non-NULL.
3903 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3905 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3907 PR tree-optimization/80788
3908 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3909 has TREE_OVERFLOW set, call drop_tree_overflow.
3911 2017-11-28 Richard Biener <rguenther@suse.de>
3913 PR tree-optimization/83158
3914 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3916 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3919 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3920 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3922 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3924 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3925 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3927 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3930 * varasm.c (compare_constant): Take the mode of the constants into
3931 account when comparing floating point constants.
3933 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3935 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3936 from explicit instantiation of debug_helper.
3937 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3939 2017-11-27 Richard Biener <rguenther@suse.de>
3941 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3944 2017-11-27 Richard Biener <rguenther@suse.de>
3946 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3947 (type_hash_canon): Also clear int_cst_hash_table entry for
3949 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3951 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3953 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3954 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3955 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3957 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3959 * hash-map.h (gt_cleare_cache): Avoid UB.
3961 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3963 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3964 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3965 (decide_unroll_runtime_iterations): Likewise.
3966 (decide_unroll_stupid): Likewise.
3968 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3971 * config/i386/i386.md: Add a loop with incssp.
3973 2017-11-27 Martin Jambor <mjambor@suse.cz>
3975 PR tree-optimization/81248
3976 * tree-sra.c (splice_param_accesses): Remove size check.
3977 (decide_one_param_reduction): Fix size check.
3978 * gimple-pretty-print.c (dump_profile): Silence warning.
3979 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3981 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3984 * dbxout.c (lastlineno): New variable.
3985 (dbx_debug_hooks): Use dbxout_switch_text_section as
3986 switch_text_section debug hook.
3987 (dbxout_function_end): Switch to current_function_section
3988 rather than function_section. If crtl->has_bb_partition,
3989 output just one N_FUN, depending on in_cold_section_p.
3990 (dbxout_source_line): Remember last lineno in lastlineno.
3991 (dbxout_switch_text_section): New function.
3992 (dbxout_function_decl): Adjust dbxout_block caller.
3993 (dbx_block_with_cold_children): New function.
3994 (dbxout_block): Return true if any LBRAC/RBRAC have been
3995 emitted. Use dbx_block_with_cold_children at depth == 0
3996 in second partition. Add PARENT_BLOCKNUM argument, pass
3997 it optionally adjusted to children. Output LBRAC/RBRAC
3998 around recursive call only if the block is in the current
3999 partition, if not and anything was output, emit empty
4001 * final.c (final_scan_insn): Compute cold_function_name
4002 before calling switch_text_section debug hook. Call
4003 that hook even if dwarf2out_do_frame if not emitting
4007 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4008 TREE_READONLY decls.
4010 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4012 PR rtl-optimization/82488
4013 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4015 2017-11-26 Julia Koval <julia.koval@intel.com>
4017 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4019 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4022 2017-11-26 Julia Koval <julia.koval@intel.com>
4024 * config/i386/driver-i386.c (host_detect_local_cpu):
4025 Detect skylake-avx512.
4027 2017-11-26 Julia Koval <julia.koval@intel.com>
4029 * config.gcc: Add -march=cannonlake.
4030 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4031 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4032 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4033 (PTA_CANNONLAKE): New.
4034 (processor_target_table): Add cannonlake.
4035 (ix86_option_override_internal): Ditto.
4036 (fold_builtin_cpu): Ditto.
4037 (get_builtin_code_for_version): Handle cannonlake.
4038 (M_INTEL_COREI7_CANNONLAKE): New.
4039 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4040 * doc/invoke.texi: Add -march=cannonlake.
4042 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4044 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4045 (try_init_one_plugin): Alternative implementation for MinGW.
4046 * Makefile.in (plugin_implib): New.
4047 (gengtype-lex.c): Fix broken AIX workaround.
4048 * configure: Regenerate.
4049 * doc/plugins.texi: Document support for MinGW.
4051 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4053 PR rtl-optimization/81553
4054 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4055 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4056 is a shift where C1 has different mode than the whole shift, use C1's
4057 mode for MULT rather than the shift's mode.
4060 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4061 builtins not enabled in the currently selected ISA.
4063 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4065 PR tree-optimization/71026
4066 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4067 (insert_reciprocals): Change to insert reciprocals before a division
4068 by a square and to insert the square of a reciprocal.
4069 (execute_cse_reciprocals_1): Change to consider division by a square.
4070 (register_division_in): Add importance parameter.
4072 2017-11-24 Richard Biener <rguenther@suse.de>
4074 PR tree-optimization/82402
4075 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4076 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4078 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4080 * match.pd (0-ptr): New transformation.
4082 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4085 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4086 when skipping checking.
4088 2017-11-24 Richard Biener <rguenther@suse.de>
4090 PR tree-optimization/83128
4091 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4092 (vn_reference_lookup_3): Likewise.
4094 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4097 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4098 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4101 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4102 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4103 check for the latter with gimple_call_builtin_p. Do not handle
4104 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4106 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4108 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4111 2017-11-23 Julia Koval <julia.koval@intel.com>
4113 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4114 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4115 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4116 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4117 _mm512_maskz_expandloadu_epi16): New intrinsics.
4118 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4119 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4120 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4121 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4122 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4123 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4124 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4125 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4126 _mm256_maskz_expandloadu_epi8): New intrinsics.
4127 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4128 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4129 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4130 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4131 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4132 * config/i386/sse.md (VI248_VLBW): New iterator.
4133 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4135 2017-11-23 Julia Koval <julia.koval@intel.com>
4137 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4138 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4139 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4140 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4141 _mm512_mask_compressstoreu_epi16): New.
4142 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4143 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4144 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4145 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4146 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4147 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4148 _mm256_mask_compressstoreu_epi8): New.
4149 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4150 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4151 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4152 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4153 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4154 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4155 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4156 __builtin_ia32_compresshi128_mask,
4157 __builtin_ia32_compressstoreuqi512_mask,
4158 __builtin_ia32_compressstoreuhi512_mask,
4159 __builtin_ia32_compressstoreuqi256_mask,
4160 __builtin_ia32_compressstoreuqi128_mask,
4161 __builtin_ia32_compressstoreuhi256_mask,
4162 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4163 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4165 (ix86_expand_special_args_builtin): Handle new types.
4166 (s4fma_expand): Handle new builtin array.
4167 * config/i386/immintrin.h: Include new headers.
4168 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4169 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4171 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4173 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4174 fixes. Declare temp and g variables at the top in order to avoid
4175 {} in most of the cases.
4177 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4179 * match.pd (ptr-0): New transformation.
4181 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4183 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4184 (struct addr_mode_cost_table): New.
4185 (struct tune_params): Add field addr_mode_costs.
4186 * config/arm/arm.c (generic_addr_mode_costs): New.
4187 (arm_slowmul_tune): Initialise addr_mode_costs field.
4188 (arm_fastmul_tune): Likewise.
4189 (arm_strongarm_tune): Likewise.
4190 (arm_xscale_tune): Likewise.
4191 (arm_9e_tune): Likewise.
4192 (arm_marvell_pj4_tune): Likewise.
4193 (arm_v6t2_tune): Likewise.
4194 (arm_cortex_tune): Likewise.
4195 (arm_cortex_a8_tune): Likewise.
4196 (arm_cortex_a7_tune): Likewise.
4197 (arm_cortex_a15_tune): Likewise.
4198 (arm_cortex_a35_tune): Likewise.
4199 (arm_cortex_a53_tune): Likewise.
4200 (arm_cortex_a57_tune): Likewise.
4201 (arm_exynosm1_tune): Likewise.
4202 (arm_xgene1_tune): Likewise.
4203 (arm_cortex_a5_tune): Likewise.
4204 (arm_cortex_a9_tune): Likewise.
4205 (arm_cortex_a12_tune): Likewise.
4206 (arm_cortex_a73_tune): Likewise.
4207 (arm_v7m_tune): Likewise.
4208 (arm_cortex_m7_tune): Likewise.
4209 (arm_v6m_tune): Likewise.
4210 (arm_fa726te_tune): Likewise.
4211 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4213 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4215 * config/arm/arm.c (arm_mem_costs): New function.
4216 (arm_rtx_costs_internal): Use arm_mem_costs.
4218 2017-11-23 Mark Wielaard <mark@klomp.org>
4220 * dwarf2out.c (init_sections_and_labels): Use generation to create
4221 unique ranges_section_label and ranges_base_label. Return generation.
4222 (output_rnglists): Add generation argument. Use generation to create
4223 unique ranges labels.
4224 (dwarf2out_finish): Get generation from init_sections_and_labels
4225 and pass generation to output_rnglists.
4227 2017-11-23 Mike Stump <mikestump@comcast.net>
4228 Eric Botcazou <ebotcazou@adacore.com>
4230 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4231 * cfgloop.h (struct loop): Add unroll field.
4232 * function.h (struct function): Add has_unroll bitfield.
4233 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4234 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4235 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4236 (pass_rtl_unroll_loops::gate): Likewise.
4237 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4238 for which loop->unroll==1.
4239 (decide_unroll_constant_iterations): Use note for consistency and
4240 take loop->unroll into account. Return early if loop->unroll is set.
4241 Fix thinko in existing test.
4242 (decide_unroll_runtime_iterations): Use note for consistency and
4243 take loop->unroll into account.
4244 (decide_unroll_stupid): Likewise.
4245 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4246 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4247 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4249 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4250 (print_loop): Print loop->unroll if set.
4251 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4252 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4253 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4255 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4256 loop->unroll is set and smaller than the trip count. Otherwise bypass
4257 entirely the heuristics if loop->unroll is set. Remove dead note.
4258 Fix off-by-one bug in other note.
4259 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4260 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4262 (tree_unroll_loops_completely): Make static.
4263 (pass_complete_unroll::execute): Use correct type for variable.
4264 (pass_complete_unrolli::execute): Fix formatting.
4265 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4267 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4269 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4270 enable when TARGET_PREFER_AVX128 is set.
4272 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4274 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4276 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4277 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4279 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4281 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4282 frequencies. Fix estimation of aggregate parameters.
4284 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4286 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4289 2017-11-23 Tom de Vries <tom@codesourcery.com>
4291 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4293 2017-11-23 Tom de Vries <tom@codesourcery.com>
4295 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4298 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4301 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4302 sibcall_value_pcrel_fdpic): Use local variable instead of
4304 (calli_tbr_rel): Add missing operand 2.
4305 (call_valuei_tbr_rel): Add missing operand 3.
4307 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4310 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4311 bitpos/bitsize covering the whole destination, use store_expr only if
4312 the complex mode is the same. Otherwise, use expand_normal and if
4313 it returns CONCAT, subreg each part separately instead of trying to
4314 subreg the whole result.
4316 2017-11-23 Richard Biener <rguenther@suse.de>
4318 PR tree-optimization/23094
4319 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4320 come here from walking over backedges in the first iteration.
4321 (vn_reference_lookup_3): Skip clobbers that store the same value.
4323 2017-11-23 Richard Biener <rguenther@suse.de>
4325 PR tree-optimization/81403
4326 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4327 a block we need a leader relative to.
4328 (phi_translate_1): For nary processing require a leader from
4329 get_representative_for given we run expression simplification
4330 using match-and-simplify. Remove previous fix.
4332 2017-11-22 Jeff Law <law@redhat.com>
4334 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4335 Use new method allocate_value_range rather than accessing the
4336 vrp_value_range_pool data member directly.
4337 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4338 to use extract_range_from_stmt method to avoid need for
4339 extract_range_from_assignment method.
4340 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4341 method rather than setting values_propgated data member directly.
4342 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4343 and values propagated data members and extract_range_from_assignment
4344 method. Reorder private data members to conform to standards.
4345 Add new methods set_lattice_propagation_complete and
4346 allocate_value_range.
4348 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4350 PR rtl-optimization/83030
4351 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4352 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4353 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4354 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4355 CROSSING_JUMP_P flag is already set before setting it.
4356 * cfgrtl.c (fixup_partition_crossing): Likewise.
4357 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4360 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4362 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4363 Handle the case where both arguments are using gen_const_vec_series.
4365 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4368 * pretty-print.c (pp_format): Move quoting implementation to
4369 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4370 to pass address of "quote" local.
4371 (pp_begin_quote): New function.
4372 (pp_end_quote): New function.
4373 * pretty-print.h (printer_fn): Convert penultimate param from bool
4375 (pp_begin_quote): New decl.
4376 (pp_end_quote): New decl.
4377 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4378 param from bool to bool *.
4379 * tree-diagnostic.h (default_tree_printer): Likewise.
4381 2017-11-22 Jeff Law <law@redhat.com>
4383 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4384 Initialize vr_values.
4385 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4386 vr_values via vr_values class instance rather than delegators.
4387 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4388 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4389 (evrp_range_analyzer::push_value_range): Likewise.
4390 (evrp_range_analyzer::pop_value_range): Likewise.
4391 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4392 most delegators. Those remaining are exposed as public interfaces.
4393 Make vr_values a pointer and private.
4394 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4396 (evrp_range_analyzer::get_vr_value): New method.
4397 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4398 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4399 (class evrp_dom_walker): Attach evrp_folder class, initialize
4400 it in the ctor. Remove temporary delegators.
4401 (evrp_dom_walker::before_dom_children): Call methods in attached
4402 evrp_range_analyzer class via class instance pointer. Use
4403 free value_range_constant_singleton to remove need for
4404 op_with_constant_singleton_value delegator method. Do not
4405 create a vrp_prop class instance for every call! Narrow
4406 scope of a couple variables.
4407 (evrp_dom_walker::cleanup): Call methods in attached
4408 evrp_range_analyzer class via class instance pointer.
4409 * vr-values.h (class vr_values): Privatize many methods and
4412 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4414 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4415 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4416 * expr.c (expand_expr_real_2): Likewise.
4417 * fold-const.c (const_unop): Likewise.
4418 * optabs-tree.c (optab_for_tree_code): Likewise.
4419 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4420 * tree-inline.c (estimate_operator_cost): Likewise.
4421 * tree-pretty-print.c (dump_generic_node): Likewise.
4422 (op_code_prio): Likewise.
4423 (op_symbol_code): Likewise.
4424 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4425 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4426 * internal-fn.c (direct_internal_fn_optab): New function.
4427 (direct_internal_fn_array, direct_internal_fn_supported_p
4428 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4429 * fold-const-call.c (fold_const_reduction): New function.
4430 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4432 * tree-vect-loop.c: Include internal-fn.h.
4433 (reduction_code_for_scalar_code): Rename to...
4434 (reduction_fn_for_scalar_code): ...this and return an internal
4436 (vect_model_reduction_cost): Take an internal_fn rather than
4438 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4440 (vectorizable_reduction): Use internal functions rather than tree
4441 codes for the reduction operation. Update calls to the functions
4443 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4444 Use calls to internal functions rather than REDUC tree codes.
4445 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4447 2017-11-22 Olivier Hainque <hainque@adacore.com>
4449 * config/vxworks.c (vxworks_override_options): Pick default
4450 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4451 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4452 DWARF_GNAT_ENCODINGS_DEFAULT.
4453 * config/vxworksae.h: Likewise.
4455 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4457 PR tree-optimization/83104
4458 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4461 2017-11-22 Marek Polacek <polacek@redhat.com>
4462 H.J. Lu <hongjiu.lu@intel.com>
4463 Jason Merrill <jason@redhat.com>
4468 * calls.c (initialize_argument_information): Call
4469 warn_parameter_passing_abi target hook.
4470 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4471 argument onto stack.
4472 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4473 * common.opt: Update -fabi-version description.
4474 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4475 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4477 (ix86_is_empty_record): New function.
4478 (ix86_warn_parameter_passing_abi): New function.
4479 (TARGET_EMPTY_RECORD_P): Redefine.
4480 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4481 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4482 * doc/tm.texi: Regenerated.
4483 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4484 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4485 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4486 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4487 instead of int_size_in_bytes.
4488 * expr.c (copy_blkmode_to_reg): Likewise.
4489 * function.c (aggregate_value_p): Return 0 for empty types.
4490 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4492 (locate_and_pad_parm): Call arg size_in_bytes instead
4494 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4495 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4496 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4498 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4499 (std_gimplify_va_arg_expr): Skip empty records. Call
4500 arg_size_in_bytes instead size_in_bytes.
4501 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4502 * tree-core.h (tree_type_common): Add empty_flag.
4503 (tree_decl_common): Update comments.
4504 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4506 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4507 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4509 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4510 * tree.c (default_is_empty_type): New function.
4511 (default_is_empty_record): New function.
4512 (arg_int_size_in_bytes): New function.
4513 (arg_size_in_bytes): New function.
4514 (get_ultimate_context): New function.
4515 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4516 TRANSLATION_UNIT_WARN_EMPTY_P.
4517 (default_is_empty_record, arg_int_size_in_bytes,
4518 arg_size_in_bytes, get_ultimate_context): Declare.
4520 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4522 * config/arm/arm.c (cmse_clear_registers): New function.
4523 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4524 code by call to cmse_clear_registers.
4525 (cmse_nonsecure_entry_clear_before_return): Likewise.
4527 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4529 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4530 (vdot_s32, vdotq_s32): New.
4531 (vdot_lane_u32, vdotq_lane_u32): New.
4532 (vdot_lane_s32, vdotq_lane_s32): New.
4535 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4538 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4539 for unsigned values with fewer HWIs than the precision.
4540 (test_overflow): New function.
4541 (wide_int_cc_tests): Call it.
4543 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4545 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4546 has the same mode class as Pmode.
4548 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4550 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4551 padding_bits_to_clear_ptr.
4552 (cmse_nonsecure_entry_clear_before_return): Likewise.
4554 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4556 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4557 auto_sbitap instead of integer bitfield to control register needing
4560 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4562 PR tree-optimization/83044
4563 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4564 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4565 that need the upper bound. Subtract offset from
4566 get_addr_base_and_unit_offset only if positive and subtract it
4567 before division by eltsize rather than after it.
4570 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4571 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4572 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4576 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4577 expand_binop with *mul_widen_optab, make sure at least one of the
4578 operands doesn't have VOIDmode.
4581 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4583 PR rtl-optimization/82044
4584 PR tree-optimization/82042
4585 * dse.c (record_store): Check for overflow.
4586 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4587 add_wild_read instead of clear_rhs_from_active_local_stores on
4588 overflow and log it into dump_file.
4590 2017-11-22 Richard Biener <rguenther@suse.de>
4592 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4593 fake edges to exit when looking for a place to insert.
4594 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4596 (insert_into_preds_of_block): Commit edge insertion immediately,
4597 assert that doesn't require new BBs.
4598 (fini_pre): Release expressions.
4599 (pass_pre::execute): Shuffle things around a bit, if the fn
4600 is too large do not compute AVAIL either as this is really the
4603 2017-11-22 Richard Biener <rguenther@suse.de>
4605 PR tree-optimization/83089
4606 * tree-if-conv.c (pass_if_conversion::execute): If anything
4607 changed reset SCEV and free the number of iteration estimates.
4609 2017-11-21 Martin Sebor <msebor@redhat.com>
4611 PR tree-optimization/82945
4612 * calls.h (warn_nonstring_bound): Remove unused function.
4614 2017-11-21 Martin Sebor <msebor@redhat.com>
4616 PR tree-optimization/82945
4617 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4618 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4619 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4621 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4622 * calls.h (get_attr_nonstring_decl): Declare new function.
4623 * doc/extend.texi (attribute nonstring): Update.
4624 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4625 get_attr_nonstring_decl and handle it.
4626 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4627 detection of nul-termination.
4628 (strlen_to_stridx): Change to a pointer.
4629 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4630 (pass_strlen::execute): Same.
4632 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4634 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4635 for the new option -mprefer-vector-width=[none|128|256|512].
4636 * config/i386/i386.c (ix86_target_string): remove old style options
4637 -mprefer-avx256 and make -mprefer-avx128 as alias.
4638 (ix86_option_override_internal): Apply defaults for the
4639 -mprefer-vector-width=[128|256] option.
4640 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4641 Implement macros to work with -mprefer-vector-width=.
4642 * config/i386/i386.opt: Implemented option
4643 -mprefer-vector-width=[none|128|256|512].
4644 * doc/invoke.texi: Documentation for
4645 -mprefer-vector-width=[none|128|256|512].
4647 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4649 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4651 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4653 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4655 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4657 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4659 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4660 * expr.c (expand_expr_real_2): Likewise.
4661 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4662 fold_binary_loc): Likewise.
4663 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4664 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4665 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4666 MINUS_EXPR transformations.
4667 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4668 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4669 * tree-inline.c (estimate_operator_cost): Likewise.
4670 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4671 op_symbol_code): Likewise.
4672 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4673 * vr-values.c (extract_range_from_binary_expr): Likewise.
4674 * varasm.c (initializer_constant_valid_p_1): Likewise.
4675 * tree.def: New tree code POINTER_DIFF_EXPR.
4677 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4679 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4680 integer suffix to movbe mnemonic.
4681 (*bswaphi2_movbe): Ditto.
4682 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4684 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4687 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4688 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4689 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4691 2017-11-21 Martin Liska <mliska@suse.cz>
4693 * tree-inline.c (expand_call_inline): Remove not needed
4696 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4697 Jakub Jelinek <jakub@redhat.com>
4700 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4701 Remove static keyword from f variable.
4703 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4705 PR tree-optimization/83086
4706 * gimple-ssa-store-merging.c
4707 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4708 rather than n.base_addr.
4710 2017-11-21 Martin Liska <mliska@suse.cz>
4712 PR rtl-optimization/82044
4713 PR tree-optimization/82042
4714 * dse.c (check_mem_read_rtx): Check for overflow.
4716 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4718 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4721 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4723 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4724 and its corresponding call arg location note.
4726 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4727 Andrew Burgess <andrew.burgess@embecosm.com>
4729 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4731 (arc_return_slot_offset): Likewise.
4732 (arc_eh_return_address_location): New declaration.
4733 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4734 (MUST_SAVE_REGISTER): Add exception handler case.
4735 (MUST_SAVE_RETURN_ADDR): Likewise.
4736 (arc_frame_pointer_required): Likewise.
4737 (arc_frame_pointer_needed): New function.
4738 (arc_compute_frame_size): Changed.
4739 (arc_expand_prologue): Likewise.
4740 (arc_expand_epilogue): Likewise.
4741 (arc_initial_elimination_offset): Likewise.
4742 (arc_return_slot_offset): Delete.
4743 (arc_eh_return_address_location): New function.
4744 (arc_builtin_setjmp_frame_value): Likewise.
4745 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4746 (EH_RETURN_STACKADJ_RTX): Define.
4747 (EH_RETURN_HANDLER_RTX): Likewise.
4748 * config/arc/arc.md (eh_return): Delete.
4750 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4752 * print-tree.h (debug_vec_tree): Remove prototype.
4753 * gdbinit.in (pvt): Remove macro.
4755 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4757 PR tree-optimization/83047
4758 * gimple-ssa-store-merging.c
4759 (imm_store_chain_info::output_merged_store): If the loads with the
4760 same vuse are in different basic blocks, for load_gsi pick a load
4761 location that is dominated by the other loads.
4764 * config/i386/i386.c (ix86_memmodel_check): Start
4765 -Winvalid-memory-model diagnostics with lowercase letter.
4768 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4769 set in_cold_section_p to the partition containing loc_list->first.
4770 When seeing loc_list->last_before_switch node, update secname and
4771 perform range_across_switch second partition handling only after that.
4774 * run-rtl-passes.c: Include debug.h.
4775 (run_rtl_passes): Call debug_hooks->assembly_start.
4776 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4780 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4781 OPTAB_DIRECT in calls to expand_simple_binop.
4783 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4786 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4788 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4791 * spellcheck.h (best_match::blithely_get_best_candidate): New
4794 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4796 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4797 with lower case letter.
4799 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4801 * config/i386/i386.md (bswaphi2): New expander.
4802 (*bswaphi2_movbe): New insn pattern.
4803 (bswaphi -> rorhi pepehole2): New peephole pattern.
4805 2017-11-20 Jeff Law <law@redhat.com>
4807 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4808 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4809 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4810 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4811 gimple-ssa-evrp-analyze.h.
4813 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4814 set BB_VISITED here.
4815 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4817 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4819 * tree-predcom.c: Add general comment on Store-Store chains.
4820 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4822 (get_chain_last_ref_at): Rename into...
4823 (get_chain_last_write_at): ...this.
4824 (get_chain_last_write_before_load): New function.
4825 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4826 CT_STORE_STORE when write reference is added.
4827 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4828 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4829 (initialize_root_vars_store_elim_1): Ditto.
4830 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4831 definition is created.
4832 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4833 chain by replacing it with dominant stored value.
4835 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4837 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4839 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4841 * vr-values.c (extract_range_from_binary_expr): Use a full range
4844 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4846 * config/arm/arm.md (R4_REGNUM): Define constant.
4847 (nonsecure_call_internal): Remove r4 clobber.
4848 (nonsecure_call_value_internal): Likewise.
4849 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4850 clobber and resequence match_operands.
4851 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4852 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4853 (nonsecure_call_value_reg_thumb2): Likewise.
4855 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4857 PR tree-optimization/78821
4858 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4859 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4860 to the address of the base rather than the base itself.
4861 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4862 (find_bswap_or_nop_finalize): New function.
4863 (find_bswap_or_nop): Use it.
4864 (bswap_replace): Return a tree rather than bool, change first
4865 argument from gimple * to gimple_stmt_iterator, allow inserting
4866 into an empty sequence, allow ins_stmt to be NULL - then emit
4867 all stmts into gsi. Fix up MEM_REF address gimplification.
4868 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4869 (struct store_immediate_info): Add N and INS_STMT non-static
4871 (store_immediate_info::store_immediate_info): Initialize them
4872 from newly added ctor args.
4873 (merged_store_group::apply_stores): Formatting fixes. Sort by
4875 (stmts_may_clobber_ref_p): For stores call also
4876 refs_anti_dependent_p.
4877 (gather_bswap_load_refs): New function.
4878 (imm_store_chain_info::try_coalesce_bswap): New method.
4879 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4880 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4881 (imm_store_chain_info::output_merged_store): Fail if number of
4882 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4883 and NOP_EXPR rhs_code.
4884 (pass_store_merging::process_store): Compute n and ins_stmt, if
4885 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4886 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4888 (pass_store_merging::execute): Calculate dominators.
4890 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4891 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4892 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4893 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4894 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4895 class pass_optimize_bswap, bswap_replace,
4896 pass_optimize_bswap::execute): Moved to ...
4897 * gimple-ssa-store-merging.c: ... this file.
4898 Include optabs-tree.h.
4899 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4900 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4901 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4902 anonymous namespace, remove static keywords.
4903 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4904 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4906 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4909 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4911 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4913 * vec.h (debug_helper): New function.
4914 (DEFINE_DEBUG_VEC): New macro.
4915 * hash-set.h (debug_helper): New function.
4916 (DEFINE_DEBUG_HASH_SET): New macro.
4917 * cfg.c (debug_slim (edge)): New function.
4918 Call DEFINE_DEBUG_VEC for edges.
4919 Call DEFINE_DEBUG_HASH_SET for edges.
4920 * cfghooks.c (debug_slim (basic_block)): New function.
4921 Call DEFINE_DEBUG_VEC for basic blocks.
4922 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4923 * print-tree.c (debug_slim): New function to handle trees.
4924 Call DEFINE_DEBUG_VEC for trees.
4925 Call DEFINE_DEBUG_HASH_SET for trees.
4926 (debug (vec<tree, va_gc>) &): Remove.
4927 (debug (<vec<tree, va_gc>) *): Remove.
4928 * print-rtl.c (debug_slim): New function to handle const_rtx.
4929 Call DEFINE_DEBUG_VEC for rtx_def.
4930 Call DEFINE_DEBUG_VEC for rtx_insn.
4931 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4932 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4933 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4934 (debug (vec<rtx_insn *> *ptr): Remove.
4935 (debug_insn_vector): Remove.
4936 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4938 2017-11-20 Tom de Vries <tom@codesourcery.com>
4940 PR rtl-optimization/82020
4941 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4942 IF_THEN_ELSE condition.
4944 2017-11-19 Jeff Law <law@redhat.com>
4946 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4947 of degenerates resulting from ignoring an edge.
4949 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4952 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4954 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4957 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4958 for uninitialied counts.
4960 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4963 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4965 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4968 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4970 2017-11-19 Tom de Vries <tom@codesourcery.com>
4972 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4975 2017-11-19 Tom de Vries <tom@codesourcery.com>
4977 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4980 2017-11-19 Tom de Vries <tom@codesourcery.com>
4982 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4985 2017-11-19 Tom de Vries <tom@codesourcery.com>
4987 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4989 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4991 2017-11-19 Tom de Vries <tom@codesourcery.com>
4993 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4995 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4996 semicolon after MCORE_EXPORT_NAME call.
4998 2017-11-19 Tom de Vries <tom@codesourcery.com>
5001 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5002 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5004 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5006 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5007 (cgraph_edge::clone): Cleanup updating of profile.
5008 * ipa-cp.c (update_profiling_info): Likewise.
5009 * ipa-inline-transform.c (inline_transform): Likewise.
5010 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5011 * ipa-split.c (execute_split_functions): Do not split when function
5013 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5014 * profile-count.c (profile_count::dump): Add global0.
5015 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5017 (profile_count::to_sreal_scale): Likewise.
5018 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5019 (profile_count::combine_with_ipa_count): New function.
5020 * profile-count.h (profile_guessed_global0adjusted): New.
5021 (profile_count::adjusted_zero): New.
5022 (profile_count::global0adjusted): New.
5023 (profile_count::combine_with_ipa_count): New.
5024 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5025 correct profile of return block of split functions.
5026 (copy_cfg_body): Remove unused profile_count.
5027 (copy_body): Likewise.
5028 (expand_call_inline): Update.
5029 (tree_function_versioning): Update.
5031 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5033 * hash-set.h (hash_set::empty): New.
5034 * tree-ssa-threadbackward.h: Delete.
5035 * tree-ssa-threadbackward.c (class thread_jumps): New.
5036 Move max_threaded_paths into class.
5037 (fsm_find_thread_path): Remove arguments that are now in class.
5038 (profitable_jump_thread_path): Rename to...
5039 (thread_jumps::profitable_jump_thread_path): ...this.
5040 (convert_and_register_jump_thread_path): Rename to...
5041 (thread_jumps::convert_and_register_current_path): ...this.
5042 (check_subpath_and_update_thread_path): Rename to...
5043 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5044 (register_jump_thread_path_if_profitable): Rename to...
5045 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5046 (handle_phi): Rename to...
5047 (thread_jumps::handle_phi): ...this.
5048 (handle_assignment): Rename to...
5049 (thread_jumps::handle_assignment): ...this.
5050 (fsm_find_control_statement_thread_paths): Rename to...
5051 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5052 (find_jump_threads_backwards): Rename to...
5053 (thread_jumps::find_jump_threads_backwards): ...this.
5054 Initialize path local data.
5055 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5056 from within thread_jumps class.
5057 (pass_early_thread_jumps::execute): Same.
5059 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5061 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5063 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5065 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5066 to canonical location.
5068 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5071 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5073 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5075 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5077 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5078 rcrt1.o%s/grcrt1.o%s for -static-pie.
5080 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5082 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5083 ix86_shift_rotate_cost): Break out from ...
5084 (ix86_rtx_costs): ... here.
5085 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5088 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5090 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5091 when entry block was promoted unlikely.
5092 (estimate_bb_frequencies): Increase frequency scale.
5093 * profile-count.h (profile_count): Export precision info.
5095 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5097 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5098 disturbin profile of entry block.
5100 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5102 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5105 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5107 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5108 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5110 2017-11-17 Jeff Law <law@redhat.com>
5112 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5113 from evrp_dom_walker class. Various methods moved into new class.
5114 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5115 (evrp_range_analyzer::enter): New method.
5116 (evrp_range_analyzer::leave): New method.
5117 (evrp_dom_walker): Remove delegators no longer needed by this class.
5118 Replace vr_values data member with evrp_range_analyzer
5120 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5121 method extracted from evrp_dom_walker::before_dom_children.
5122 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5123 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5125 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5126 Add private copy constructor and move assignment operators.
5127 Privatize methods and class data where trivially possible.
5128 (evrp_dom_walker::cleanup): New function, extracted from
5129 execute_early_vrp. Simplify access to class data.
5131 * vr-values.h (get_output_for_vrp): Prototype.
5132 * vr-values.c (get_output_for_vrp): New function extracted from
5133 vrp_visit_assignment_or_call and extract_range_from_stmt.
5134 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5136 2017-11-17 Luis Machado <luis.machado@linaro.org>
5138 * config/aarch64/aarch64.c
5139 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5140 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5141 tune_params::AUTOPREFETCHER_WEAK.
5143 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5146 * config/arm/arm.c (arm_valid_target_attribute_rec):
5147 Parse "arch=" and "+<ext>".
5148 (arm_valid_target_attribute_tree): Re-init global options.
5149 (arm_option_override): Make non-static.
5150 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5151 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5152 (__ARM_FEATURE_CRC32): Support undef.
5153 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5154 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5156 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5158 * gdbinit.in (break-on-diagnostic): New command.
5160 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5162 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5163 return to indirect jump for EH return if control-flow
5164 protection is enabled. Change explicit 'false' argument in
5165 pro_epilogue_adjust_stack with a value of
5167 * config/i386/i386.md (simple_return_indirect_internal):
5168 Remove SImode restriction to support 64-bit.
5170 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5172 * combine.c (added_notes_insn): New.
5173 (try_combine): Handle added_notes_insn like added_links_insn.
5174 Rewrite return value code.
5175 (distribute_notes): Set added_notes_insn to the earliest insn we added
5178 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5180 PR rtl-optimization/82621
5181 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5182 dest of one of those SETs is unused.
5184 2017-11-17 Richard Biener <rguenther@suse.de>
5187 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5188 * tree-pretty-print.c (dump_generic_node): Handle
5189 annot_expr_parallel_kind.
5190 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5191 * gimplify.c (gimple_boolify): Likewise.
5193 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5195 * config.gcc (extra_headers): Add cet.h for x86 targets.
5196 * config/i386/cet.h: New file.
5197 * doc/install.texi: Add --enable-cet/--disable-cet.
5199 2017-11-17 Richard Biener <rguenther@suse.de>
5201 PR tree-optimization/83017
5202 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5203 (gen_parallel_loop): Properly count iterations.
5204 (parallelize_loops): Handle loop->can_be_parallel independent
5205 of flag_loop_parallelize_all. Make static profitability test match
5207 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5208 * invoke.texi (parloops-min-per-thread): Document.
5210 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5212 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5213 upstreaming review comments.
5215 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5217 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5218 with fast unaligned access.
5219 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5221 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5223 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5224 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5226 2017-11-17 Richard Biener <rguenther@suse.de>
5228 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5229 folding of references.
5231 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5234 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5235 of replacing call to strncmp with corresponding call to strcmp when
5238 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5240 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5241 option prefer-avx256 for skylake-avx512 configuration.
5242 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5243 (get_builtin_code_for_version): Ditto.
5245 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5246 Monk Chiang <sh.chiang04@gmail.com>
5248 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5249 (FIXED_REGISTERS): Reserve more register numbers.
5250 (CALL_USED_REGISTERS): Likewise.
5251 (REG_ALLOC_ORDER): Likewise.
5252 (REG_CLASS_CONTENTS): Likewise.
5253 (REGISTER_NAMES): Likewise.
5255 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5256 Kito Cheng <kito.cheng@gmail.com>
5258 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5260 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5262 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5264 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5265 parameter list for vec_splats.
5267 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5269 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5271 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5273 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5276 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5277 with the required version of automake 1.11.6.
5279 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5281 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5283 (DU_C2_3_power9): Correct reservation combinations.
5284 (FP_DIV_power9, VEC_DIV_power9): New.
5285 (power9-alu): Split out rotate/shift...
5286 (power9-rot): ...to here, correct dispatch resource.
5287 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5289 (power9-fp): Correct latency.
5290 (power9-sdiv): Add div/sqrt resource.
5291 (power9-ddiv): Correct latency, add div/sqrt resource.
5292 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5293 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5295 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5297 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5299 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5300 switch statement mapping KF built-ins to TF built-ins if we don't
5301 have the proper ISA 3.0 assembler support.
5303 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5305 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5306 (lower_emutls_function_body): Do not compute it.
5308 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5310 * ipa-split.c (split_bb_info): Turn time to sreal.
5311 (split_point): Likewise.
5312 (dump_split_point): Likewise.
5313 (fine_split_points): Likewise.
5314 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5316 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5318 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5321 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5323 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5325 (expensive_function_p): Remove useless assert.
5326 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5328 2017-11-16 Martin Sebor <msebor@redhat.com>
5330 PR tree-optimization/82588
5331 PR tree-optimization/82583
5332 * tree-vrp.c (check_array_ref): Handle flexible array members,
5333 string literals, and inner indices.
5334 (search_for_addr_array): Add detail to diagnostics.
5336 2017-11-16 Nathan Sidwell <nathan@acm.org>
5340 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5341 (SET_DECL_ASSEMBLER_NAME): Forward to
5342 overwrite_decl_assembler_name.
5343 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5344 (overwrite_decl_assembler_name): Declare.
5345 * tree.c (overwrite_decl_assembler_name): New.
5346 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5347 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5348 (LANG_HOOKS_INITIALIZER): Add it.
5349 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5350 * langhooks.c (lhd_set_decl_assembler_name): Use
5351 SET_DECL_ASSEMBLER_NAME.
5352 (lhd_overwrite_decl_assembler_name): Default implementation.
5354 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5355 Jackson Woodruff <jackson.woodruff@arm.com>
5357 PR tree-optimization/71026
5358 * match.pd: Canonicalize constant multiplies in division.
5360 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5362 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5363 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5364 Remove OPT_fomit_frame_pointer entry.
5365 * common/config/arc/arc-common.c: Likewise.
5366 * common/config/arm/arm-common.c: Likewise.
5367 * common/config/avr/avr-common.c: Likewise.
5368 * common/config/c6x/c6x-common.c: Likewise.
5369 * common/config/cr16/cr16-common.c: Likewise.
5370 * common/config/cris/cris-common.c: Likewise.
5371 * common/config/epiphany/epiphany-common.c: Likewise.
5372 * common/config/fr30/fr30-common.c: Likewise.
5373 * common/config/frv/frv-common.c: Likewise.
5374 * common/config/ia64/ia64-common.c: Likewise.
5375 * common/config/iq2000/iq2000-common.c: Likewise.
5376 * common/config/lm32/lm32-common.c: Likewise.
5377 * common/config/m32r/m32r-common.c: Likewise.
5378 * common/config/mcore/mcore-common.c: Likewise.
5379 * common/config/microblaze/microblaze-common.c: Likewise.
5380 * common/config/mips/mips-common.c: Likewise.
5381 * common/config/mmix/mmix-common.c: Likewise.
5382 * common/config/mn10300/mn10300-common.c: Likewise.
5383 * common/config/nios2/nios2-common.c: Likewise.
5384 * common/config/pa/pa-common.c: Likewise.
5385 * common/config/pdp11/pdp11-common.c: Likewise.
5386 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5387 * common/config/riscv/riscv-common.c: Likewise.
5388 * common/config/rs6000/rs6000-common.c: Likewise.
5389 * common/config/rx/rx-common.c: Likewise.
5390 * common/config/s390/s390-common.c: Likewise.
5391 * common/config/sh/sh-common.c: Likewise.
5392 * common/config/sparc/sparc-common.c: Likewise.
5393 * common/config/tilegx/tilegx-common.c: Likewise.
5394 * common/config/tilepro/tilepro-common.c: Likewise.
5395 * common/config/v850/v850-common.c: Likewise.
5396 * common/config/visium/visium-common.c: Likewise.
5397 * common/config/xstormy16/xstormy16-common.c: Likewise.
5398 * common/config/xtensa/xtensa-common.c: Likewise.
5399 * invoke.texi (-fomit-frame-pointer): Update documentation.
5401 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5403 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5405 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5407 * cfg.c (scale_bbs_frequencies_int,
5408 cale_bbs_frequencies_gcov_type): Remove.
5409 * cfg.h (scale_bbs_frequencies_int,
5410 cale_bbs_frequencies_gcov_type): Remove.
5412 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5414 * tree-ssa-loop-manip.c
5415 (scale_dominated_blocks_in_loop): Update to profile counts.
5416 (tree_transform_and_unroll_loop): Likewise.
5418 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5420 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5421 scale_bbs_frequencies_int.
5423 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5425 * final.c (compute_alignments): Use counts rather than frequencies.
5427 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5429 * cfgloopanal.c: Include sreal.h
5430 (average_num_loop_insns): Use counts and sreal for accounting.
5432 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5434 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5437 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5438 Kito Cheng <kito.cheng@gmail.com>
5440 * config/nds32/constraints.md: Provide more constraints.
5441 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5442 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5443 support constraints usage.
5445 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5447 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5449 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5450 Kito Cheng <kito.cheng@gmail.com>
5452 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5453 * config/nds32/nds32.opt: Refine the layout.
5454 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5455 TARGET_EXT_STRING): Support new options.
5456 * config/nds32/nds32.h: Likewise.
5457 * config/nds32/nds32.md: Likewise.
5458 * config/nds32/nds32-predicates.c: Likewise.
5459 * config/nds32/constraints.md: Likewise.
5460 * common/config/nds32/nds32-common.c: Likewise.
5462 2017-11-16 Julia Koval <julia.koval@intel.com>
5465 * config/i386/gfniintrin.h: Add sse check.
5466 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5468 2017-11-16 Julia Koval <julia.koval@intel.com>
5470 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5471 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5472 (ix86_handle_option): Handle -mavx512vbmi2.
5473 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5474 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5475 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5476 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5477 (ix86_valid_target_attribute_inner_p): Ditto.
5478 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5479 * config/i386/i386.opt (mavx512vbmi2): New option.
5480 * doc/invoke.texi: Add new option.
5482 2017-11-16 Julia Koval <julia.koval@intel.com>
5484 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5485 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5486 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5487 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5488 _mm512_gf2p8mul_epi8): New intrinsics.
5489 * config/i386/i386-builtin-types.def
5490 (V64QI_FTYPE_V64QI_V64QI): New type.
5491 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5492 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5493 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5494 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5495 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5496 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5498 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5500 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5501 explicitly as a stream of bytes.
5503 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5505 * config/rs6000/altivec.h (vec_xst_be): New #define.
5506 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5507 and externalize from *altivec_vperm_<mode>_internal.
5508 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5510 (XL_BE_V8HI): Likewise.
5511 (XL_BE_V4SI): Likewise.
5512 (XL_BE_V4SI): Likewise.
5513 (XL_BE_V2DI): Likewise.
5514 (XL_BE_V4SF): Likewise.
5515 (XL_BE_V2DF): Likewise.
5516 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5517 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5518 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5519 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5520 VSX_BUILTIN_VEC_XST_BE.
5521 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5522 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5524 (altivec_init_builtins): Replace conditional calls to def_builtin
5525 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5526 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5527 with unconditional calls. Remove calls to def_builtin for
5528 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5529 __builtin_vec_xst_be.
5530 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5531 to define_expand, and add alternate RTL generation for P8.
5532 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5533 vsx_ld_elemrev_v8hi.
5534 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5535 add alternate RTL generation for P8.
5536 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5537 vsx_ld_elemrev_v16qi.
5538 (vsx_st_elemrev_v8hi): Convert define_insn
5539 to define_expand, and add alternate RTL generation for P8.
5540 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5541 vsx_st_elemrev_v8hi.
5542 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5543 add alternate RTL generation for P8.
5544 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5545 vsx_st_elemrev_v16qi.
5547 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5550 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5551 TARGET_AVX512ER check.
5552 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5553 neither -mzeroupper nor -mno-zeroupper is used and
5554 TARGET_EMIT_VZEROUPPER is set.
5555 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5556 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5558 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5560 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5561 folding of vector compares.
5562 (fold_build_vec_cmp): New helper function.
5563 (fold_compare_helper): New helper function.
5564 (builtin_function_type): Add compare builtins to the list of functions
5565 having unsigned arguments. Cosmetic updates to comment indentation.
5566 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5567 the not+eq combination.
5569 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5571 PR tree-optimization/82726
5572 PR tree-optimization/70754
5573 * tree-predcom.c (order_drefs_by_pos): New function.
5574 (combine_chains): Move code setting has_max_use_after to...
5575 (try_combine_chains): ...here. New parameter. Sort combined chains
5576 according to position information.
5577 (tree_predictive_commoning_loop): Update call to above function.
5578 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5580 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5582 PR tree-optimization/82726
5584 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5586 PR tree-optimization/70754
5587 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5588 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5589 combined stmt before it if not NULL.
5590 (combine_chains): Process refs reversely and compute dominance point
5594 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5596 PR tree-optimization/79663
5597 * tree-predcom.c (combine_chains): Process refs in reverse order
5598 only for ZERO length chains, and add explaining comment.
5600 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5602 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5603 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5604 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5606 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5608 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5610 2017-11-15 Martin Liska <mliska@suse.cz>
5612 * tree-cfg.c (pass_warn_function_return::execute):
5613 Compare warn_return_type for greater than zero.
5615 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5619 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5620 to return true on Xeon and not on Xeon Phi.
5621 (ix86_check_avx256_register): Changed to ...
5622 (ix86_check_avx_upper_register): ... this. Add extra check for
5623 VALID_AVX512F_REG_OR_XI_MODE.
5624 (ix86_avx_u128_mode_needed): Changed
5625 ix86_check_avx256_register to ix86_check_avx_upper_register.
5626 (ix86_check_avx256_stores): Changed to ...
5627 (ix86_check_avx_upper_stores): ... this. Changed
5628 ix86_check_avx256_register to ix86_check_avx_upper_register.
5629 (ix86_avx_u128_mode_after): Changed
5630 avx_reg256_found to avx_upper_reg_found. Changed
5631 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5632 (ix86_avx_u128_mode_entry): Changed
5633 ix86_check_avx256_register to ix86_check_avx_upper_register.
5634 (ix86_avx_u128_mode_exit): Ditto.
5635 * config/i386/i386.h: (host_detect_local_cpu): New define.
5637 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5639 * config/arm/xgene1.md (xgene1): Split into automatons
5640 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5641 (xgene1_f_load): Adjust reservations and/or types.
5642 (xgene1_f_store): Likewise.
5643 (xgene1_load_pair): Likewise.
5644 (xgene1_store_pair): Likewise.
5645 (xgene1_fp_load1): Likewise.
5646 (xgene1_load1): Likewise.
5647 (xgene1_store1): Likewise.
5648 (xgene1_move): Likewise.
5649 (xgene1_alu): Likewise.
5650 (xgene1_simd): Likewise.
5651 (xgene1_bfm): Likewise.
5652 (xgene1_neon_load1): Likewise.
5653 (xgene1_neon_store1): Likewise.
5654 (xgene1_neon_logic): Likewise.
5655 (xgene1_neon_st1): Likewise.
5656 (xgene1_neon_ld1r): Likewise.
5657 (xgene1_alu_cond): Added.
5658 (xgene1_shift_reg): Likwise.
5659 (xgene1_bfx): Likewise.
5660 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5662 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5665 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5667 (can_widen_mult_without_libcall): New function.
5668 (expand_mul_overflow): If only checking unsigned mul overflow,
5669 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5670 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5671 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5672 (expand_DIVMOD): Formatting fix.
5673 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5674 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5675 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5677 PR tree-optimization/82977
5678 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5679 constructed temporary to strlen_to_stridx.put.
5681 2017-11-15 Martin Liska <mliska@suse.cz>
5683 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5684 * configure: Regenerate.
5686 2017-11-15 Martin Liska <mliska@suse.cz>
5689 * config/sh/sh-mem.cc: Use proper probability for
5692 2017-11-14 Jeff Law <law@redhat.com>
5694 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5695 the red zone for stack_clash_protection_final_dynamic_probe targets
5696 when the total dynamic stack size is zero bytes.
5698 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5699 blocks is post order.
5701 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5703 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5704 * final.c (rest_of_clean_state): Set it for the
5705 -fcompare-debug dump.
5706 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5707 class when TDF_COMPARE_DEBUG is set.
5709 * dwarf2out.c (gen_producer_string): Discard
5712 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5715 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5716 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5717 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5718 (__TGMATH_REAL_2_3): Remove macros.
5719 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5720 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5722 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5724 (remquo): Define using __TGMATH_REAL_3.
5726 2017-11-14 Jeff Law <law@redhat.com>
5728 * vr-values.c: New file with contents extracted from tree-vrp.c.
5729 * Makefile.in (OBJS): Add vr-values.o
5730 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5731 (set_value_range, set_and_canonicalize_value_range): Likewise.
5732 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5733 (value_range_constant_singleton, symbolic_range_p): Likewise.
5734 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5735 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5736 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5737 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5738 (find_case_label_range, find_case_label_index): Likewise.
5739 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5740 (range_int_cst_singleton_p, value_inside_range): Likewise.
5741 (get_single_symbol): Likewise.
5742 (switch_update): Move structure definition here.
5743 (to_remove_edges, to_update_switch_stmts): Provide externs.
5744 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5745 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5746 (vrp_val_is_min, set_value_range): Likewise.
5747 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5748 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5749 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5750 (range_is_nonnull, range_int_cst_p): Likewwise.
5751 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5752 (get_single_symbol, operand_less_p): Likewise
5753 (compare_values_warnv, compare_values): Likewise.
5754 (value_inside_range, value_range_constant_singleton): Likewise.
5755 (zero_nonzero_bitgs_from_vr): Likewise.
5756 (extract_range_from_binary_expr_1): Likewise.
5757 (overflow_comparison_p): Likewise.
5758 (to_remove_edges, to_update_switch_stmts): Likewise.
5759 (find_case_label-index, find_case_label_range): Likewise.
5760 (switch_update, set_value_range_to_nonnegative): Remove.
5761 (set_value_range_to_truthvalue): Likewise.
5762 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5763 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5764 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5765 (find_case_label_ranges, test_for_singularity): Likewise.
5766 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5767 (x_vr_values): Move to its remaining use site.
5769 2017-11-10 Jeff Law <law@redhat.com>
5771 * vr-values.h (VR_INITIALIZER): Move #define here.
5772 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5773 * Makefile.in (OBJS): Add tree-evrp.o
5774 * tree-vrp.h (assert_info): Move structure definition here.
5775 (set_value_range_to_varying): Prototype.
5776 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5777 (infer_value_range, register_edge_assert_for): Likewise.
5778 (stmt_interesting_for_vrp): Likewise.
5779 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5780 (set_value_range_to_varying): No longer static.
5781 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5782 (infer_value_range, register_edge_assert_for): Likewise.
5784 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5786 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5787 generate the XXBRD instruction.
5789 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5790 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5791 as being compatible if -mabi=ieeelongdouble.
5792 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5793 to setup float128 built-ins with hardware support.
5794 (BU_FLOAT128_HW_2): Likewise.
5795 (BU_FLOAT128_HW_3): Likewise.
5796 (BU_FLOAT128_HW_VSX_1): Likewise.
5797 (BU_FLOAT128_HW_VSX_2): Likewise.
5798 (scalar_extract_expq): Change float128 built-in functions to
5799 accommodate having both KFmode and TFmode functions. Use the
5800 KFmode variant as the default.
5801 (scalar_extract_sigq): Likewise.
5802 (scalar_test_neg_qp): Likewise.
5803 (scalar_insert_exp_q): Likewise.
5804 (scalar_insert_exp_qp): Likewise.
5805 (scalar_test_data_class_qp): Likewise.
5806 (sqrtf128_round_to_odd): Delete processing the round to odd
5807 built-in functions as special built-in functions, and define them
5808 as float128 built-ins. Use the KFmode variant as the default.
5809 (truncf128_round_to_odd): Likewise.
5810 (addf128_round_to_odd): Likewise.
5811 (subf128_round_to_odd): Likewise.
5812 (mulf128_round_to_odd): Likewise.
5813 (divf128_round_to_odd): Likewise.
5814 (fmaf128_round_to_odd): Likewise.
5815 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5816 support for KFmode and TFmode xststdcqp calls.
5817 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5818 point, switch the built-in handlers for the get/set float128
5819 exponent, get float128 mantissa, float128 test built-ins, and the
5820 float128 round to odd built-in functions. Eliminate creating the
5821 float128 round to odd built-in functions as special built-ins.
5822 (rs6000_init_builtins): Eliminate special creation of the float128
5823 round to odd built-in functions.
5824 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5825 function insns to support both TFmode and KFmode variants.
5826 (xsxsigqp_<mode>): Likewise.
5827 (xsiexpqpf_<mode>): Likewise.
5828 (xsiexpqp_<mode>): Likewise.
5829 (xststdcqp_<mode>): Likewise.
5830 (xststdcnegqp_<mode>): Likewise.
5831 (xststdcqp_<mode>): Likewise.
5833 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5835 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5836 unused path_in_freq_ptr parameter.
5837 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5839 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5841 * ipa-inline.c (edge_badness): Dump sreal frequency.
5842 (compute_inlined_call_time): Match natural implementaiton ...
5843 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5844 forgotten division by CGRAPH_FREQ_BASE.
5846 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5848 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5849 Solaris 11. Update comment.
5850 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5852 * config/sol2.h (STARTFILE_SPEC): Likewise.
5853 * configure: Regenerate.
5855 2017-11-14 Carl Love <cel@us.ibm.com>
5857 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5858 le_ and be_ prefixes to swap* variables. Remove
5859 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5862 2017-11-14 Jason Merrill <jason@redhat.com>
5864 Support GTY((cache)) on hash_map.
5865 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5866 (ggc_cache_remove): Override it instead of ggc_mx.
5867 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5868 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5869 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5870 (simple_cache_map_traits): Override maybe_mx.
5871 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5872 (hash_map): Friend gt_cleare_cache.
5873 (gt_cleare_cache): New.
5874 * tree.h (tree_cache_traits): New hash_map traits class.
5875 (tree_cache_map): New typedef.
5877 2017-11-14 Richard Biener <rguenther@suse.de>
5879 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5880 paramter and handling.
5881 (cleanup_control_flow_bb): Likewise.
5882 (cleanup_control_flow_pre): New helper performing a DFS walk
5883 to call cleanup_control_flow_bb in PRE order.
5884 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5885 via cleanup_control_flow_pre.
5887 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5889 * config/aarch64/aarch64-simd.md
5890 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5891 (*aarch64_simd_bsl<mode>_alt): Likewise.
5892 (aarch64_simd_bsldi_internal): New.
5893 (aarch64_simd_bsldi_alt): Likewise.
5895 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5897 * tracer.c (better_p): Do not compare frequencies.
5898 * reg-stack.c (better_edge): Likewise.
5899 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5902 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5904 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5905 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5906 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5907 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5908 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5909 * predict.c (maybe_hot_frequency_p): Inline to ...
5910 (maybe_hot_count_p): ... here; rewrite to counts.
5911 (counts_to_freqs): Rename to ...
5912 (update_max_bb_count): ... this one.
5913 (expensive_function_p): Use counts.
5914 (estimate_bb_frequencies): Update.
5915 (rebuild_frequencies): Update.
5916 * predict.h (counts_to_freqs): Rename to ...
5917 (update_max_bb_count): ... this one.
5918 * profile.c (compute_branch_probabilities): Add debug info
5919 * tree-inline.c (expand_call_inline): Update debug info.
5920 (optimize_inline_calls): Use update_max_bb_count..
5921 (tree_function_versioning): Use update_max_bb_count..
5922 * value-prof.c (gimple_value_profile_transformations):
5923 Do not use update_max_bb_count.
5925 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5927 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5928 always use frequencies.
5930 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5932 * bb-reorder.c: Remove frequencies from comments.
5933 (better_edge_p): Use profile counts.
5934 (find_traces): Dump profile counts.
5935 (rotate_loop): Use profile counts.
5936 (find_traces_1_round): Likewise.
5937 (connect_better_edge_p): Use counts instead of probabilities for
5939 (copy_bb_p): Drop early check for non-0 frequency.
5940 (sanitize_hot_paths): Update comments.
5942 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5944 * ipa-split.c (struct split_point): Add count.
5945 (consider_split): Do not compute incoming frequency; compute incoming
5946 count and store it to split_point.
5947 (split_function): Set count of the call to split part correctly.
5949 2017-11-13 Carl Love <cel@us.ibm.com>
5951 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5953 2017-11-13 Tom Tromey <tom@tromey.com>
5955 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5957 2017-11-13 Carl Love <cel@us.ibm.com>
5959 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5960 Add support for builtins:
5961 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5962 vector {un,}signed {char,int,short},
5963 vector {un,}signed {char,int,short}) arguments.
5964 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5965 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5966 Add BU_P9V_AV_2 expansions for the builtins.
5967 * config/rs6000/altivec.h (vec_first_match_index,
5968 vec_first_mismatch_index, vec_first_match_or_eos_index,
5969 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5970 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5971 new extern declaration.
5972 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5973 * config/rs6000/vsx.md (first_match_index_<mode>,
5974 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5975 first_mismatch_or_eos_index_<mode>): Add define expand.
5976 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5977 * doc/extend.texi: Update the built-in documenation file for the new
5980 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5982 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5983 into the min/max operations for _Float<N> and _Float<N>X types.
5985 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5988 * dwarf2out.c (do_eh_frame): New static variable.
5989 (dwarf2out_begin_prologue): Set it.
5990 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5992 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5994 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5996 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5998 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6001 * debug.h (dwarf2out_do_eh_frame): Declare.
6002 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6003 (dwarf2out_do_frame): Use it.
6004 (dwarf2out_do_cfi_asm): Likewise.
6005 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6006 (dwarf2out_assembly_start): Likewise.
6007 (dwarf2out_begin_prologue): Fix comment.
6008 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6009 if the target needs either debug or unwind DWARF2 info.
6010 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6011 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6012 -ftrapping-math, -ftrapv and -fwrapv.
6014 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6016 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6017 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6018 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6019 (estimate_edge_size_and_time): Likewise.
6020 (ipa_merge_fn_summary_after_inlining): Likewise.
6021 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6022 (compute_uninlined_call_time): Use sreal_frequency.
6023 (compute_inlined_call_time): Likewise.
6024 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6025 * profile-count.c: Include sreal.h.
6026 (profile_count::to_sreal_scale): New.
6027 * profile-count.h: Forward declare sreal.
6028 (profile_count::to_sreal_scale): Declare.
6030 2017-11-13 Nathan Sidwell <nathan@acm.org>
6032 * diagnostic.c (maybe_line_and_column): New.
6033 (diagnostic_get_location_text): Use it.
6034 (diagnostic_report_current_module): Likewise.
6035 (test_diagnostic_get_location_text): Add tests.
6037 2017-11-13 Luis Machado <luis.machado@linaro.org>
6039 * doc/md.texi (Specifying processor pipeline description): Fix
6040 incorrect latency for the div instruction example.
6042 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6044 PR tree-optimization/78821
6045 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6046 that bit_not_p is the same.
6047 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6048 (split_group): Count precisely bit_not_p bits in each statement.
6049 (invert_op): New function.
6050 (imm_store_chain_info::output_merged_store): Use invert_op to
6051 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6052 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6054 2017-11-13 Martin Liska <mliska@suse.cz>
6056 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6057 (struct source_info): Likewise.
6058 (add_branch_counts): Likewise.
6059 (add_line_counts): Likewise.
6060 (function_summary): Likewise.
6061 (output_intermediate_line): Likewise.
6062 (generate_results): Likewise.
6064 2017-11-13 Martin Liska <mliska@suse.cz>
6066 * gcov.c (struct block_info): Remove typedef for block_t.
6067 (struct line_info): Likewise.
6068 (line_info::has_block): Likewise.
6069 (EXIT_BLOCK): Likewise.
6070 (unblock): Likewise.
6071 (circuit): Likewise.
6072 (get_cycles_count): Likewise.
6073 (process_file): Likewise.
6074 (read_graph_file): Likewise.
6075 (solve_flow_graph): Likewise.
6076 (find_exception_blocks): Likewise.
6077 (add_line_counts): Likewise.
6078 (accumulate_line_info): Likewise.
6079 (output_line_details): Likewise.
6081 2017-11-13 Martin Liska <mliska@suse.cz>
6083 * gcov.c (struct arc_info): Remove typedef for arc_t.
6084 (struct line_info): Likewise.
6085 (add_branch_counts): Likewise.
6086 (output_branch_count): Likewise.
6087 (function_info::~function_info): Likewise.
6088 (circuit): Likewise.
6089 (output_intermediate_line): Likewise.
6090 (read_graph_file): Likewise.
6091 (solve_flow_graph): Likewise.
6092 (find_exception_blocks): Likewise.
6093 (add_line_counts): Likewise.
6094 (accumulate_line_info): Likewise.
6095 (output_line_details): Likewise.
6096 (output_function_details): Likewise.
6098 2017-11-13 Martin Liska <mliska@suse.cz>
6100 * gcov.c (struct function_info): Remove typedef for function_t.
6101 (struct source_info): Likewise.
6102 (source_info::get_functions_at_location): Likewise.
6103 (solve_flow_graph): Likewise.
6104 (find_exception_blocks): Likewise.
6105 (add_line_counts): Likewise.
6106 (output_intermediate_file): Likewise.
6107 (process_file): Likewise.
6108 (generate_results): Likewise.
6109 (release_structures): Likewise.
6110 (read_graph_file): Likewise.
6111 (read_count_file): Likewise.
6112 (accumulate_line_counts): Likewise.
6113 (output_lines): Likewise.
6115 2017-11-13 Martin Liska <mliska@suse.cz>
6117 * gcov.c (function_info::function_info): Remove num_counts
6118 and add vector<gcov_type>.
6119 (function_info::~function_info): Use the vector.
6120 (process_file): Likewise.
6121 (read_graph_file): Likewise.
6122 (read_count_file): Likewise.
6123 (solve_flow_graph): Likewise.
6125 2017-11-13 Martin Liska <mliska@suse.cz>
6127 * gcov.c (function_info::is_artificial): New function.
6128 (process_file): Erase all artificial early.
6129 (generate_results): Skip as all artificial are already
6132 2017-11-13 Martin Liska <mliska@suse.cz>
6134 * gcov.c (read_graph_file): Store to global vector of functions.
6135 (read_count_file): Iterate the vector.
6136 (process_file): Likewise.
6137 (generate_results): Likewise.
6138 (release_structures): Likewise.
6140 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6142 PR tree-optimization/82954
6143 * gimple-ssa-store-merging.c
6144 (imm_store_chain_info::coalesce_immediate_stores): If
6145 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6147 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6149 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6150 Upddate call to ENDIAN_LANE_N.
6151 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6152 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6153 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6154 and use aarch64_endian_lane_rtx.
6155 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6157 2017-11-12 Tom de Vries <tom@codesourcery.com>
6159 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6161 2017-11-12 Tom de Vries <tom@codesourcery.com>
6163 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6165 2017-11-12 Tom de Vries <tom@codesourcery.com>
6167 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6169 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6170 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6171 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6173 2017-11-11 Martin Sebor <msebor@redhat.com>
6176 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6179 * prefic.c (translate_name): Replace strncpy with memcpy to
6180 avoid -Wstringop-truncation.
6182 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6184 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6186 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6188 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6190 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6192 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6195 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6197 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6198 enable generating XXBRH if the value is in a vector register.
6199 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6200 value is in a vector register.
6201 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6202 register to register bswap64's instead of doing the GPR sequence
6203 used on previous machines.
6204 (bswapdi2_xxbrd): New insn.
6205 (bswapdi2_reg): Disallow on ISA 3.0.
6206 (register to register bswap64 splitter): Do not split the insn on
6207 ISA 3.0 systems that use XXBRD.
6209 2017-11-10 Martin Sebor <msebor@redhat.com>
6212 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6214 (find_subframework_file): Same.
6216 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6218 * auto-profile.c (afdo_indirect_call): Drop frequency.
6219 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6220 (cgraph_node::create_edge): Drop frequency argument.
6221 (cgraph_node::create_indirect_edge): Drop frequency argument.
6222 (cgraph_edge::make_speculative): Drop frequency arguments.
6223 (cgraph_edge::resolve_speculation): Do not update frequencies
6224 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6225 (cgraph_node::dump): Check consistency in IPA mode.
6226 (cgraph_edge::maybe_hot_p): Use IPA counter.
6227 (cgraph_edge::verify_count_and_frequency): Rename to ...
6228 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6229 (cgraph_node::verify_node): Update.
6230 * cgraph.h (struct cgraph_edge): Drop frequency.
6231 (cgraph_edge::frequency): New function.
6232 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6234 (cgraph_edge::rebuild_edges): Likewise.
6235 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6236 (duplicate_thunk_for_node): Do not pass frequency.
6237 (cgraph_node::create_clone): Scale only counts.
6238 (cgraph_node::create_virtual_clone): Do not pass frequency.
6239 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6240 (cgraph_node::create_version_clone): Do not pass frequency.
6241 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6242 (cgraph_node::expand_thunk): Do not pass frequency.
6243 (cgraph_node::create_wrapper): Do not pass frequency.
6244 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6246 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6247 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6248 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6249 (gather_caller_stats): Use frequency function.
6250 (ipcp_cloning_candidate_p): Use frequency function.
6251 (ipcp_propagate_stage): Use frequency function.
6252 (get_info_about_necessary_edges): Use frequency function.
6253 (update_profiling_info): Update only IPA profile.
6254 (update_specialized_profile): Use frequency functoin.
6255 (perhaps_add_new_callers): Update only IPA profile.
6256 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6257 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6258 (dump_ipa_call_summary): Use frequency function.
6259 (estimate_edge_size_and_time): Use frequency function.
6260 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6261 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6262 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6263 (update_noncloned_counts): ... ths one; scale counts only.
6264 (clone_inlined_nodes): Do not scale frequency.
6265 (inline_call): Do not pass frequency.
6266 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6267 (compute_inlined_call_time): Use IPA profile.
6268 (want_inline_small_function_p): Use IPA profile.
6269 (want_inline_self_recursive_call_p): Use IPA profile.
6270 (edge_badness): Use IPA profile.
6271 (lookup_recursive_calls): Use IPA profile.
6272 (recursive_inlining): Do not pass frequency.
6273 (resolve_noninline_speculation): Do not update frequency.
6274 (inline_small_functions): Collect max of IPA profile.
6275 (dump_overall_stats): Dump IPA porfile.
6276 (dump_inline_stats): Dump IPA porfile.
6277 (ipa_inline): Collect IPA stats.
6278 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6279 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6280 (ipa_propagate_frequency): Use frequency function.
6281 (ipa_profile): Cleanup.
6282 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6283 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6284 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6285 (input_node): Do not stream frequency.
6286 (input_edge): Do not stream frequency.
6287 (merge_profile_summaries): Scale only IPA profiles.
6288 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6289 * predict.c (drop_profile): Do not recompute frequency.
6290 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6291 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6292 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6293 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6294 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6295 * tree-inline.c (copy_bb): Do not scale frequency.
6296 (expand_call_inline): Do not scale frequency.
6297 (tree_function_versioning): Do not scale frequency.
6298 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6300 2017-11-10 Julia Koval <julia.koval@intel.com>
6302 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6303 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6304 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6305 (_mm256_maskz_gf2p8affine_epi64_epi8)
6306 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6307 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6308 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6309 (__builtin_ia32_vgf2p8affineqb_v32qi)
6310 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6311 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6313 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6317 (arm_option_override): Refactor.
6318 (arm_option_reconfigure_globals): New.
6319 (arm_options_perform_arch_sanity_checks): New.
6320 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6322 (arm_options_perform_arch_sanity_checks): Likewise
6324 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6326 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6327 (power9-qpmul): New.
6328 * rs6000/rs6000.md ("type" attr): Add qmul.
6329 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6330 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6331 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6333 2017-11-10 Martin Sebor <msebor@redhat.com>
6336 * builtins.c (compute_objsize): Handle arrays that
6337 compute_builtin_object_size likes to fail for. Make extern.
6338 * builtins.h (compute_objsize): Declare.
6339 (check_strncpy_sizes): New function.
6340 (expand_builtin_strncpy): Call check_strncpy_sizes.
6341 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6342 -Wstringop-truncation.
6343 (gimple_fold_builtin_strncat): Same.
6344 * gimple.c (gimple_build_call_from_tree): Set call location.
6345 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6346 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6347 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6348 (handle_builtin_strlen): Use strlen_to_stridx.
6349 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6351 Use strlen_to_stridx.
6352 (pass_strlen::execute): Release strlen_to_stridx.
6353 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6354 (-Wstringop-truncation): Document new option.
6356 2017-11-10 Martin Liska <mliska@suse.cz>
6358 PR gcov-profile/82702
6359 * gcov.c (main): Handle intermediate files in a different
6361 (get_gcov_intermediate_filename): New function.
6362 (output_gcov_file): Remove support of intermediate files.
6363 (generate_results): Allocate intermediate file.
6364 (release_structures): Clean-up properly fn_end.
6365 (output_intermediate_file): Start iterating with line 1.
6367 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6369 PR tree-optimization/82929
6370 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6371 ops_swapped_p non-static data member.
6372 (store_immediate_info::store_immediate_info): Clear it.
6373 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6374 ops set ops_swapped_p.
6375 (count_multiple_uses): Handle ops_swapped_p.
6377 2017-11-10 Martin Liska <mliska@suse.cz>
6379 * coverage.c (coverage_init): Stream information about
6380 support of has_unexecuted_blocks.
6381 * doc/gcov.texi: Document that.
6382 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6383 * gcov.c (read_graph_file): Likewise.
6384 (output_line_beginning): Fix a small issue with
6387 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6389 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6390 reference of trivial component.
6392 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6395 * gimple-ssa-store-merging.c
6396 (pass_store_merging::terminate_all_aliasing_chains): For
6397 gimple_store_p stmts also call refs_output_dependent_p.
6399 PR rtl-optimization/82913
6400 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6403 2017-11-09 Jeff Law <law@redhat.com>
6405 * vr-values.h: New file with vr_values class.
6406 * tree-vrp.c: Include vr-values.h
6407 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6408 data objects into the vr_values class.
6409 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6410 (get_value_range): Make it a member function within vr_values class.
6411 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6412 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6413 (op_with_constant_singleton_value_range): Likewise.
6414 (extract_range_for_var_from_comparison_expr): Likewise.
6415 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6416 (extract_range_from_binary_expr): Likewise.
6417 (extract_range_from_unary_expr): Likewise.
6418 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6419 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6420 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6421 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6422 (compare_name_with_value, compare_names): Likewise.
6423 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6424 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6425 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6426 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6427 (extract_range_from_phi_node): Likewise.
6428 (simplify_truth_ops_using_ranges): Likewise.
6429 (simplify_div_or_mod_using_ranges): Likewise.
6430 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6431 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6432 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6433 (simplify_float_conversion_using_ranges): Likewise.
6434 (simplify_internal_call_using_ranges): Likewise.
6435 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6436 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6437 poitner via x_vr_values for calls into gimple folder.
6438 (vrp_initialize_lattice): Make this the vr_values ctor.
6439 (vrp_free_lattice): Make this the vr_values dtor.
6440 (set_vr_value): New function.
6441 (class vrp_prop): Add vr_values data member. Add various member
6442 functions as well as member functions that delegate to vr_values.
6443 (check_array_ref): Make a member function within vrp_prop class.
6444 (search_for_addr_array, vrp_initialize): Likewise.
6445 (vrp_finalize): Likewise. Revamp to avoid direct access to
6446 vr_value, values_propagated, etc.
6447 (check_array_bounds): Extract vrp_prop class instance pointer from
6448 walk info structure. Use it to call member functions.
6449 (check_all_array_refs): Make a member function within vrp_prop class.
6450 Smuggle class instance pointer via walk info structure.
6451 (x_vr_values): New local static.
6452 (vrp_valueize): Use x_vr_values to get class instance.
6453 (vr_valueize_1): Likewise.
6454 (class vrp_folder): Add vr_values data member. Add various member
6455 functions as well as member functions that delegate to vr_values.
6456 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6457 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6458 class instance from vr_values. Use it to call member functions.
6459 (vrp_dom_walker): Add vr_values data member.
6460 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6461 instance via x_vr_values.
6462 (identify_jump_threads): Accept vr_values as argument. Store
6463 it into the walker structure.
6464 (evrp_dom_walker): Add vr_values class data member. Add various
6466 (evrp_dom_walker::try_find_new_range): Use vr_values data
6467 member to access the memory allocator.
6468 (evrp_dom_walker::before_dom_children): Store vr_values class
6469 instance into the vrp_folder class.
6470 (evrp_dom_walker::push_value_range): Rework to avoid direct
6471 access to num_vr_values and vr_value.
6472 (evrp_dom_walker::pop_value_range): Likewise.
6473 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6474 Use vr_values to get to dump_all_value_ranges member function.
6475 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6476 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6477 Pass vr_values class instance down to identify_jump_threads.
6478 Remove call to vrp_free_lattice.
6479 (debug_all_value_ranges): Remove.
6481 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6482 (vrp_folder): Likewise.
6484 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6485 Get it from the existing bitmap instead.
6486 (vrp_intersect_ranges_1): Likewise.
6488 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6490 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6492 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6493 set bit_not_p to it.
6494 (imm_store_chain_info::coalesce_immediate_stores): Break group
6495 if bit_not_p is different.
6496 (count_multiple_uses, split_group,
6497 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6498 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6499 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6500 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6501 to store_immediate_info ctor.
6503 2017-11-09 Jim Wilson <jimw@sifive.com>
6505 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6506 (scan_prog_file): Likewise.
6508 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6510 * bb-reorder.c (max_entry_frequency): Remove.
6511 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6512 connect_traces, push_to_next_round_p): Remove prototypes.
6513 (find_traces_1_round): Use counts only.
6514 (push_to_next_round_p): Likewise.
6515 (find_traces): Likewise.
6516 (rotate_loop): Likewise.
6517 (find_traces_1_round): Likewise.
6518 (connect_traces): Likewise.
6519 (edge_order): Likewise.
6521 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6523 * config/arm/arm.c (output_return_instruction): Add comments to
6524 indicate requirement for cmse_nonsecure_entry return to account
6525 for the size of clearing instruction output here.
6526 (thumb_exit): Likewise.
6527 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6528 return in hardfloat mode.
6530 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6532 * config/rs6000/rs6000.c (machine_function): Add a bool,
6533 "toc_is_wrapped_separately".
6534 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6535 if it wasn't explicitly set or unset, we are optimizing for speed, and
6536 doing separate shrink-wrapping.
6537 (rs6000_get_separate_components): Enable the TOC component if
6538 saving the TOC register in the prologue.
6539 (rs6000_components_for_bb): Handle the TOC component.
6540 (rs6000_emit_prologue_components): Store the TOC register where needed.
6541 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6542 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6544 2017-11-09 Martin Jambor <mjambor@suse.cz>
6546 * ipa-param-manipulation.c: New file.
6547 * ipa-param-manipulation.h: Likewise.
6548 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6549 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6550 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6551 (ipa_parm_adjustment): Likewise.
6552 (ipa_parm_adjustment_vec): Likewise.
6553 (ipa_get_vector_of_formal_parms): Moved declaration to
6554 ipa-param-manipulation.h.
6555 (ipa_get_vector_of_formal_parm_types): Likewise.
6556 (ipa_modify_formal_parameters): Likewise.
6557 (ipa_modify_call_arguments): Likewise.
6558 (ipa_combine_adjustments): Likewise.
6559 (ipa_dump_param_adjustments): Likewise.
6560 (ipa_modify_expr): Likewise.
6561 (ipa_get_adjustment_candidate): Likewise.
6562 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6563 ipa-param-manipulation.c.
6564 (ipa_get_vector_of_formal_parm_types): Likewise.
6565 (ipa_modify_formal_parameters): Likewise.
6566 (ipa_modify_call_arguments): Likewise.
6567 (ipa_modify_expr): Likewise.
6568 (get_ssa_base_param): Likewise.
6569 (ipa_get_adjustment_candidate): Likewise.
6570 (index_in_adjustments_multiple_times_p): Likewise.
6571 (ipa_combine_adjustments): Likewise.
6572 (ipa_dump_param_adjustments): Likewise.
6573 * tree-sra.c: Also include ipa-param-manipulation.h
6574 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6577 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6578 Alan Hayward <alan.hayward@arm.com>
6579 David Sherwood <david.sherwood@arm.com>
6581 * doc/sourcebuild.texi (vect_masked_store): Document.
6583 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6584 Alan Hayward <alan.hayward@arm.com>
6585 David Sherwood <david.sherwood@arm.com>
6587 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6589 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6590 Alan Hayward <alan.hayward@arm.com>
6591 David Sherwood <david.sherwood@arm.com>
6593 * doc/sourcebuild.texi (vect_variable_length): Document.
6595 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6596 Alan Hayward <alan.hayward@arm.com>
6597 David Sherwood <david.sherwood@arm.com>
6599 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6601 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6602 Alan Hayward <alan.hayward@arm.com>
6603 David Sherwood <david.sherwood@arm.com>
6605 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6607 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6608 Alan Hayward <alan.hayward@arm.com>
6609 David Sherwood <david.sherwood@arm.com>
6611 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6612 previously undocumented selectors.
6613 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6615 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6617 * doc/rtl.texi (const_vector): Say that elements can be
6618 const_wide_ints too.
6619 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6620 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6621 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6622 * optabs.c (expand_vector_broadcast): Likewise.
6624 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6625 Alan Hayward <alan.hayward@arm.com>
6626 David Sherwood <david.sherwood@arm.com>
6628 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6629 scaled index even if the unscaled address was invalid.
6630 Don't increase the complexity of using a scale in that case.
6632 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6633 Alan Hayward <alan.hayward@arm.com>
6634 David Sherwood <david.sherwood@arm.com>
6636 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6637 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6638 UNITS_PER_WORD bytes.
6639 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6640 based on REGMODE_NATURAL_SIZE of the inner mode.
6641 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6642 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6643 * expmed.c (lowpart_bit_field_p): Divide the value up into
6644 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6645 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6646 whether something is likely to occupy more than one register.
6648 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6651 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6652 profile_count::adjust_for_ipa_scaling.
6653 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6654 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6656 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6658 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6660 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6661 (split_group): Add total_orig and total_new arguments, estimate the
6662 number of statements related to the store group without store merging
6663 and with store merging.
6664 (imm_store_chain_info::output_merged_store): Adjust split_group
6665 callers, punt if estimated number of statements with store merging
6666 is not smaller than estimated number of statements without it.
6668 (handled_load): Remove has_single_use checks.
6669 (pass_store_merging::process_store): Likewise.
6671 2017-11-09 Richard Biener <rguenther@suse.de>
6673 PR tree-optimization/82902
6674 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6676 2017-11-09 Martin Liska <mliska@suse.cz>
6679 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6682 2017-11-09 Martin Liska <mliska@suse.cz>
6684 PR tree-optimization/82669
6685 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6687 2017-11-09 Martin Liska <mliska@suse.cz>
6689 PR gcov-profile/48463
6690 * coverage.c (coverage_begin_function): Output also end locus
6691 of a function and information whether the function is
6693 * gcov-dump.c (tag_function): Parse and print the information.
6694 * gcov.c (INCLUDE_MAP): Add include.
6695 (INCLUDE_SET): Likewise.
6696 (struct line_info): Move earlier in the source file because
6697 of vector<line_info> in function_info structure.
6698 (line_info::line_info): Likewise.
6699 (line_info::has_block): Likewise.
6700 (struct source_info): Add new member index.
6701 (source_info::get_functions_at_location): New function.
6702 (function_info::group_line_p): New function.
6703 (output_intermediate_line): New function.
6704 (output_intermediate_file): Use the mentioned function.
6705 (struct function_start): New.
6706 (struct function_start_pair_hash): Likewise.
6707 (process_file): Add code that identifies group functions.
6708 Assign lines either to global or function scope.
6709 (generate_results): Skip artificial functions.
6710 (find_source): Assign index for each source file.
6711 (read_graph_file): Read new flag artificial and end_line.
6712 (add_line_counts): Assign it either to global of function scope.
6713 (accumulate_line_counts): Isolate core of the function to
6714 accumulate_line_info and call it for both function and global
6716 (accumulate_line_info): New function.
6717 (output_line_beginning): Fix GNU coding style.
6718 (print_source_line): New function.
6719 (output_line_details): Likewise.
6720 (output_function_details): Likewise.
6721 (output_lines): Iterate both source (global) scope and function
6723 (struct function_line_start_cmp): New class.
6724 * doc/gcov.texi: Reflect changes in documentation.
6726 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6729 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6730 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6731 and similarly for not instead of neg.
6733 2017-11-08 Andi Kleen <ak@linux.intel.com>
6735 * config/i386/i386.opt: Add -mforce-indirect-call.
6736 * config/i386/predicates.md: Check for flag_force_indirect_call.
6737 * doc/invoke.texi: Document -mforce-indirect-call
6739 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6741 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6743 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6744 config/riscv/riscv.c (predict.h): New include.
6745 (riscv_slow_unaligned_access_p): No longer static.
6746 (riscv_block_move_straight): Add require.
6747 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6749 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6752 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6753 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6754 nonimmediate_operand predicate for operand 1 instead of
6757 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6759 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6761 * config/aarch64/constraints.md (Uml): New constraint.
6762 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6765 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6767 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6768 of two vec_duplicates into a vec_concat.
6770 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6772 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6773 Simplify vec_merge of vec_duplicate and vec_concat.
6774 * config/aarch64/constraints.md (Utq): New constraint.
6775 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6778 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6780 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6781 Simplify vec_merge of vec_duplicate and const_vector.
6782 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6784 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6785 mode iterator. Update predicate on operand 1 to
6786 handle non-const_vec constants. Delete constraints.
6787 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6789 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6791 PR tree-optimization/78821
6792 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6794 (store_operand_info::store_operand_info): Initialize it to false.
6795 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6796 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6797 store in the group, and if chain_info is non-NULL, to ignore altogether
6799 (compatible_load_p): Fail if bit_not_p does not match.
6800 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6801 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6802 (pass_store_merging::process_store): Adjust
6803 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6804 call terminate_all_aliasing_chains newly when adding a store into
6805 a chain with non-NULL chain_info.
6807 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6809 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6811 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6813 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6815 (aarch64_layout_frame): Initialise emit_frame_chain.
6816 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6817 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6819 2017-11-08 Martin Liska <mliska@suse.cz>
6821 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6822 of gimple_call_internal_p.
6824 2017-11-07 Tom de Vries <tom@codesourcery.com>
6826 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6828 2017-11-07 Tom de Vries <tom@codesourcery.com>
6830 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6833 2017-11-08 Martin Liska <mliska@suse.cz>
6836 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6838 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6840 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6842 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6845 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6846 (*cmp<mode>_ccz_1): New insn with $k alternative.
6849 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6850 RTX_COMM_COMPARE as commutative as well.
6851 (ix86_binary_operator_ok): Formatting fix.
6852 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6853 *<code><mode>3<mask_name><round_saeonly_name>,
6854 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6855 *<s>mul<mode>3_highpart<mask_name>,
6856 *vec_widen_umult_even_v16si<mask_name>,
6857 *vec_widen_umult_even_v8si<mask_name>,
6858 *vec_widen_umult_even_v4si<mask_name>,
6859 *vec_widen_smult_even_v16si<mask_name>,
6860 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6861 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6862 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6863 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6864 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6865 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6866 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6867 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6868 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6869 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6870 ix86_binary_operator_ok. Formatting fixes.
6871 (*<plusminus_insn><mode>3<mask_name><round_name>,
6872 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6875 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6877 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6878 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6879 eq and ne if TARGET_ISEL.
6880 (cmp): New code_iterator.
6881 (UNS, UNSU_, UNSIK): New code_attrs.
6882 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6883 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6885 ("eq<mode>3"): ... this.
6886 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6888 ("ne<mode>3"): ... this.
6890 2017-11-07 Julia Koval <julia.koval@intel.com>
6893 * common/config/i386/i386-common.c
6894 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6895 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6896 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6897 * config/i386/i386.opt: Ditto.
6898 * config/i386/i386.c (ix86_target_string): Ditto.
6899 (ix86_option_override_internal): Ditto.
6900 (ix86_init_mpx_builtins): Move MPX to args2.
6901 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6902 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6903 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6904 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6905 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6906 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6907 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6909 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6912 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6913 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6914 (zero-extendsidi peephole2): Remove peephole.
6916 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6919 * stor-layout.c: Include attribs.h.
6920 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6921 explicit lookup of "aligned" attribute.
6923 2017-11-07 Andrew Waterman <andrew@sifive.com>
6925 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6926 (riscv_expand_block_move): Likewise.
6927 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6929 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6930 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6931 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6932 (riscv_adjust_block_mem): Likewise.
6933 (riscv_block_move_loop): Likewise.
6934 (riscv_expand_block_move): Likewise.
6935 * config/riscv/riscv.md (movmemsi): New pattern.
6937 2017-11-07 Michael Clark <michaeljclark@mac.com>
6939 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6940 (MUSL_DYNAMIC_LINKER): Likewise.
6942 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6944 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6945 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6948 2017-11-07 Richard Biener <rguenther@suse.de>
6950 * match.pd: Fix build.
6952 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6953 Jackson Woodruff <jackson.woodruff@arm.com>
6955 PR tree-optimization/71026
6956 * match.pd: Canonicalize negate in division.
6958 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6961 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6963 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6965 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6966 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6968 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6970 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6971 non-scalar integral types.
6972 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6973 (-(A-B), -(~A)): New transformations.
6975 2017-11-07 Tom de Vries <tom@codesourcery.com>
6977 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6978 semicolon after "do {} while (0)".
6979 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6980 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6981 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6982 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6983 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6985 2017-11-07 Tom de Vries <tom@codesourcery.com>
6987 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6988 after "do {} while (0)".
6989 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6990 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6991 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6992 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6993 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6995 2017-11-07 Tom de Vries <tom@codesourcery.com>
6998 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7000 (arm_rtx_costs_internal): Add missing semicolon after
7001 HANDLE_NARROW_SHIFT_ARITH call.
7003 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7006 disable isel if it was not set explicitly.
7008 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7010 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7011 (add_type_attribute) likewise.
7013 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7015 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7016 of struct ix86_frame.
7017 (ix86_initial_elimination_offset): Likewise.
7018 (ix86_expand_split_stack_prologue): Likewise.
7020 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7022 * tree-vrp.h (enum value_range_type): Update stale comment.
7024 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7025 Alan Hayward <alan.hayward@arm.com>
7026 David Sherwood <david.sherwood@arm.com>
7028 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7029 (aarch64_expand_vec_perm_const): Take the number of units too.
7030 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7031 (aarch64_expand_vec_perm_const): Likewise.
7032 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7033 (vec_perm<mode>): Update accordingly.
7035 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7036 Alan Hayward <alan.hayward@arm.com>
7037 David Sherwood <david.sherwood@arm.com>
7039 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7040 Take the number of units too.
7041 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7042 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7043 but check for a vector mode before rather than after the call.
7044 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7045 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7046 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7047 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7048 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7049 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7050 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7051 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7052 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7053 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7054 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7055 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7056 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7057 (aarch64_sqdmull2_n<mode>): Update accordingly.
7059 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7060 Alan Hayward <alan.hayward@arm.com>
7061 David Sherwood <david.sherwood@arm.com>
7063 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7064 the number of units too.
7065 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7066 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7067 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7068 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7069 (vec_store_lanesxi<mode>): Update accordingly.
7071 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7072 Alan Hayward <alan.hayward@arm.com>
7073 David Sherwood <david.sherwood@arm.com>
7075 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7076 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7077 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7078 of units rather than the mode.
7079 * config/aarch64/iterators.md (nunits): New mode attribute.
7080 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7081 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7082 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7083 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7084 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7085 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7086 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7087 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7088 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7089 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7090 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7091 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7092 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7093 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7094 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7095 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7096 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7097 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7098 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7099 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7100 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7101 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7102 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7103 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7104 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7105 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7106 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7107 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7108 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7109 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7110 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7111 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7112 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7113 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7114 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7115 (aarch64_simd_vec_setv2di): Likewise.
7117 2017-11-06 Carl Love <cel@us.ibm.com>
7119 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7121 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7122 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7124 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7126 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7127 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7128 Add power 8 macro expansions.
7129 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7130 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7131 power 8 instructions. (VSX_XXBR): Add iterator.
7133 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7135 * config/arm/arm.md (predicable_short_it): Change default to "no",
7136 improve documentation, remove uses that are identical to the default.
7137 (enabled_for_depr_it): Rename to enabled_for_short_it.
7138 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7139 Remove default uses.
7140 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7141 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7142 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7143 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7145 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7148 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7149 float128 helper macros, which are no longer used after deleting
7150 the old 'q' built-in functions, and moving the round to odd
7151 built-in functions to being special built-in functions.
7152 (BU_FLOAT128_2): Likewise.
7153 (BU_FLOAT128_1_HW): Likewise.
7154 (BU_FLOAT128_2_HW): Likewise.
7155 (BU_FLOAT128_3_HW): Likewise.
7156 (FABSQ): Delete old 'q' built-in functions.
7157 (COPYSIGNQ): Likewise.
7158 (SQRTF128_ODD): Move round to odd built-in functions to be
7159 special built-in functions, so that we can handle
7160 -mabi=ieeelongdouble.
7161 (TRUNCF128_ODD): Likewise.
7162 (ADDF128_ODD): Likewise.
7163 (SUBF128_ODD): Likewise.
7164 (MULF128_ODD): Likewise.
7165 (DIVF128_ODD): Likewise.
7166 (FMAF128_ODD): Likewise.
7167 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7168 built-in names to 'f128'.
7169 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7170 old 'q' built-in functions, as the machine independent code for
7171 'f128' built-in functions handles this.
7172 (rs6000_expand_builtin): Add expansion for float128 round to odd
7173 functions, keying off on -mabi=ieeelongdouble of whether to use
7174 the KFmode or TFmode variant.
7175 (rs6000_init_builtins): Initialize the _Float128 round to odd
7177 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7178 _Float128 'q' built-in functions are now mapped into the new
7179 'f128' built-in functions.
7181 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7183 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7184 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7186 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7189 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7190 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7191 * toplev.c: Include "ipa-fnsummary.h".
7192 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7194 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7196 PR tree-optimization/82838
7197 * gimple-ssa-store-merging.c
7198 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7199 on a separate gimple_seq which is then appended to seq.
7201 2017-11-06 Jeff Law <law@redhat.com>
7204 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7205 (get_probe_interval): New functions.
7206 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7207 (ix86_adjust_stack_and_probe): Likewise.
7208 (output_adjust_stack_and_probe): Likewise.
7209 (ix86_emit_probe_stack_range): Likewise.
7210 (ix86_expand_prologue): Likewise.
7212 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7214 PR tree-optimization/82816
7215 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7216 if the modes of the two types are the same.
7217 (convert_plusminus_to_widen): Likewise.
7219 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7221 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7223 (usadv16qi): New define_expand.
7224 (usadv8hi): New define_expand.
7226 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7229 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7231 (inline_transform): Likewise.
7232 * predict.c (counts_to_freqs): Remove useless conditional.
7233 * profile-count.h (profile_count::apply_scale): Move sanity check.
7234 * tree-inline.c (copy_bb): Always scale.
7235 (copy_cfg_body): Likewise.
7237 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7240 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7242 (*cmp_ite0): Add enabled_for_depr_it attribute.
7243 (*cmp_ite1): Likewise.
7245 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7247 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7250 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7252 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7253 return the value by pointer.
7254 (extract_range_from_multiplicative_op_1): Update accordingly.
7255 Return as soon as an operation fails.
7257 2017-11-05 Tom de Vries <tom@codesourcery.com>
7260 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7261 (DEF_SANITIZER_BUILTIN): ... here.
7262 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7263 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7265 2017-11-05 Tom de Vries <tom@codesourcery.com>
7268 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7270 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7271 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7272 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7274 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7275 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7276 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7278 2017-11-05 Tom de Vries <tom@codesourcery.com>
7281 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7284 2017-11-04 Michael Clark <michaeljclark@mac.com>
7286 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7287 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7289 (*addsi3_extended): Likewise.
7290 (*addsi3_extended2): Likewise.
7291 (<optab>si3): Likewise.
7292 (<optab>di3): Likewise.
7293 (<optab><mode>3): Likewise.
7294 (<*optabe>si3_internal): Likewise.
7295 (zero_extendqi<SUPERQI:mode>2): Likewise.
7296 (*add<mode>hi3): Likewise.
7297 (*xor<mode>hi3): Likewise.
7298 (<optab>di3): Likewise.
7299 (*<optab>si3_extend): Likewise.
7300 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7301 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7302 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7304 2017-11-04 Andrew Waterman <andrew@sifive.com>
7306 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7307 TARGET_STRICT_ALIGN based upon -mtune argument.
7309 2017-11-04 Andrew Waterman <andrew@sifive.com>
7311 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7313 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7315 * config/i386/i386.c (choose_basereg): Use optional scratch
7316 register and add assertion.
7317 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7318 needed, and don't allocate stack.
7319 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7320 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7321 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7323 2017-11-03 Jeff Law <law@redhat.com>
7325 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7326 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7327 to probe at the start of a noreturn function.
7329 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7331 PR tree-optimization/78821
7332 * gimple-ssa-store-merging.c: Update the file comment.
7333 (MAX_STORE_ALIAS_CHECKS): Define.
7334 (struct store_operand_info): New type.
7335 (store_operand_info::store_operand_info): New constructor.
7336 (struct store_immediate_info): Add rhs_code and ops data members.
7337 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7338 and op1r arguments to the ctor, initialize corresponding data members.
7339 (struct merged_store_group): Add load_align_base and load_align
7341 (merged_store_group::merged_store_group): Initialize them.
7342 (merged_store_group::do_merge): Update them.
7343 (merged_store_group::apply_stores): Pick the constant for
7344 encode_tree_to_bitpos from one of the two operands, or skip
7345 encode_tree_to_bitpos if neither operand is a constant.
7346 (class pass_store_merging): Add process_store method decl. Remove
7347 bool argument from terminate_all_aliasing_chains method decl.
7348 (pass_store_merging::terminate_all_aliasing_chains): Remove
7349 var_offset_p argument and corresponding handling.
7350 (stmts_may_clobber_ref_p): New function.
7351 (compatible_load_p): New function.
7352 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7353 if there is overlap and rhs_code is not INTEGER_CST. For
7354 non-overlapping stores terminate group if rhs is not mergeable.
7355 (get_alias_type_for_stmts): Change first argument from
7356 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7357 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7358 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7360 (get_location_for_stmts): Change first argument from
7361 auto_vec<gimple *> & to vec<gimple *> &.
7362 (struct split_store): Remove orig_stmts data member, add orig_stores.
7363 (split_store::split_store): Create orig_stores rather than orig_stmts.
7364 (find_constituent_stmts): Renamed to ...
7365 (find_constituent_stores): ... this. Change second argument from
7366 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7367 to info structures rather than the statements.
7368 (split_group): Rename ALLOW_UNALIGNED argument to
7369 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7370 it. Adjust find_constituent_stores caller.
7371 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7372 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7373 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7374 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7375 (mem_valid_for_store_merging): New function.
7376 (handled_load): New function.
7377 (pass_store_merging::process_store): New method.
7378 (pass_store_merging::execute): Use process_store method. Adjust
7379 terminate_all_aliasing_chains caller.
7381 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7383 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7384 Return true for more constants, symbols and label references.
7385 (aarch64_valid_floating_const): Remove unused function.
7387 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7390 * config/aarch64/aarch64.c (aarch64_layout_frame):
7391 Undo forcing of LR at bottom of frame.
7393 2017-11-03 Jeff Law <law@redhat.com>
7396 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7397 for int_registers_saved.
7399 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7400 extracted from tree-ssa-dom.c.
7401 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7402 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7403 (record_equivalences_from_incoming_edge): Add additional argument
7404 to single_pred_edge_ignoring_loop_edges call.
7405 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7406 (uncprop_dom_walker::before_dom_children): Add additional argument
7407 to single_pred_edge_ignoring_loop_edges call.
7408 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7409 single_pred_edge_ignoring_loop_edges rather than open coding.
7410 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7412 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7414 * match.pd (-(-A)): Rewrite.
7416 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7418 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7419 (rs6000_emit_int_cmove): New declaration.
7420 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7421 (rs6000_emit_sISEL): Delete.
7422 (rs6000_emit_int_cmove): Make non-static.
7423 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7424 instead of rs6000_emit_sISEL.
7426 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7428 * asan.c (create_cond_insert_point): Maintain profile.
7429 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7431 * basic-block.h (struct basic_block_def): Remove frequency.
7432 (EDGE_FREQUENCY): Use to_frequency
7433 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7435 (find_traces): Update to use to_frequency.
7436 (find_traces_1_round): Likewise; use only IPA counts.
7437 (bb_to_key): Likewise.
7438 (connect_traces): Use IPA counts only.
7439 (copy_bb_p): Update to use to_frequency.
7440 (fix_up_crossing_landing_pad): Likewise.
7441 (sanitize_hot_paths): Likewise.
7442 * bt-load.c (basic_block_freq): Likewise.
7443 * cfg.c (init_flow): Set count_max to uninitialized.
7444 (check_bb_profile): Remove frequencies; check counts.
7445 (dump_bb_info): Do not dump frequencies.
7446 (update_bb_profile_for_threading): Update counts only.
7447 (scale_bbs_frequencies_int): Likewise.
7448 (MAX_SAFE_MULTIPLIER): Remove.
7449 (scale_bbs_frequencies_gcov_type): Update counts only.
7450 (scale_bbs_frequencies_profile_count): Update counts only.
7451 (scale_bbs_frequencies): Update counts only.
7452 * cfg.h (struct control_flow_graph): Add count-max.
7453 (update_bb_profile_for_threading): Update prototype.
7454 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7455 (find_many_sub_basic_blocks): Likewise.
7456 * cfgcleanup.c (try_forward_edges): Likewise.
7457 (try_crossjump_to_edge): Likewise.
7458 * cfgexpand.c (expand_gimple_cond): Likewise.
7459 (expand_gimple_tailcall): Likewise.
7460 (construct_init_block): Likewise.
7461 (construct_exit_block): Likewise.
7462 * cfghooks.c (verify_flow_info): Check consistency of counts.
7463 (dump_bb_for_graph): Do not dump frequencies.
7464 (split_block_1): Do not update frequencies.
7465 (split_edge): Do not update frequencies.
7466 (make_forwarder_block): Do not update frequencies.
7467 (duplicate_block): Do not update frequencies.
7468 (account_profile_record): Do not update frequencies.
7469 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7470 for global heuristics.
7471 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7472 (expected_loop_iterations_unbounded): Use counts only.
7473 * cfgloopmanip.c (scale_loop_profile): Simplify.
7474 (create_empty_loop_on_edge): Simplify
7476 (duplicate_loop_to_header_edge): Simplify
7477 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7478 (update_br_prob_note): Take care of removing note when profile
7480 (relink_block_chain): Do not dump frequency.
7481 (rtl_account_profile_record): Use to_frequency.
7482 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7483 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7484 (cgraph_update_edges_for_call_stmt_node): Likewise.
7485 (cgraph_edge::verify_count_and_frequency): Update.
7486 (cgraph_node::verify_node): Temporarily disable frequency verification.
7487 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7488 to_cgraph_frequency.
7489 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7490 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7492 (cgraph_node::expand_thunk): Update profile.
7493 * except.c (dw2_build_landing_pads): Do not update frequency.
7494 * final.c (compute_alignments): Use to_frequency.
7495 (dump_basic_block_info): Do not dump frequency.
7496 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7497 (dump_gimple_bb_header): Do not dump frequency.
7498 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7500 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7501 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7502 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7503 (init_before_recovery): Do not update frequency.
7504 (sched_create_recovery_edges): Do not update frequency.
7505 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7506 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7507 (ipa_cp_c_finalize): Set max_count to uninitialized.
7508 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7509 (param_change_prob): Use counts.
7510 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7512 * ipa-split.c (consider_split): Use to_frequency.
7513 (split_function): Use to_frequency.
7514 * ira-build.c (loop_compare_func): Likewise.
7515 (mark_loops_for_removal): Likewise.
7516 (mark_all_loops_for_removal): Likewise.
7517 * loop-doloop.c (doloop_modify): Do not update frequency.
7518 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7520 * lto-streamer-in.c (input_function): Update count_max.
7521 * omp-expand.c (expand_omp_taskreg): Update count_max.
7522 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7523 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7524 (maybe_hot_count_p): Use ipa counts only.
7525 (maybe_hot_bb_p): Simplify.
7526 (maybe_hot_edge_p): Simplify.
7527 (probably_never_executed): Do not take frequency argument.
7528 (probably_never_executed_bb_p): Do not pass frequency.
7529 (probably_never_executed_edge_p): Likewise.
7530 (combine_predictions_for_bb): Check that profile is nonzero.
7531 (propagate_freq): Do not set frequency.
7532 (drop_profile): Simplify.
7533 (counts_to_freqs): Simplify.
7534 (expensive_function_p): Use to_frequency.
7535 (propagate_unlikely_bbs_forward): Simplify.
7536 (determine_unlikely_bbs): Simplify.
7537 (estimate_bb_frequencies): Add hack to silence graphite issues.
7538 (compute_function_frequency): Use ipa counts.
7539 (pass_profile::execute): Update.
7540 (rebuild_frequencies): Use counts only.
7541 (force_edge_cold): Use counts only.
7542 * profile-count.c (profile_count::dump): Dump new count types.
7543 (profile_count::differs_from_p): Check compatiblity.
7544 (profile_count::to_frequency): New function.
7545 (profile_count::to_cgraph_frequency): New function.
7546 * profile-count.h (struct function): Declare.
7547 (enum profile_quality): Add profile_guessed_local and
7548 profile_guessed_global0.
7549 (class profile_proability): Decrease number of bits to 29;
7550 update from_reg_br_prob_note and to_reg_br_prob_note.
7551 (class profile_count: Update comment; decrease number of bits
7552 to 61. Check compatibility.
7553 (profile_count::compatible_p): New private member function.
7554 (profile_count::ipa_p): New member function.
7555 (profile_count::operator<): Handle global zero correctly.
7556 (profile_count::operator>): Handle global zero correctly.
7557 (profile_count::operator<=): Handle global zero correctly.
7558 (profile_count::operator>=): Handle global zero correctly.
7559 (profile_count::nonzero_p): New member function.
7560 (profile_count::force_nonzero): New member function.
7561 (profile_count::max): New member function.
7562 (profile_count::apply_scale): Handle IPA scalling.
7563 (profile_count::guessed_local): New member function.
7564 (profile_count::global0): New member function.
7565 (profile_count::ipa): New member function.
7566 (profile_count::to_frequency): Declare.
7567 (profile_count::to_cgraph_frequency): Declare.
7568 * profile.c (OVERLAP_BASE): Delete.
7569 (compute_frequency_overlap): Delete.
7570 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7571 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7572 * sched-ebb.c (rank): Use counts only.
7573 * shrink-wrap.c (handle_simple_exit): Use counts only.
7574 (try_shrink_wrapping): Use counts only.
7575 (place_prologue_for_one_component): Use counts only.
7576 * tracer.c (find_best_predecessor): Use to_frequency.
7577 (find_trace): Use to_frequency.
7578 (tail_duplicate): Use to_frequency.
7579 * trans-mem.c (expand_transaction): Do not update frequency.
7580 * tree-call-cdce.c: Do not update frequency.
7581 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7582 (gimple_merge_blocks): Likewise.
7583 (gimple_split_edge): Likewise.
7584 (gimple_duplicate_sese_region): Likewise.
7585 (gimple_duplicate_sese_tail): Likewise.
7586 (move_sese_region_to_fn): Likewise.
7587 (gimple_account_profile_record): Likewise.
7588 (insert_cond_bb): Likewise.
7589 * tree-complex.c (expand_complex_div_wide): Likewise.
7590 * tree-eh.c (lower_resx): Update profile.
7591 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7593 (initialize_cfun): Do not initialize frequencies
7594 (freqs_to_counts): Delete.
7595 (copy_cfg_body): Ignore count parameter.
7596 (copy_body): Update.
7597 (expand_call_inline): Update count_max.
7598 (optimize_inline_calls): Update count_max.
7599 (tree_function_versioning): Update count_max.
7600 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7601 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7603 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7604 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7605 (try_peel_loop): Likewise.
7606 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7608 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7609 (tree_transform_and_unroll_loop): Do not use frequencies
7610 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7611 Use reliable prediction only.
7612 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7613 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7614 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7615 probability scaling.
7616 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7617 not update frequency
7618 (any_remaining_duplicated_blocks): Likewise.
7619 (update_profile): Likewise.
7620 (estimated_freqs_path): Delete.
7621 (freqs_to_counts_path): Delete.
7622 (clear_counts_path): Delete.
7623 (ssa_fix_duplicate_block_edges): Likewise.
7624 (duplicate_thread_path): Likewise.
7625 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7626 * tree-tailcall.c (decrease_profile): Do not update frequency.
7627 (eliminate_tail_call): Likewise.
7628 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7629 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7630 (optimize_mask_stores): Likewise.
7631 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7632 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7633 (ubsan_expand_ptr_ifn): Update profile.
7634 * value-prof.c (gimple_ic): Simplify.
7635 * value-prof.h (gimple_ic): Update prototype.
7636 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7637 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7639 (want_inline_self_recursive_call_p): Likewise.
7640 (resolve_noninline_speculation): Only cummulate defined counts.
7641 (inline_small_functions): Use nonzero_p.
7642 (ipa_inline): Do not access freed node.
7644 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7646 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7647 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7649 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7651 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7652 non-legitimate address.
7654 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7656 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7657 (*lt0_<mode>di, *lt0_<mode>si): New.
7659 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7661 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7662 TARGET_PAIRED_FLOAT.
7664 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7665 Jim Wilson <jim.wilson@linaro.org>
7667 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7668 * config/aarch64/aarch64-tune.md: Regenerated.
7669 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7670 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7672 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7674 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7675 (arc_expand_prologue): Restore blink for millicode.
7676 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7678 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7681 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7682 gen_vec_duplicate after forcing the scalar into a register.
7684 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7686 * combine (try_combine): Print the insns input to try_combine to the
7689 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7692 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7693 Remove second argument from aarch64_process_target_attr call.
7694 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7696 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7698 (aarch64_handle_attr_arch): Remove second argument.
7699 (aarch64_handle_attr_cpu): Ditto.
7700 (aarch64_handle_attr_tune): Ditto.
7701 (aarch64_handle_attr_isa_flags): Ditto.
7702 (aarch64_process_one_target_attr): Ditto.
7703 (aarch64_process_target_attr): Ditto.
7704 (aarch64_option_valid_attribute_p): Remove second argument.
7705 on aarch64_process_target_attr call.
7707 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7709 * diagnostic.c: Include "selftest-diagnostic.h".
7710 (selftest::assert_location_text): New function.
7711 (selftest::test_diagnostic_get_location_text): New function.
7712 (selftest::diagnostic_c_tests): Call it.
7714 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7716 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7717 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7718 (class selftest::test_diagnostic_context): Move to...
7719 * selftest-diagnostic.c: New file.
7720 * selftest-diagnostic.h: New file.
7722 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7724 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7725 offset range for FT32B.
7726 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7727 * config/ft32/ft32.md: Add TARGET_NOPM.
7728 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7729 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7731 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7733 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7735 2017-11-02 Jeff Law <law@redhat.com>
7737 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7738 virtual keyword on FINAL OVERRIDE members.
7740 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7741 virtuals together. Add virtual destructor.
7742 (substitute_and_fold_engine): Similarly.
7744 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7746 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7748 2017-11-02 Richard Biener <rguenther@suse.de>
7750 PR tree-optimization/82795
7751 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7753 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7755 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7757 (gcc_GAS_CHECK_FEATURE): Remove.
7758 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7759 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7760 * configure: Regenerate.
7762 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7764 * config/arc/arc.c (hwloop_optimize): Account for empty
7767 2017-11-02 Richard Biener <rguenther@suse.de>
7770 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7771 Truncate ARRAY_REF index and element size.
7773 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7775 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7777 2017-11-01 Jeff Law <law@redhat.com>
7779 * tree-ssa-ccp.c (ccp_folder): New class derived from
7780 substitute_and_fold_engine.
7781 (ccp_folder::get_value): New member function.
7782 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7783 (ccp_fold_stmt): Remove prototype.
7784 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7785 * tree-ssa-copy.c (copy_folder): New class derived from
7786 substitute_and_fold_engine.
7787 (copy_folder::get_value): Renamed from get_value.
7788 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7789 * tree-vrp.c (vrp_folder): New class derived from
7790 substitute_and_fold_engine.
7791 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7792 (vrp_folder::get_value): New member function.
7793 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7794 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7795 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7796 provide a class interface to folder/substitute routines.
7797 (ssa_prop_fold_stmt_fn): Remove typedef.
7798 (ssa_prop_get_value_fn): Likewise.
7799 (subsitute_and_fold): Remove prototype.
7800 (replace_uses_in): Likewise.
7801 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7802 Renamed from replace_uses_in. Call the virtual member function
7803 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7804 (substitute_and_fold_dom_walker): Remove initialization of
7805 data member entries for calbacks. Add substitute_and_fold_engine
7806 member and initialize it.
7807 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7808 member functions for get_value, replace_phi_args_in c
7809 replace_uses_in, and fold_stmt calls.
7810 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7811 substitute_and_fold. Remove assert. Update ctor call.
7813 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7814 (ssa_prop_visit_phi_fn): Likewise.
7815 (class ssa_propagation_engine): New class to provide an interface
7817 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7819 (ssa_prop_visit_phi): Likewise.
7820 (ssa_propagation_engine::simulate_stmt): Moved into class.
7821 Call visit_phi/visit_stmt from the class rather than via
7822 file scoped static variables.
7823 (ssa_propagation_engine::simulate_block): Moved into class.
7824 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7825 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7826 set file scoped statics for the visit_stmt/visit_phi callbacks.
7827 * tree-complex.c (complex_propagate): New class derived from
7828 ssa_propagation_engine.
7829 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7830 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7831 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7833 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7834 ssa_propagation_engine.
7835 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7836 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7837 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7838 * tree-ssa-copy.c (copy_prop): New class derived from
7839 ssa_propagation_engine.
7840 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7841 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7842 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7843 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7844 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7845 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7846 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7848 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7850 PR rtl-optimization/82778
7851 PR rtl-optimization/82597
7852 * compare-elim.c (struct comparison): Add in_a_setter field.
7853 (find_comparison_dom_walker::before_dom_children): Remove killed
7854 bitmap and df_simulate_find_defs call, instead walk the defs.
7855 Compute last_setter and initialize in_a_setter. Merge definitions
7856 with first initialization for a few variables.
7857 (try_validate_parallel): Use insn_invalid_p instead of
7858 recog_memoized. Return insn rather than just the pattern.
7859 (try_merge_compare): Fix up comment. Don't uselessly test if
7860 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7862 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7865 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7866 Alan Hayward <alan.hayward@arm.com>
7867 David Sherwood <david.sherwood@arm.com>
7869 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7870 aarch64_hard_regno_nregs to get the number of registers
7873 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7874 Alan Hayward <alan.hayward@arm.com>
7875 David Sherwood <david.sherwood@arm.com>
7877 * config/aarch64/constraints.md (Upl): Rename to...
7879 * config/aarch64/aarch64.md
7880 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7883 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7884 Alan Hayward <alan.hayward@arm.com>
7885 David Sherwood <david.sherwood@arm.com>
7887 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7888 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7891 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7892 Alan Hayward <alan.hayward@arm.com>
7893 David Sherwood <david.sherwood@arm.com>
7895 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7896 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7897 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7899 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7900 of a single element.
7901 * config/aarch64/iterators.md: Add a comment above the permute
7902 unspecs to say that they are generated directly by
7903 aarch64_expand_vec_perm_const.
7904 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7906 2017-11-01 Nathan Sidwell <nathan@acm.org>
7908 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7910 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7912 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7913 and medany code models, and describe what they do.
7915 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7917 Revert accidental duplicate:
7919 * combine.c (can_change_dest_mode): Reject changes in
7920 REGMODE_NATURAL_SIZE.
7922 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7924 PR rtl-optimization/64682
7925 PR rtl-optimization/69567
7926 PR rtl-optimization/69737
7927 PR rtl-optimization/82683
7928 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7929 register mentioned in the note, drop the note, unless it came from I3,
7930 in which case it should go to I3 again.
7932 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7934 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7935 and return false if not.
7936 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7938 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7940 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7941 range is known to be empty.
7943 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7944 Alan Hayward <alan.hayward@arm.com>
7945 David Sherwood <david.sherwood@arm.com>
7947 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7948 and CONST_VECTOR_NUNITS instead of computing the number of units from
7949 the byte sizes of the vector and element.
7950 (simplify_binary_operation_1): Likewise.
7951 (simplify_const_binary_operation): Likewise.
7952 (simplify_ternary_operation): Likewise.
7954 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7955 Alan Hayward <alan.hayward@arm.com>
7956 David Sherwood <david.sherwood@arm.com>
7958 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7959 (int_mem_offset): ...this new function.
7960 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7961 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7962 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7965 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7966 Alan Hayward <alan.hayward@arm.com>
7967 David Sherwood <david.sherwood@arm.com>
7969 * lower-subreg.c (interesting_mode_p): New function.
7970 (compute_costs, find_decomposable_subregs, decompose_register)
7971 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7972 (resolve_clobber, dump_choices): Use it.
7974 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7975 Alan Hayward <alan.hayward@arm.com>
7976 David Sherwood <david.sherwood@arm.com>
7978 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7980 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7981 Alan Hayward <alan.hayward@arm.com>
7982 David Sherwood <david.sherwood@arm.com>
7984 * alias.c (find_base_value, find_base_term): Only process integer
7985 truncations. Check the precision rather than the size.
7987 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7988 Alan Hayward <alan.hayward@arm.com>
7989 David Sherwood <david.sherwood@arm.com>
7991 * machmode.h (is_narrower_int_mode): New function
7992 * optabs.c (expand_float, expand_fix): Use it.
7993 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7995 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7996 Alan Hayward <alan.hayward@arm.com>
7997 David Sherwood <david.sherwood@arm.com>
7999 * rtl.h (narrower_subreg_mode): New function.
8000 * ira-color.c (update_costs_from_allocno): Use it.
8002 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8003 Alan Hayward <alan.hayward@arm.com>
8004 David Sherwood <david.sherwood@arm.com>
8006 * optabs-query.h (convert_optab_p): New function, split out from...
8007 (convert_optab_handler): ...here.
8008 (widening_optab_handler): Delete.
8009 (find_widening_optab_handler): Remove permit_non_widening parameter.
8010 (find_widening_optab_handler_and_mode): Likewise. Provide an
8011 override that operates on mode class wrappers.
8012 * optabs-query.c (widening_optab_handler): Delete.
8013 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8014 parameter. Assert that the two modes are the same class and that
8015 the "from" mode is narrower than the "to" mode. Use
8016 convert_optab_handler instead of widening_optab_handler.
8017 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8018 instead of widening_optab_handler.
8019 * expr.c (expand_expr_real_2): Update calls to
8020 find_widening_optab_handler.
8021 * optabs.c (expand_widen_pattern_expr): Likewise.
8022 (expand_binop_directly): Take the insn_code as a parameter.
8023 (expand_binop): Only call find_widening_optab_handler for
8024 conversion optabs; use optab_handler otherwise. Update calls
8025 to find_widening_optab_handler and expand_binop_directly.
8026 Use convert_optab_handler instead of widening_optab_handler.
8027 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8028 find_widening_optab_handler and use scalar_mode rather than
8030 (convert_plusminus_to_widen): Likewise.
8032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8033 Alan Hayward <alan.hayward@arm.com>
8034 David Sherwood <david.sherwood@arm.com>
8036 * machmode.h (fixed_size_mode): New class.
8037 * rtl.h (get_pool_mode): Return fixed_size_mode.
8038 * gengtype.c (main): Add fixed_size_mode.
8039 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8040 (get_raw_arg_mode): Likewise.
8041 * doc/tm.texi: Regenerate.
8042 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8043 * targhooks.c (default_get_reg_raw_mode): Likewise.
8044 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8045 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8046 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8047 (msp430_get_raw_result_mode): Likewise.
8048 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8049 * dbxout.c (dbxout_parms): Require fixed-size modes.
8050 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8051 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8052 * omp-low.c (lower_oacc_reductions): Likewise.
8053 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8054 (simplify_subreg): Update accordingly.
8055 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8056 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8057 that aren't fixed-size.
8058 (get_pool_mode): Return a fixed_size_mode.
8059 (output_constant_pool_2): Take a fixed_size_mode.
8061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8062 Alan Hayward <alan.hayward@arm.com>
8063 David Sherwood <david.sherwood@arm.com>
8065 * doc/rtl.texi (vec_series): Document.
8066 (const): Say that the operand can be a vec_series.
8067 * rtl.def (VEC_SERIES): New rtx code.
8068 * rtl.h (const_vec_series_p_1): Declare.
8069 (const_vec_series_p): New function.
8070 * emit-rtl.h (gen_const_vec_series): Declare.
8071 (gen_vec_series): Likewise.
8072 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8073 (gen_vec_series): Likewise.
8074 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8075 * simplify-rtx.c (simplify_unary_operation): Handle negations
8077 (simplify_binary_operation_series): New function.
8078 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8079 (test_vector_ops_series): New function.
8080 (test_vector_ops): Call it.
8081 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8082 gen_const_vec_series.
8083 (altivec_lvsr): Likewise.
8084 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8086 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8087 Alan Hayward <alan.hayward@arm.com>
8088 David Sherwood <david.sherwood@arm.com>
8090 * doc/rtl.texi (const): Update description of address constants.
8091 Say that vector constants are allowed too.
8092 * common.md (E, F): Use CONSTANT_P instead of checking for
8094 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8095 checking for CONST_VECTOR.
8096 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8098 * expr.c (expand_expr_real_2): Check for vector modes instead
8099 of checking for CONST_VECTOR.
8100 * rtl.h (const_vec_p): New function.
8101 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8102 (unwrap_const_vec_duplicate): Handle them here too.
8104 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8105 David Malcolm <dmalcolm@redhat.com>
8106 Alan Hayward <alan.hayward@arm.com>
8107 David Sherwood <david.sherwood@arm.com>
8109 * rtl.h (vec_duplicate_p): New function.
8110 * selftest-rtl.c (assert_rtx_eq_at): New function.
8111 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8112 (assert_rtx_eq_at): Declare.
8113 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8114 * selftest-run-tests.c (selftest::run_tests): Call it.
8115 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8116 (simplify_unary_operation_1): Recursively handle vector duplicates.
8117 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8119 (simplify_subreg): Handle subregs of vector duplicates.
8120 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8121 (selftest::simplify_rtx_c_tests): New functions.
8123 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8124 Alan Hayward <alan.hayward@arm.com>
8125 David Sherwood <david.sherwood@arm.com>
8127 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8128 (gen_vec_duplicate): Likewise.
8129 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8131 (gen_const_vector): ...here.
8132 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8133 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8134 whose elements are all equal.
8135 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8136 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8137 (simplify_relational_operation): Likewise.
8138 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8140 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8141 (aarch64_expand_vector_init): Likewise.
8142 * config/arm/arm.c (neon_vdup_constant): Likewise.
8143 (neon_expand_vector_init): Likewise.
8144 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8145 (arm_block_set_unaligned_vect): Likewise.
8146 (arm_block_set_aligned_vect): Likewise.
8147 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8148 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8149 (expand_vec_perm_even_odd_pack): Likewise.
8150 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8151 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8152 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8153 gen_const_vec_duplicate.
8154 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8155 * config/mips/mips.c (mips_gen_const_int_vector): Use
8156 gen_const_vec_duplicate.
8157 (mips_expand_vector_init): Use CONST0_RTX.
8158 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8159 (define_split): Use gen_const_vec_duplicate.
8160 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8161 (define_split): Use gen_const_vec_duplicate.
8162 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8163 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8164 * config/spu/spu.c (spu_const): Likewise.
8166 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8167 Alan Hayward <alan.hayward@arm.com>
8168 David Sherwood <david.sherwood@arm.com>
8170 * combine.c (can_change_dest_mode): Reject changes in
8171 REGMODE_NATURAL_SIZE.
8173 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8175 * configure.ac (--enable-libssp): New.
8176 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8177 trying to determine target-specific default. Adjust indentation.
8178 * configure: Regenerated.
8179 * doc/install.texi (Configuration): Expand --disable-libssp
8182 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8184 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8187 2017-10-31 Martin Jambor <mjambor@suse.cz>
8190 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8192 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8194 * auto-profile.c (autofdo_source_profile::read): Use
8195 UNKNOWN_LOCATION rather than 0.
8196 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8197 (warning_at): ...this overload.
8198 (warning_at_rich_loc_n): Rename to...
8199 (warning_n): ...this overload.
8200 (error_at_rich_loc): Rename to...
8201 (error_at): ...this overload.
8202 (pedwarn_at_rich_loc): Rename to...
8203 (pedwarn): ...this overload.
8204 (permerror_at_rich_loc): Rename to...
8205 (permerror): ...this overload.
8206 (inform_at_rich_loc): Rename to...
8207 (inform): ...this overload.
8208 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8209 (diagnostic_n_impl_richloc): Rename to...
8210 (diagnostic_n_impl): ...this rich_location *-based decl.
8211 (inform_at_rich_loc): Rename to...
8212 (inform): ...this, and add an assertion.
8213 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8214 (warning_at_rich_loc): Rename to...
8215 (warning_at): ...this, and add an assertion.
8216 (warning_at_rich_loc_n): Rename to...
8217 (warning_n): ...this, and add an assertion.
8218 (warning_n): Update location_t-based implementation for removal of
8219 location_t-based diagnostic_n_impl.
8220 (pedwarn_at_rich_loc): Rename to...
8221 (pedwarn): ...this, and add an assertion.
8222 (permerror_at_rich_loc): Rename to...
8223 (permerror): ...this, and add an assertion.
8224 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8225 (error_at_rich_loc): Rename to...
8226 (error_at): ...this, and add an assertion.
8227 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8228 (driver::do_spec_on_infiles): Likewise.
8229 * substring-locations.c (format_warning_va): Update for renaming
8230 of inform_at_rich_loc.
8232 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8234 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8235 _Float<N>X built-in functions so that the variant without the
8236 "__builtin_" prefix is only enabled for the GNU C and Objective C
8237 languages when they are in non-strict ANSI/ISO mode.
8238 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8239 * target.def (floatn_builtin_p): Add a target hook to control
8240 whether _Float<N> and _Float<N>X built-in functions without the
8241 "__builtin_" prefix are enabled, and return true for C and
8242 Objective C in the default hook. Include langhooks.h in
8244 * targhooks.h (default_floatn_builtin_p): Likewise.
8245 * targhooks.c (default_floatn_builtin_p): Likewise.
8246 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8247 floatn_builtin_p target hook.
8248 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8250 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8251 Eric Botcazou <ebotcazou@adacore.com>
8253 PR rtl-optimization/81803
8254 * lra-constraints.c (curr_insn_transform): Also reload the whole
8255 register for a strict subreg no wider than a word if this is for
8256 a WORD_REGISTER_OPERATIONS target.
8258 2017-10-31 Jason Merrill <jason@redhat.com>
8260 * gdbinit.in: Skip over inlines from timevar.h.
8262 2017-10-31 Martin Liska <mliska@suse.cz>
8264 * doc/gcov.texi: Document new option.
8265 * gcov.c (print_usage): Likewise print it.
8266 (process_args): Support the argument.
8267 (format_count): New function.
8268 (format_gcov): Use the function.
8270 2017-10-31 Martin Liska <mliska@suse.cz>
8272 * gcov.c (struct name_map): do not use typedef.
8273 Define operator== and operator<.
8274 (name_search): Remove.
8275 (name_sort): Remove.
8276 (main): Do not allocate names.
8277 (process_file): Add vertical space.
8278 (generate_results): Use std::find.
8279 (release_structures): Do not release memory.
8280 (find_source): Use std::find.
8282 2017-10-31 Martin Liska <mliska@suse.cz>
8284 * gcov.c (struct line_info): Remove it's typedef.
8285 (line_info::line_info): Add proper ctor.
8286 (line_info::has_block): Do not use a typedef.
8287 (struct source_info): Do not use typedef.
8288 (circuit): Likewise.
8289 (get_cycles_count): Likewise.
8290 (output_intermediate_file): Iterate via vector iterator.
8291 (add_line_counts): Use std::vector methods.
8292 (accumulate_line_counts): Likewise.
8293 (output_lines): Likewise.
8295 2017-10-31 Martin Liska <mliska@suse.cz>
8297 * gcov.c (struct source_info): Remove typedef.
8298 (source_info::source_info): Add proper ctor.
8299 (accumulate_line_counts): Use struct, not it's typedef.
8300 (output_gcov_file): Likewise.
8301 (output_lines): Likewise.
8302 (main): Do not allocate an array.
8303 (output_intermediate_file): Use size of vector container.
8304 (process_file): Resize the vector.
8305 (generate_results): Do not preallocate, use newly added vector
8307 (release_structures): Do not release sources.
8308 (find_source): Use vector methods.
8309 (add_line_counts): Do not use typedef.
8311 2017-10-31 Martin Liska <mliska@suse.cz>
8313 * doc/gcov.texi: Document that.
8314 * gcov.c (add_line_counts): Mark lines with a non-executed
8316 (output_line_beginning): Handle such lines.
8317 (output_lines): Pass new argument.
8318 (output_intermediate_file): Print it in intermediate format.
8320 2017-10-31 Martin Liska <mliska@suse.cz>
8322 * color-macros.h: New file.
8323 * diagnostic-color.c: Factor out color related to macros to
8325 * doc/gcov.texi: Document -k option.
8326 * gcov.c (INCLUDE_STRING): Include string.h.
8327 (print_usage): Add -k option.
8328 (process_args): Parse it.
8329 (pad_count_string): New function.
8330 (output_line_beginning): Likewise.
8331 (DEFAULT_LINE_START): New macro.
8332 (output_lines): Support color output.
8334 2017-10-31 Martin Liska <mliska@suse.cz>
8336 PR gcov-profile/82633
8337 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8338 their interaction with GCOV infrastructure.
8339 * configure.ac: Add -fkeep-{inline,static}-functions to
8341 * configure: Regenerate.
8343 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8346 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8348 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8351 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8352 into a register if it does not fit into an immediate offset field.
8354 2017-10-31 Olivier Hainque <hainque@adacore.com>
8356 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8358 2017-10-31 Julia Koval <julia.koval@intel.com>
8360 * config.gcc: Add gfniintrin.h.
8361 * config/i386/gfniintrin.h: New.
8362 * config/i386/i386-builtin-types.def
8363 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8364 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8365 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8366 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8367 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8368 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8369 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8370 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8371 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8372 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8373 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8374 * config/i386/immintrin.h: Include gfniintrin.h.
8375 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8377 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8379 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8381 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8383 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8384 (arm_ashldi3_1bit): Remove pattern.
8385 (ashrdi3): Remove shift by 1 expansion.
8386 (arm_ashrdi3_1bit): Remove pattern.
8387 (lshrdi3): Remove shift by 1 expansion.
8388 (arm_lshrdi3_1bit): Remove pattern.
8389 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8390 cost of ashldi3 by 1.
8391 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8392 (<shift>di3_neon): Likewise.
8394 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8396 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8397 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8398 and (*aarch64_simd_mov<VQ:mode>).
8399 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8400 pattern alternative.
8401 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8402 attributes to match pattern alternative.
8404 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8406 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8407 * config/rs6000/emmintrin.h: New file.
8408 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8410 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8412 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8413 * (movdi_vfp_cortexa8): Remove pattern.
8415 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8417 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8419 (Specific, alpha*-dec-osf5.1): Remove.
8420 (Specific, mips-sgi-irix5): Remove.
8421 (Specific, mips-sgi-irix6): Remove.
8423 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8426 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8427 arguments to clear_bit_region_be.
8429 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8431 * gimplify.c: Include memmodel.h.
8433 2017-10-30 Martin Jambor <mjambor@suse.cz>
8435 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8436 condition whether loop should be executed at all.
8438 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8440 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8441 gimple folding of vec_madd() intrinsics.
8442 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8443 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8444 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8446 2017-10-30 Richard Biener <rguenther@suse.de>
8448 PR tree-optimization/82762
8450 2017-10-23 Richard Biener <rguenther@suse.de>
8452 PR tree-optimization/82129
8454 2017-08-01 Richard Biener <rguenther@suse.de>
8456 PR tree-optimization/81181
8457 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8458 (compute_antic): ... end of iteration here.
8460 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8462 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8463 -std=iso9899:2017 and -std=gnu17.
8464 * doc/standards.texi (C Language): Document C17 support.
8465 * doc/cpp.texi (Overview): Mention -std=c17.
8466 (Standard Predefined Macros): Document C11 and C17 values of
8467 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8468 * doc/extend.texi (Inline): Do not list individual options for
8469 standards newer than C99.
8470 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8472 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8475 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8477 * asan.c (asan_finish_file): Align asan globals array by shadow
8480 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8483 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8484 (struct store_immediate_info): Add bitregion_start and bitregion_end
8486 (store_immediate_info::store_immediate_info): Add brs and bre
8487 arguments and initialize bitregion_{start,end} from those.
8488 (struct merged_store_group): Add bitregion_start, bitregion_end,
8489 align_base and mask fields. Drop unnecessary struct keyword from
8490 struct store_immediate_info. Add do_merge method.
8491 (clear_bit_region_be): Use memset instead of loop storing zeros.
8492 (merged_store_group::do_merge): New method.
8493 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8494 stores as long as the surrounding bitregions have no gaps.
8495 (merged_store_group::merge_overlapping): Use do_merge.
8496 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8497 is byte aligned, rather than requiring that start and width are
8498 byte aligned. Drop unnecessary struct keyword from
8499 struct store_immediate_info. Allocate and populate also mask array.
8500 Make start of the arrays relative to bitregion_start rather than
8501 start and size them according to bitregion_{end,start} difference.
8502 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8503 struct store_immediate_info.
8504 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8505 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8506 struct keyword from struct store_immediate_info.
8507 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8508 between stores as long as the surrounding bitregions have no gaps.
8510 (struct split_store): Add orig non-static data member.
8511 (split_store::split_store): Initialize orig to false.
8512 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8513 if there is exactly a single original stmt. Change stmts argument
8514 to pointer from reference, if NULL, don't push anything to it. Add
8515 first argument, use it to optimize skipping over orig stmts that
8516 are known to be before bitpos already. Simplify.
8517 (split_group): Return unsigned int count how many stores are or
8518 would be needed rather than a bool. Add allow_unaligned argument.
8519 Change split_stores argument from reference to pointer, if NULL,
8520 only do a dry run computing how many stores would be produced.
8521 Rewritten algorithm to use both alignment and misalign if
8522 !allow_unaligned and handle bitfield stores with gaps.
8523 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8524 from bitregion_start instead of start. Compute allow_unaligned
8525 here, if true, do 2 split_group dry runs to compute which one
8526 produces fewer stores and prefer aligned if equal. Punt if
8527 new count is bigger or equal than original before emitting any
8528 statements, rather than during that. Remove no longer needed
8529 new_ssa_names tracking. Replace num_stmts with
8530 split_stores.length (). Use 32-bit stack allocated entries
8531 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8532 if possible. Handle bitfields with gaps.
8533 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8534 Compute bitregion_{start,end} for the stores and construct
8535 store_immediate_info with that. Formatting fixes.
8537 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8540 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8541 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8543 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8545 * gimplify.c: Include tm_p.h.
8547 * common.opt (gcoff): Re-add as ignored option.
8548 (gcoff1, gcoff2, gcoff3): Likewise.
8550 * Makefile.in (OBJS): Delete sdbout.o.
8551 (GTFILES): Delete $(srcdir)/sdbout.c.
8552 * debug.h: Delete sdb_debug_hooks.
8553 * final.c: Delete sdbout.h include.
8554 (final_scan_insn): Delete SDB_DEBUG check.
8555 (rest_of_clean_state): Likewise.
8556 * output.h: Delete sdb_begin_function_line.
8559 * toplev.c: Delete sdbout.h include.
8560 (process_options): Delete SDB_DEBUG check.
8561 * tree-core.h (tree_type_common): Delete pointer field of
8563 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8564 TYPE_SYMTAB_POINTER.
8565 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8566 (TYPE_SYMTAB_IS_POINTER): Delete.
8567 (TYPE_SYMTAB_IS_DIE): Renumber.
8568 * xcoffout.c: Refer to former sdbout.c file.
8569 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8571 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8572 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8573 (Debugging Options): Delete -gcoff.
8574 (-gstabs): Delete SDB reference.
8576 (-gcoff@var{level}): Delete.
8577 * doc/passes.texi (Debugging information output): Delete SDB and
8578 sdbout.c references.
8579 * doc/tm.texi: Regenerate.
8580 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8581 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8583 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8584 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8585 (SDB_DEBUGGING_INFO): Delete.
8586 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8587 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8588 * target.def (output_source_filename): Delete COFF reference.
8590 * common.opt (gcoff): Delete.
8591 (gxcoff+): Update Negative chain.
8592 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8594 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8595 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8596 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8598 (expand_function_start): Change sdb reference to past tense.
8599 (expand_function_end): Change sdb reference to past tense.
8600 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8601 * opts.c (debug_type_names): Delete coff entry.
8602 (common_handle_option): Delete OPT_gcoff case.
8603 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8605 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8606 * config/cris/cris.h: Delete SDB reference in comment.
8607 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8608 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8609 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8610 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8612 (ix86_expand_prologue): Likewise.
8613 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8614 * config/ia64/ia64.h: Likewise.
8615 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8616 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8618 * config/mmix/mmix.h: Likewise.
8619 * config/nds32/nds32.c: Likewise.
8620 * config/stormy/storym16.h: Likewise.
8621 * config/visium/visium.h: Likewise.
8622 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8624 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8626 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8627 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8628 FRAME_POINTER_REGNUM point at high end of local var area.
8630 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8632 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8633 Move comment around. Do not reset best_edge for a copiable
8634 destination if the copy would cause a partition change.
8635 (better_edge_p): Remove redundant check.
8637 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8639 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8641 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8643 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8644 for math functions that have _Float<N> and _Float<N>X variants.
8645 (mathfn_built_in_2): Add support for math functions that have
8646 _Float<N> and _Float<N>X variants.
8647 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8648 (expand_builtin_mathfn_ternary): Add support for fma with
8649 _Float<N> and _Float<N>X variants.
8650 (expand_builtin): Likewise.
8651 (fold_builtin_3): Likewise.
8652 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8653 create math function _Float<N> and _Float<N>X variants as external
8655 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8656 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8657 the __builtin_ prefix and if not strict ansi, without the prefix.
8658 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8659 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8660 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8661 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8662 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8663 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8664 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8665 function signatures for fma _Float<N> and _Float<N>X variants.
8666 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8667 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8668 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8669 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8670 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8671 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8672 * gencfn-macros.c (print_case_cfn): Add support for math functions
8673 that have _Float<N> and _Float<N>X variants.
8674 (print_define_operator_list): Likewise.
8675 (fltfn_suffixes): Likewise.
8677 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8678 for math functions that have _Float<N> and _Float<N>X variants.
8679 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8680 and _Float<N>X variants.
8681 (COPYSIGN): Likewise.
8684 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8685 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8687 (integer_valued_read_call_p): Likewise.
8688 * fold-const-call.c (fold_const_call_ss): Likewise.
8689 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8690 _Float<N> and _Float<N>X variants.
8691 (fold_const_call_ssss): Add support for fma _Float<N> and
8692 _Float<N>X variants.
8693 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8694 support for copysign and fma _Float<N> and _Float<N>X variants.
8695 (backprop::process_builtin_call_use): Likewise.
8696 * tree-call-cdce.c (can_test_argument_range); Add support for
8697 sqrt _Float<N> and _Float<N>X variants.
8698 (edom_only_function): Likewise.
8699 (get_no_error_domain): Likewise.
8700 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8701 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8702 copysign _Float<N> and _Float<N>X variants.
8703 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8704 handled by machine independent code.
8705 (FMAF128): Likewise.
8706 * doc/cpp.texi (Common Predefined Macros): Document defining
8707 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8708 fma _Float<N> and _Float<N>X variants.
8710 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8713 * config/i386/i386-modes.def (CCFPU): Remove definition.
8714 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8715 (ix86_cc_modes_compatible): Ditto.
8716 (ix86_expand_carry_flag_compare): Ditto.
8717 (ix86_expand_int_movcc): Ditto.
8718 (ix86_expand_int_addcc): Ditto.
8719 (ix86_reverse_condition): Ditto.
8720 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8721 Return true/false for unordered/ordered fp comparisons.
8722 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8723 (ix86_prepare_fp_compare_args): Update for rename.
8724 (ix86_expand_fp_compare): Update for rename. Generate unordered
8725 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8726 (ix86_expand_sse_compare_and_jump): Ditto.
8727 * config/i386/predicates.md (fcmov_comparison_operator):
8728 Remove CCFPU mode handling.
8729 (ix86_comparison_operator): Ditto.
8730 (ix86_carry_flag_operator): Ditto.
8731 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8732 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8733 (*cmpu<mode>_cc_i387): Ditto.
8734 (FPCMP): Remove mode iterator.
8735 (unord): Remove mode attribute.
8736 (unord_subst): New define_subst transformation
8737 (unord): New define_subst attribute.
8739 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8740 (*cmpi<unord>xf_i387): Ditto.
8741 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8742 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8743 using unord_subst transformation.
8744 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8745 (round_saeonly): Also handle CCFP mode.
8746 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8747 Remove UNSPEC_SAHF unspec handling.
8749 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8751 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8753 2017-10-27 Jeff Law <law@redhat.com>
8755 * gimple-ssa-sprintf.c: Include domwalk.h.
8756 (class sprintf_dom_walker): New class, derived from dom_walker.
8757 (sprintf_dom_walker::before_dom_children): New function.
8758 (struct call_info): Moved into sprintf_dom_walker class
8759 (compute_formath_length, handle_gimple_call): Likewise.
8760 (sprintf_length::execute): Call the dominator walker rather
8761 than walking the statements.
8763 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8764 gimple statement locations.
8765 (check_array_bounds): Corresponding changes. Get the statement's
8766 location directly from wi->stmt.
8768 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8771 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8773 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8775 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8776 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8778 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8781 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8782 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8783 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8784 maybe_get_pool_constant.
8785 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8788 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8790 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8791 2.26 caveat. Update gas and gld versions.
8792 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8795 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8797 * cgraph.h (set_malloc_flag): Declare.
8798 * cgraph.c (set_malloc_flag_1): New function.
8799 (set_malloc_flag): Likewise.
8800 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8801 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8803 (read_ipa_call_summary): Add support for reading is_return_callee.
8804 (write_ipa_call_summary): Stream is_return_callee.
8805 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8806 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8807 ipa-prop.h, ipa-fnsummary.h.
8808 (pure_const_names): Change to static.
8809 (malloc_state_e): Define.
8810 (malloc_state_names): Define.
8811 (funct_state_d): Add field malloc_state.
8812 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8813 (check_retval_uses): New function.
8814 (malloc_candidate_p): Likewise.
8815 (analyze_function): Add support for malloc attribute.
8816 (pure_const_write_summary): Stream malloc_state.
8817 (pure_const_read_summary): Add support for reading malloc_state.
8818 (dump_malloc_lattice): New function.
8819 (propagate_malloc): New function.
8820 (warn_function_malloc): New function.
8821 (ipa_pure_const::execute): Call propagate_malloc and
8822 ipa_free_fn_summary.
8823 (pass_local_pure_const::execute): Add support for malloc attribute.
8824 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8825 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8827 2017-10-27 Martin Liska <mliska@suse.cz>
8829 PR gcov-profile/82457
8830 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8831 for fork-like functions to be properly instrumented.
8833 2017-10-27 Richard Biener <rguenther@suse.de>
8836 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8837 info when we redirected EH.
8839 2017-10-26 Michael Collison <michael.collison@arm.com>
8841 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8843 (<optab>_trunchf<GPI:mode>2: New pattern.
8844 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8845 * config/aarch64/iterators.md (wv): New mode attribute.
8846 (vf, VF): New mode attributes.
8847 (vgp, VGP): New mode attributes.
8848 (s): Update attribute with SImode and DImode prefixes.
8850 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8852 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8853 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8854 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8855 (nios2_option_overide): Initialize it. Don't allow R0-relative
8856 addressing with PIC.
8857 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8858 (nios2_symbolic_constant_p): Likewise.
8859 (nios2_legitimate_address_p): Likewise.
8860 (nios2_r0rel_section_name_p): New.
8861 (nios2_symbol_ref_in_r0rel_data_p): New.
8862 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8863 (r0rel_constant_p): New.
8864 (nios2_print_operand_address): Handle r0rel_constant_p.
8865 (nios2_cdx_narrow_form_p): Likewise.
8866 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8867 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8868 (Nios II Options): Document -mr0rel-sec.
8870 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8872 * config/nios2/nios2.c: Include xregex.h.
8873 (nios2_gprel_sec_regex): New.
8874 (nios2_option_overide): Initialize it. Don't allow GP-relative
8875 addressing with PIC.
8876 (nios2_small_section_name_p): Check for regex match.
8877 * config/nios2/nios2.opt (mgprel-sec=): New option.
8878 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8879 (Nios II Options): Document -mgprel-sec.
8881 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8883 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8885 2017-10-26 Tom de Vries <tom@codesourcery.com>
8887 PR tree-optimization/82707
8888 * gimple.c (gimple_copy): Fix unsharing of
8889 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8891 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8893 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8894 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8895 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8896 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8897 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8898 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8899 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8900 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8901 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8902 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8903 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8904 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8905 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8906 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8907 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8908 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8909 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8911 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8913 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8915 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8916 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8917 warning to rs6000.c. Remove the Undocumented flag, since it has
8919 (-mabi=ibmlongdouble): Likewise.
8920 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8921 already set, set the default format for long double.
8922 (rs6000_debug_reg_global): Print whether long double is IBM or
8924 (rs6000_option_override_internal): Rework setting long double
8925 format. Only warn if the user is changing the long double default
8926 and they did not use -Wno-psabi.
8927 * doc/invoke.texi (PowerPC options): Update the documentation for
8928 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8930 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8931 Alan Hayward <alan.hayward@arm.com>
8932 David Sherwood <david.sherwood@arm.com>
8934 * rtl.h (wider_subreg_mode): New function.
8935 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8936 rather than an unsigned int *.
8937 * ira-color.c (regno_max_ref_width): Replace with...
8938 (regno_max_ref_mode): ...this new variable.
8939 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8940 Use wider_subreg_mode.
8941 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8942 rather than an unsigned int *.
8943 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8944 (process_alt_operands): Likewise.
8945 (invariant_p): Likewise.
8946 * lra-spills.c (assign_mem_slot): Likewise.
8947 (add_pseudo_to_slot): Likewise.
8948 * lra.c (collect_non_operand_hard_regs): Likewise.
8949 (add_regs_to_insn_regno_info): Likewise.
8950 * reload1.c (regno_max_ref_width): Replace with...
8951 (regno_max_ref_mode): ...this new variable.
8952 (reload): Update accordingly. Update call to
8953 ira_sort_regnos_for_alter_reg.
8954 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8955 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8956 (scan_paradoxical_subregs): Likewise.
8958 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8960 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8961 (aarch64_frame): Add emit_frame_chain boolean.
8962 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8963 Move eh_return case to aarch64_layout_frame.
8964 (aarch64_layout_frame): Initialize emit_frame_chain.
8965 (aarch64_expand_prologue): Use emit_frame_chain.
8967 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8969 * config/aarch64/aarch64.c (aarch64_layout_frame):
8970 Ensure LR is always stored at the bottom of the callee-saves.
8971 Remove rarely used frame layout which saves callee-saves at top of
8972 frame, so the store of LR can be used as a valid probe in all cases.
8974 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8976 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8977 Improve unaligned TImode/TFmode base/offset split.
8979 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8980 Alan Hayward <alan.hayward@arm.com>
8981 David Sherwood <david.sherwood@arm.com>
8983 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8984 * combine.c (find_single_use_1): Likewise.
8985 (expand_field_assignment): Likewise.
8986 (move_deaths): Likewise.
8987 * lra-constraints.c (simplify_operand_subreg): Likewise.
8988 (curr_insn_transform): Likewise.
8989 * lra.c (collect_non_operand_hard_regs): Likewise.
8990 (add_regs_to_insn_regno_info): Likewise.
8991 * rtlanal.c (reg_referenced_p): Likewise.
8992 (covers_regno_no_parallel_p): Likewise.
8994 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8996 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8997 Don't print any bits outside the precision of the value.
8998 * wide-int.cc (test_printing): Add some new tests.
9000 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9002 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9003 supports -xbrace_comment option.
9004 * configure: Regenerate.
9005 * config.in: Regenerate.
9006 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9007 (ASM_CPU_SPEC): Use it.
9009 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9011 * target.def (static_rtx_alignment): New hook.
9012 * targhooks.h (default_static_rtx_alignment): Declare.
9013 * targhooks.c (default_static_rtx_alignment): New function.
9014 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9015 * doc/tm.texi: Regenerate.
9016 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9017 instead of targetm.constant_alignment. Remove call to
9019 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9020 (cris_preferred_mininum_alignment): New function, split out from...
9021 (cris_constant_alignment): ...here.
9022 (cris_static_rtx_alignment): New function.
9023 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9025 (ix86_constant_alignment): ...here.
9026 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9027 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9028 (mmix_static_rtx_alignment): New function.
9029 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9030 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9032 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9035 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9036 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9038 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9040 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9041 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9042 mark as a sign-extending load.
9043 (local_pic_load_u): Define.
9045 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9048 * fold-const.c (operand_equal_for_comparison_p): Also return true
9049 if ARG0 is a simple variant of ARG1 with narrower precision.
9050 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9052 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9054 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9056 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9057 scatter_static, scatter_per_elt.
9058 * x86-tune-costs.h: Add new cost entries.
9060 2017-10-25 Richard Biener <rguenther@suse.de>
9062 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9063 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9064 class pass_fre): Move to ...
9065 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9066 class pass_fre): ... here and adjust for statistics.
9068 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9071 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9072 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9073 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9074 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9077 2017-10-25 Richard Biener <rguenther@suse.de>
9079 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9080 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9081 eliminate_push_avail, eliminate_insert): Move inside...
9082 (class eliminate_dom_walker): ... this class in preparation
9084 (fini_eliminate): Remove by merging with ...
9085 (eliminate): ... this function. Adjust for class changes.
9086 (pass_pre::execute): Remove fini_eliminate call.
9087 (pass_fre::execute): Likewise.
9089 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9092 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9093 (VPERMI2, VPERMI2I): New mode iterators.
9094 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9095 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9097 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9098 mode iterator. Remove 3 old define_insn patterns.
9099 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9100 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9101 VPERMI2 mode iterator, remove the other two expanders.
9102 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9103 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9104 instructions, remove the other two patterns.
9105 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9106 mode iterator, remove the other two patterns.
9107 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9108 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9109 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9110 and adjust argument order accordingly.
9111 (ix86_expand_vec_perm): Adjust caller.
9112 (expand_vec_perm_1): Likewise.
9113 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9114 (expand_vec_perm_vpermt2_vpshub2): ... this.
9115 (ix86_expand_vec_perm_const_1): Adjust caller.
9116 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9119 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9120 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9121 (vec_shl_<mode>): Remove unused expander.
9122 (avx512bw_<shift_insn><mode>3): New define_insn.
9123 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9124 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9126 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9129 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9132 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9134 PR rtl-optimization/82396
9135 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9136 (autopref_multipass_init): Simplify initialization.
9137 (autopref_rank_data): Simplify sort order.
9138 * gcc/sched-int.h (autopref_multipass_data_): Remove
9139 multi_mem_insn_p, min_offset and max_offset.
9141 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9144 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9145 Check special value of flag_omit_frame_pointer.
9146 (aarch64_can_eliminate): Likewise.
9147 (aarch64_override_options_after_change_1): Simplify handling of
9148 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9150 2017-10-24 Richard Biener <rguenther@suse.de>
9152 PR tree-optimization/82697
9153 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9154 zero for conditional load and unconditional store.
9156 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9158 * doc/install.texi: Document bootstrap-cet.
9160 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9163 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9164 ENDBR instruction at function entrance if function is only
9167 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9170 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9171 patterns to better describe from which operation the CF is computed.
9172 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9173 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9174 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9175 is 0, use _0 suffixed expanders instead of emitting a comparison
9178 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9180 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9181 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9183 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9186 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9187 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9188 * loop-iv.c (iv_get_reaching_def): Likewise.
9189 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9190 variable is promoted and the partition contains undefined values.
9192 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9194 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9196 (nios2_address_cost): Define.
9197 (nios2_legitimize_address): Recognize (exp + constant) directly.
9198 (TARGET_ADDRESS_COST): Define.
9200 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9202 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9203 (nios2_symbolic_memory_operand_p): Declare.
9204 (nios2_split_large_constant): Declare.
9205 (nios2_split_symbolic_memory_operand): Declare.
9206 * config/nios2/nios2.c: Adjust includes.
9207 (nios2_symbolic_constant_allowed): New.
9208 (nios2_symbolic_constant_p): New.
9209 (nios2_plus_symbolic_constant_p): New.
9210 (nios2_valid_addr_expr_p): Recognize addresses involving
9212 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9213 (nios2_symbolic_memory_operand_p): New.
9214 (nios2_large_constant_p): New.
9215 (nios2_split_large_constant): New.
9216 (nios2_split_plus_large_constant): New.
9217 (nios2_split_symbolic_memory_operand): New.
9218 (nios2_legitimize_address): Code refactoring. Handle addresses
9219 involving symbolic constants.
9220 (nios2_emit_move_sequence): Likewise.
9221 (nios2_print_operand): Improve error output.
9222 (nios2_print_operand_address): Handle LO_SUM.
9223 (nios2_cdx_narrow_form_p): Likewise.
9224 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9225 operands involving symbolic constants.
9226 (movhi_internal, movsi_internal): Likewise.
9227 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9228 (extendhisi2, extendqi<mode>2): Likewise.
9230 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9232 * tree-pass.h (PROP_rtl_split_insns): Define.
9233 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9235 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9237 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9239 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9242 * target.def (const_not_ok_for_debug_p): Default to
9243 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9244 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9245 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9246 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9247 which targetm.const_not_ok_for_debug_p returned true.
9248 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9250 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9252 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9253 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9254 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9256 (ix86_const_not_ok_for_debug_p): New function.
9257 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9258 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9260 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9263 * system.h: Conditionally include "unique-ptr.h" if
9264 INCLUDE_UNIQUE_PTR is defined.
9265 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9266 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9268 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9270 * config/rl78/rl78.md: New define_expand "subdi3".
9272 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9275 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9276 DF_REF_INSN if DF_REF_INSN_INFO is false.
9278 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9280 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9281 xmm_move instead of sse_move.
9282 (sse_store_index): New function.
9283 (ix86_register_move_cost): Be more sensible about mismatch stall;
9284 model AVX moves correctly; make difference between sse->integer and
9286 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9287 moves; make difference between SSE and AVX.
9288 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9289 and zmm_move. Increase size of sse load and store tables;
9290 add unaligned load and store tables; add ssemmx_to_integer.
9291 * x86-tune-costs.h: Update all entries according to real
9292 move latencies from Agner Fog's manual and chip documentation.
9294 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9297 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9298 * config/i386/constraints.md (Wf): New constraint.
9299 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9300 (cmp<dwi>_doubleword): Removed.
9301 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9302 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9303 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9304 expand with cmp<dwi>_doubleword. For LTU and GEU use
9305 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9307 * common.opt (gcolumn-info): Enable by default.
9308 * doc/invoke.texi (gcolumn-info): Document new default.
9310 2017-10-23 Richard Biener <rguenther@suse.de>
9312 PR tree-optimization/82672
9313 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9314 Fold the stmt if we propagated into it.
9316 2017-10-23 Richard Biener <rguenther@suse.de>
9318 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9319 (bitmap_remove_expr_from_set): ... this. All callers call this
9320 for non-constant values.
9321 (bitmap_set_subtract): Rename to...
9322 (bitmap_set_subtract_expressions): ... this. Adjust and
9324 (bitmap_set_contains_value): Remove superfluous check.
9325 (bitmap_set_replace_value): Inline into single caller ...
9326 (bitmap_value_replace_in_set): ... here and simplify.
9327 (dependent_clean): Merge into ...
9328 (clean): ... this using an overload. Adjust.
9329 (prune_clobbered_mems): Adjust.
9330 (compute_antic_aux): Likewise.
9331 (compute_partial_antic_aux): Likewise.
9333 2017-10-23 Richard Biener <rguenther@suse.de>
9335 PR tree-optimization/82129
9337 2017-08-01 Richard Biener <rguenther@suse.de>
9339 PR tree-optimization/81181
9340 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9341 (compute_antic): ... end of iteration here.
9343 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9345 * target.def (starting_frame_offset): New hook.
9346 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9347 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9348 * doc/tm.texi.in: Regenerate.
9349 * hooks.h (hook_hwi_void_0): Declare.
9350 * hooks.c (hook_hwi_void_0): New function.
9351 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9352 STARTING_FRAME_OFFSET.
9353 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9354 * reload1.c (reload): Likewise.
9355 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9356 instead of STARTING_FRAME_OFFSET.
9357 * function.c (try_fit_stack_local): Likewise.
9358 (assign_stack_local_1): Likewise
9359 (instantiate_virtual_regs): Likewise.
9360 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9361 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9362 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9363 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9364 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9365 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9366 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9367 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9368 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9369 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9370 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9371 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9372 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9373 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9374 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9375 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9376 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9377 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9378 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9379 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9380 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9381 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9382 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9383 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9384 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9385 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9386 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9387 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9388 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9389 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9390 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9391 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9392 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9393 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9394 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9395 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9396 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9397 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9398 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9399 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9400 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9401 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9402 return a HOST_WIDE_INT.
9403 (avr_builtin_setjmp_frame_value): Use it instead of
9404 STARTING_FRAME_OFFSET.
9405 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9406 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9407 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9409 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9410 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9411 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9412 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9413 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9414 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9415 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9416 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9417 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9418 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9419 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9420 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9422 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9423 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9424 * config/mips/mips.c (mips_compute_frame_info): Refer to
9425 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9426 (mips_starting_frame_offset): New function.
9427 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9428 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9429 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9430 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9431 and return a HOST_WIDE_INT.
9432 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9433 (mmix_initial_elimination_offset): Refer to
9434 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9435 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9436 * config/pa/pa.c (pa_starting_frame_offset): New function.
9437 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9438 (pa_expand_prologue): Likewise.
9439 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9440 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9441 !FRAME_GROWS_DOWNWARD handling to...
9442 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9443 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9444 !FRAME_GROWS_DOWNWARD handling to...
9445 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9446 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9447 !FRAME_GROWS_DOWNWARD handling to...
9448 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9449 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9451 (rs6000_starting_frame_offset): New function.
9452 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9453 !FRAME_GROWS_DOWNWARD handling to...
9454 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9455 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9456 !FRAME_GROWS_DOWNWARD handling to...
9457 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9458 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9459 !FRAME_GROWS_DOWNWARD handling to...
9460 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9461 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9462 (rs6000_starting_frame_offset): New function.
9463 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9464 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9465 * config/vax/vax.c (vax_starting_frame_offset): New function.
9466 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9467 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9468 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9469 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9470 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9471 * system.h (STARTING_FRAME_OFFSET): Poison.
9473 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9475 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9476 SCALAR_TYPE_MODE instead of TYPE_MODE.
9478 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9479 Alan Hayward <alan.hayward@arm.com>
9480 David Sherwood <david.sherwood@arm.com>
9482 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9484 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9485 Alan Hayward <alan.hayward@arm.com>
9486 David Sherwood <david.sherwood@arm.com>
9488 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9490 2017-10-23 Richard Biener <rguenther@suse.de>
9492 PR tree-optimization/82129
9493 * tree-ssa-pre.c (bitmap_set_and): Remove.
9494 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9495 canonicalizing expressions in the set to those with lowest
9496 ID rather than taking that from the first edge.
9498 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9500 * combine.c (rtx_equal_for_field_assignment_p): Use
9501 byte_lowpart_offset.
9503 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9504 Alan Hayward <alan.hayward@arm.com>
9505 David Sherwood <david.sherwood@arm.com>
9507 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9508 to a SUBREG_PROMOTED_VAR.
9510 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9511 Alan Hayward <alan.hayward@arm.com>
9512 David Sherwood <david.sherwood@arm.com>
9514 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9515 (expand_debug_source_expr): Likewise.
9516 * combine.c (combine_simplify_rtx): Likewise.
9517 * cse.c (fold_rtx): Likewise.
9518 * optabs.c (expand_float): Likewise.
9519 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9520 (simplify_binary_operation_1): Likewise.
9522 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9523 Alan Hayward <alan.hayward@arm.com>
9524 David Sherwood <david.sherwood@arm.com>
9526 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9527 (record_promoted_value): Likewise.
9528 * expr.c (expand_expr_real_2): Likewise.
9529 * ree.c (update_reg_equal_equiv_notes): Likewise.
9530 (combine_set_extension): Likewise.
9531 * rtlanal.c (low_bitmask_len): Likewise.
9532 * simplify-rtx.c (neg_const_int): Likewise.
9533 (simplify_binary_operation_1): Likewise.
9535 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9536 Alan Hayward <alan.hayward@arm.com>
9537 David Sherwood <david.sherwood@arm.com>
9539 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9540 * regcprop.c (maybe_mode_change): Likewise.
9541 * reload1.c (alter_reg): Likewise.
9543 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9545 * inchash.h (inchash::hash::add_wide_int): New function.
9546 * lto-streamer-out.c (hash_tree): Use it.
9548 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9550 * inchash.h (inchash::hash::add_wide_int): Rename to...
9551 (inchash::hash::add_hwi): ...this.
9552 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9553 (polymorphic_call_target_hasher::hash): Likewise.
9554 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9555 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9556 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9557 * lto-streamer-out.c (hash_tree): Likewise.
9558 * optc-save-gen.awk: Likewise.
9559 * tree.c (add_expr): Likewise.
9561 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9564 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9565 for ordered inequality comparisons even with TARGET_IEEE_FP.
9567 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9570 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9571 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9572 Expand with cmp<dwi>_doubleword.
9574 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9576 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9577 List CET intrinsics.
9578 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9579 specific to -fcf-protection option.
9581 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9583 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9584 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9585 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9586 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9587 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9588 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9589 (extra_objs): Add cet.o for Linux/x86 targets.
9590 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9591 * config/i386/cet.c: New file.
9592 * config/i386/cetintrin.h: Likewise.
9593 * config/i386/t-cet: Likewise.
9594 * config/i386/cpuid.h (bit_SHSTK): New.
9595 (bit_IBT): Likewise.
9596 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9597 pass IBT and SHSTK bits.
9598 * config/i386/i386-builtin-types.def
9599 (VOID_FTYPE_UNSIGNED_PVOID): New.
9600 (VOID_FTYPE_UINT64_PVOID): Likewise.
9601 * config/i386/i386-builtin.def: Add CET intrinsics.
9602 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9603 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9604 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9605 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9607 * config/i386/i386.c (rest_of_insert_endbranch): New.
9608 (pass_data_insert_endbranch): Likewise.
9609 (pass_insert_endbranch): Likewise.
9610 (make_pass_insert_endbranch): Likewise.
9611 (ix86_notrack_prefixed_insn_p): Likewise.
9612 (ix86_target_string): Add -mibt, -mshstk flags.
9613 (ix86_option_override_internal): Add flag_cf_protection
9615 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9616 (ix86_print_operand): Add 'notrack' prefix output.
9617 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9618 (ix86_expand_builtin): Expand CET intrinsics.
9619 (x86_output_mi_thunk): Add 'endbranch' instruction.
9620 * config/i386/i386.h (TARGET_IBT): New.
9621 (TARGET_IBT_P): Likewise.
9622 (TARGET_SHSTK): Likewise.
9623 (TARGET_SHSTK_P): Likewise.
9624 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9625 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9626 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9627 (builtin_setjmp_setup): New pattern.
9628 (builtin_longjmp): Likewise.
9629 (rdssp<mode>): Likewise.
9630 (incssp<mode>): Likewise.
9631 (saveprevssp): Likewise.
9632 (rstorssp): Likewise.
9633 (wrss<mode>): Likewise.
9634 (wruss<mode>): Likewise.
9635 (setssbsy): Likewise.
9636 (clrssbsy): Likewise.
9637 (nop_endbr): Likewise.
9638 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9640 * config/i386/immintrin.h: Include <cetintrin.h>.
9641 * config/i386/linux-common.h
9642 (file_end_indicate_exec_stack_and_cet): New prototype.
9643 (TARGET_ASM_FILE_END): New.
9645 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9647 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9648 latencies instead of having separate table; make difference between
9649 integer and float costs.
9650 * i386.h (processor_costs): Remove scalar_stmt_cost,
9651 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9652 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9654 * x86-tune-costs.h: Remove entries which has been removed in
9655 procesor_costs from all tables; make cond_taken_branch_cost
9656 and cond_not_taken_branch_cost COST_N_INSNS based.
9658 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9660 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9662 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9664 * config/i386/i386.md (isa): Remove fma_avx512f.
9665 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9666 <avx512>_fmadd_<mode>_mask3<round_name>,
9667 <avx512>_fmsub_<mode>_mask<round_name>,
9668 <avx512>_fmsub_<mode>_mask3<round_name>,
9669 <avx512>_fnmadd_<mode>_mask<round_name>,
9670 <avx512>_fnmadd_<mode>_mask3<round_name>,
9671 <avx512>_fnmsub_<mode>_mask<round_name>,
9672 <avx512>_fnmsub_<mode>_mask3<round_name>,
9673 <avx512>_fmaddsub_<mode>_mask<round_name>,
9674 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9675 <avx512>_fmsubadd_<mode>_mask<round_name>,
9676 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9677 (*vec_widen_umult_even_v16si<mask_name>,
9678 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9679 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9681 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9683 * extend.texi: Add 'nocf_check' documentation.
9684 * gimple.texi: Add second parameter to
9685 gimple_build_call_from_tree.
9686 * invoke.texi: Add -fcf-protection documentation.
9687 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9689 2017-10-20 Richard Biener <rguenther@suse.de>
9691 PR tree-optimization/82473
9692 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9693 the largest input type.
9695 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9697 * c-attribs.c (handle_nocf_check_attribute): New function.
9698 (c_common_attribute_table): Add 'nocf_check' handling.
9699 * gimple-parser.c: Add second argument NULL to
9700 gimple_build_call_from_tree.
9701 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9702 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9704 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9705 * common.opt: Add fcf-protection flag.
9706 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9707 * flag-types.h: Add enum cf_protection_level.
9708 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9709 Add 'nocf_check' attribute propagation to gimple call.
9710 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9711 (gimple_build_call_from_tree): Update prototype.
9712 (gimple_call_nocf_check_p): New function.
9713 (gimple_call_set_nocf_check): Likewise.
9714 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9715 * ipa-icf.c: Add nocf_check attribute in statement hash.
9716 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9717 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9718 * toplev.c (process_options): Add flag_cf_protection handling.
9720 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9722 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9724 2017-10-20 Richard Biener <rguenther@suse.de>
9726 PR tree-optimization/82603
9727 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9728 remove false predicated stores.
9730 2017-10-20 Richard Biener <rguenther@suse.de>
9732 * graphite-isl-ast-to-gimple.c
9733 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9734 Remove return value and simplify, dump copied stmt after lhs
9736 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9737 Reduce dump verbosity.
9738 (gsi_insert_earliest): Likewise.
9739 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9740 * graphite.c (print_global_statistics): Adjust dumping.
9741 (print_graphite_scop_statistics): Likewise.
9742 (print_graphite_statistics): Do not dump loops here.
9743 (graphite_transform_loops): But here.
9745 2017-10-20 Nicolas Roche <roche@adacore.com>
9747 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9748 * configure: Regenerate.
9750 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9753 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9754 functions when optimizing replace GIMPLE_RETURN stmts with
9755 calls to __builtin_unreachable ().
9758 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9759 for -fsanitize=thread link of executables.
9760 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9761 link of executables.
9764 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9766 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9767 define_insns for logical vector shifts to use VI248_AVX512BW
9768 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9769 condition, useless isa and prefix attributes. Change the first
9770 2 of these define_insns to ...
9771 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9772 define_insn for avx512vl.
9773 (<shift_insn><mode>3): ... and this, new define_insn without
9774 masking for non-avx512vl.
9777 * config/i386/sse.md (*andnot<mode>3,
9778 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9779 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9780 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9781 not applied use empty suffix even for TARGET_AVX512VL.
9782 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9783 is applied, supply evex,evex or evex,evex,evex instead of just
9786 2017-10-20 Julia Koval <julia.koval@intel.com>
9788 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9789 (OPTION_MASK_ISA_GFNI_UNSET): New.
9790 (ix86_handle_option): Handle OPT_mgfni.
9791 * config/i386/cpuid.h (bit_GFNI): New.
9792 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9793 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9794 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9795 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9796 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9797 * config/i386/i386.opt: Add mgfni.
9799 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9801 * config/msp430/msp430.c (msp430_option_override): Disable
9802 -fdelete-null-pointer-checks.
9803 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9805 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9807 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9808 of x87 and SSE instructions.
9810 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9812 * asan.c (create_cond_insert_point): Do not update edge count.
9813 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9814 (afdo_propagate_circuit): Likewise.
9815 (afdo_calculate_branch_prob): Likewise.
9816 (afdo_annotate_cfg): Likewise.
9817 * basic-block.h (struct edge_def): Remove count.
9818 (edge_def::count): New accessor.
9819 * bb-reorder.c (rotate_loop): Update.
9820 (find_traces_1_round): Update.
9821 (connect_traces): Update.
9822 (sanitize_hot_paths): Update.
9823 * cfg.c (unchecked_make_edge): Update.
9824 (make_single_succ_edge): Update.
9825 (check_bb_profile): Update.
9826 (dump_edge_info): Update.
9827 (update_bb_profile_for_threading): Update.
9828 (scale_bbs_frequencies_int): Update.
9829 (scale_bbs_frequencies_gcov_type): Update.
9830 (scale_bbs_frequencies_profile_count): Update.
9831 (scale_bbs_frequencies): Update.
9832 * cfganal.c (connect_infinite_loops_to_exit): Update.
9833 * cfgbuild.c (compute_outgoing_frequencies): Update.
9834 (find_many_sub_basic_blocks): Update.
9835 * cfgcleanup.c (try_forward_edges): Update.
9836 (try_crossjump_to_edge): Update
9837 * cfgexpand.c (expand_gimple_cond): Update
9838 (expand_gimple_tailcall): Update
9839 (construct_exit_block): Update
9840 * cfghooks.c (verify_flow_info): Update
9841 (redirect_edge_succ_nodup): Update
9842 (split_edge): Update
9843 (make_forwarder_block): Update
9844 (duplicate_block): Update
9845 (account_profile_record): Update
9846 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9847 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9848 * cfgloopmanip.c (scale_loop_profile): Update.
9850 (lv_adjust_loop_entry_edge): Update.
9851 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9852 (force_nonfallthru_and_redirect): Update.
9853 (purge_dead_edges): Update.
9854 (rtl_flow_call_edges_add): Update.
9855 * cgraphunit.c (init_lowered_empty_function): Update.
9856 (cgraph_node::expand_thunk): Update.
9857 * gimple-pretty-print.c (dump_probability): Update.
9858 (dump_edge_probability): Update.
9859 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9860 * haifa-sched.c (sched_create_recovery_edges): Update.
9861 * hsa-gen.c (convert_switch_statements): Update.
9862 * ifcvt.c (dead_or_predicable): Update.
9863 * ipa-inline-transform.c (inline_transform): Update.
9864 * ipa-split.c (split_function): Update.
9865 * ipa-utils.c (ipa_merge_profiles): Update.
9866 * loop-doloop.c (add_test): Update.
9867 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9868 * lto-streamer-in.c (input_cfg): Update.
9869 (input_function): Update.
9870 * lto-streamer-out.c (output_cfg): Update.
9871 * modulo-sched.c (sms_schedule): Update.
9872 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9873 * predict.c (maybe_hot_edge_p): Update.
9874 (unlikely_executed_edge_p): Update.
9875 (probably_never_executed_edge_p): Update.
9876 (dump_prediction): Update.
9877 (drop_profile): Update.
9878 (propagate_unlikely_bbs_forward): Update.
9879 (determine_unlikely_bbs): Update.
9880 (force_edge_cold): Update.
9881 * profile.c (compute_branch_probabilities): Update.
9882 * reg-stack.c (better_edge): Update.
9883 * shrink-wrap.c (handle_simple_exit): Update.
9884 * tracer.c (better_p): Update.
9885 * trans-mem.c (expand_transaction): Update.
9886 (split_bb_make_tm_edge): Update.
9887 * tree-call-cdce.c: Update.
9888 * tree-cfg.c (gimple_find_sub_bbs): Update.
9889 (gimple_split_edge): Update.
9890 (gimple_duplicate_sese_region): Update.
9891 (gimple_duplicate_sese_tail): Update.
9892 (gimple_flow_call_edges_add): Update.
9893 (insert_cond_bb): Update.
9894 (execute_fixup_cfg): Update.
9895 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9896 * tree-complex.c (expand_complex_div_wide): Update.
9897 * tree-eh.c (lower_resx): Update.
9898 (unsplit_eh): Update.
9899 (cleanup_empty_eh_move_lp): Update.
9900 * tree-inline.c (copy_edges_for_bb): Update.
9901 (freqs_to_counts): Update.
9902 (copy_cfg_body): Update.
9903 * tree-ssa-dce.c (remove_dead_stmt): Update.
9904 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9905 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9906 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9907 (unloop_loops): Update.
9908 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9909 * tree-ssa-loop-split.c (connect_loops): Update.
9910 (split_loop): Update.
9911 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9912 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9913 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9914 * tree-ssa-reassoc.c (branch_fixup): Update.
9915 * tree-ssa-tail-merge.c (replace_block_by): Update.
9916 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9917 (compute_path_counts): Update.
9918 (update_profile): Update.
9919 (recompute_probabilities): Update.
9920 (update_joiner_offpath_counts): Update.
9921 (estimated_freqs_path): Update.
9922 (freqs_to_counts_path): Update.
9923 (clear_counts_path): Update.
9924 (ssa_fix_duplicate_block_edges): Update.
9925 (duplicate_thread_path): Update.
9926 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9927 (case_bit_test_cmp): Update.
9928 (collect_switch_conv_info): Update.
9929 (gen_inbound_check): Update.
9930 (do_jump_if_equal): Update.
9931 (emit_cmp_and_jump_insns): Update.
9932 * tree-tailcall.c (decrease_profile): Update.
9933 (eliminate_tail_call): Update.
9934 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9935 (vect_do_peeling): Update.
9936 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9937 * ubsan.c (ubsan_expand_null_ifn): Update.
9938 (ubsan_expand_ptr_ifn): Update.
9939 * value-prof.c (gimple_divmod_fixed_value): Update.
9940 (gimple_mod_pow2): Update.
9941 (gimple_mod_subtract): Update.
9942 (gimple_ic): Update.
9943 (gimple_stringop_fixed_value): Update.
9945 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9948 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9950 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9952 PR rtl-optimization/82395
9953 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9954 based on non_spilled_static_chain_regno_p.
9956 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9958 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9959 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9960 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9962 2017-10-19 Martin Sebor <msebor@redhat.com>
9964 PR tree-optimization/82596
9965 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9967 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9969 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9970 (get_mem_refs_of_builtin_call): Likewise.
9971 * builtins.c (expand_builtin_apply): Adjust call to
9972 allocate_dynamic_stack_space.
9973 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9974 the third argument to allocate_dynamic_stack_space, otherwise -1.
9975 (expand_builtin): Deal with all alloca variants.
9976 (is_inexpensive_builtin): Likewise.
9977 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9978 * calls.c (special_function_p): Deal with all alloca variants.
9979 (initialize_argument_information): Adjust call to
9980 allocate_dynamic_stack_space.
9981 (expand_call): Likewise.
9982 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9983 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9984 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9985 use it for the stack usage computation.
9986 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9987 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9988 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9989 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9990 (in_loop_p): Remove first argument and useless check.
9991 (pass_walloca::execute): Remove useless test and adjust call to above.
9992 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9993 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9994 (gimplify_call_expr): Deal with all alloca variants.
9995 * hsa-gen.c (gen_hsa_alloca): Likewise.
9996 (gen_hsa_insns_for_call): Likewise.
9997 * ipa-pure-const.c (special_builtin_state): Likewise.
9998 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9999 * tree-object-size.c (alloc_object_size): Likewise.
10000 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10001 (call_may_clobber_ref_p_1): Likewise.
10002 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10003 (ccp_fold_stmt): Likewise.
10004 (optimize_stack_restore): Likewise.
10005 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10006 (mark_all_reaching_defs_necessary_1): Likewise.
10007 (propagate_necessity): Likewise.
10008 (eliminate_unnecessary_stmts): Likewise.
10009 * tree.c (build_common_builtin_nodes): Build
10010 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10011 (build_alloca_call_expr): New function.
10012 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10013 (CASE_BUILT_IN_ALLOCA): Likewise.
10014 (build_alloca_call_expr): Declare.
10015 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10017 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10020 * dwarf2out.c (new_die_raw): New static inline function.
10021 (new_die): Use it to create the DIE.
10022 (add_AT_external_die_ref): Likewise.
10023 (clone_die): Likewise.
10024 (clone_as_declaration): Likewise.
10025 (dwarf2out_vms_debug_main_pointer): Likewise.
10026 (base_type_die): Likewise. Remove early return for corner cases.
10027 Do not call add_pubtype on the DIE here.
10028 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10029 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10030 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10031 native order exists for base types, attach the DIE manually and call
10032 add_pubtype on it. Do not equate a reverse order DIE to the type.
10034 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10036 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10037 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10039 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10041 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10043 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10044 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10045 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10046 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10047 * opts.c (sanitizer_opts): Add builtin.
10048 * ubsan.c (instrument_builtin): New function.
10049 (pass_ubsan::execute): Call it.
10050 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10051 * doc/invoke.texi: Document -fsanitize=builtin.
10053 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10054 builtins, store max (log2 (align), 0) into uchar field instead of
10055 align into uptr field.
10056 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10057 store uchar 0 field instead of uptr 0 field.
10058 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10059 instead of passing one address of struct with 2 locations pass
10060 two addresses of structs with 1 location each.
10061 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10062 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10063 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10064 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10065 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10066 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10067 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10068 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10070 2017-10-19 Martin Liska <mliska@suse.cz>
10073 * file-find.c (remove_prefix): Remove.
10074 * file-find.h (remove_prefix): Likewise.
10075 * gcc-ar.c: Remove smartness of lookup.
10077 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10079 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10080 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10081 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10083 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10086 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10087 (setcc + and to xor + setcc): New peephole2.
10089 2017-10-19 Tom de Vries <tom@codesourcery.com>
10091 * doc/sourcebuild.texi (Test Directives, Variants of
10092 dg-require-support): Add dg-require-stack-size.
10094 2017-10-19 Martin Liska <mliska@suse.cz>
10097 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10098 that have a large alignment.
10099 (gimplify_target_expr): Likewise.
10101 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10103 PR rtl-optimization/82602
10104 * ira.c (rtx_moveable_p): Return false for volatile asm.
10106 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10109 * config/i386/i386-modes.def (CCGZ): New CC mode.
10110 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10111 * config/i386/predicates.md (ix86_comparison_operator):
10113 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10114 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10115 with double-word subtraction.
10116 (put_condition_code): Handle CCGZmode.
10118 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10120 * wide-int.cc (debug (const wide_int &)): New.
10121 (debug (const wide_int *)): New.
10122 (debug (const widest_int &)): New.
10123 (debug (const widest_int *)): New.
10125 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10127 PR middle-end/82556
10128 * lra-constraints.c (curr_insn_transform): Use non-input operand
10129 instead of output one for matched reload.
10131 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10133 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10134 (tree-ssa-loop-ivopts.h): New header file.
10135 (struct builtin_info): New fields.
10136 (classify_builtin_1): Compute and record base and offset parts for
10137 memset builtin partition by calling strip_offset.
10138 (offset_cmp, fuse_memset_builtins): New functions.
10139 (finalize_partitions): Fuse adjacent memset partitions by calling
10141 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10142 Expose the interface.
10143 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10145 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10147 PR tree-optimization/82574
10148 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10149 that data reference must be executed exactly once per iteration
10150 against the outermost loop in nest.
10151 (classify_partition): Update call to above function.
10153 2017-10-18 Richard Biener <rguenther@suse.de>
10155 PR tree-optimization/82591
10156 * graphite.c (graphite_transform_loops): Move code gen message
10158 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10159 Here. Handle scop_to_isl_ast failing.
10160 (scop_to_isl_ast): Limit the number of ISL operations.
10162 2017-10-18 Richard Biener <rguenther@suse.de>
10164 * graphite-isl-ast-to-gimple.c
10165 (translate_isl_ast_to_gimple::set_rename): Simplify.
10166 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10167 (graphite_copy_stmts_from_block): ... here.
10168 (copy_bb_and_scalar_dependences): Simplify.
10169 (add_parameters_to_ivs_params): Canonicalize.
10170 (generate_entry_out_of_ssa_copies): Simplify.
10171 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10172 by passing in ISL dimension.
10173 (parameter_index_in_region_1): Rename to ...
10174 (parameter_index_in_region): ... this.
10175 (extract_affine): Adjust assert, pass down parameter index.
10176 (add_param_constraints): Use range-info when available.
10177 (build_scop_context): Adjust.
10178 * sese.c (new_sese_info): Adjust.
10179 (free_sese_info): Likewise.
10180 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10181 Remove unused typedefs.
10182 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10184 2017-10-18 Martin Liska <mliska@suse.cz>
10186 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10188 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10190 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10191 (s390_sched_init): Do not reset s390_sched_state if we entered the
10192 current basic block via a fallthru edge and all others are unlikely.
10194 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10196 * config/s390/s390.c (NUM_SIDES): New variable.
10197 (LONGRUNNING_THRESHOLD): New variable.
10198 (LATENCY_FACTOR): New variable.
10199 (s390_sched_score): Decrease score for long-running instructions on
10201 (s390_sched_variable_issue): Perform bookkeeping for long-running
10204 2017-10-18 Richard Biener <rguenther@suse.de>
10206 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10207 Simplify with removal of the parameter rename map.
10208 (set_rename): Likewise.
10209 (should_copy_to_new_region): Likewise.
10210 (graphite_copy_stmts_from_block): Likewise.
10211 (copy_bb_and_scalar_dependences): Remove initialization of
10212 unused copied_bb_map.
10213 (copy_def): Remove.
10214 (copy_internal_parameters): Likewise.
10215 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10216 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10217 Use INTEGRAL_TYPE_P.
10218 (parameter_index_in_region_1): Rename to ...
10219 (assign_parameter_index_in_region): ... this. Assert we have
10220 a parameter we handle.
10221 (scan_tree_for_params): Adjust.
10222 * sese.h (parameter_rename_map_t): Remove.
10223 (struct sese_info_t): Remove unused parameter_rename_map and
10224 copied_bb_map members.
10225 * sese.c (new_sese_info): Adjust.
10226 (free_sese_info): Likewise.
10228 2017-10-18 Martin Liska <mliska@suse.cz>
10231 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10232 on an abnormal edge.
10234 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10236 * doc/invoke.texi (ffunction-sections and fdata-sections):
10239 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10241 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10242 the use statement can throw internally.
10244 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10246 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10247 any RTX present on the RHS of a SET.
10248 * compare-elim.c (try_eliminate_compare): Restore comment.
10250 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10252 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10253 may be also called on tcc_type nodes.
10254 * langhooks.c (lhd_tree_size): Likewise.
10256 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10258 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10259 format_warning_at_substring.
10260 (maybe_warn): Convert source_range * param to a location_t. Pass
10261 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10262 (format_directive): Remove code to extract source_ranges and
10263 source_range * in favor of just a location_t.
10264 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10266 * substring-locations.c (format_warning_va): Convert
10267 source_range * param to a location_t.
10268 (format_warning_at_substring): Likewise.
10269 * substring-locations.h (format_warning_va): Likewise.
10270 (format_warning_at_substring): Likewise.
10272 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10274 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10276 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10277 and scatter/gather ops.
10279 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10280 vec_gather_load and vec_scatter_store.
10281 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10282 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10283 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10284 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10285 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10286 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10288 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10290 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10291 Detect FTST instruction and handle its register pops. Only pop
10292 second operand if can_pop_second_op is true.
10293 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10294 set can_pop_second_op to false in the compare_for_stack_reg call.
10296 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10297 output_fp_compare for stack register operands.
10298 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10299 instructions here. Do not emit stack register pops here. Assert
10300 that FCOMPP pops next to top stack register. Rewrite function.
10302 2017-10-17 Nathan Sidwell <nathan@acm.org>
10304 PR middle-end/82577
10305 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10306 use DECL_ASSEMBLER_NAME_RAW.
10308 PR middle-end/82546
10309 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10312 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10313 Wilco Dijkstra <wilco.dijkstra@arm.com>
10315 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10316 converstion to Pmode from the buf_addr.
10318 2017-10-17 Richard Biener <rguenther@suse.de>
10320 * graphite-dependences.c (scop_get_reads_and_writes): Change
10321 output parameters to references.
10323 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10325 PR 71026/tree-optimization
10326 * fold-const.c (distribute_real_division): Removed.
10327 (fold_binary_loc): Remove calls to distribute_real_divison.
10329 2017-10-17 Richard Biener <rguenther@suse.de>
10331 * graphite-scop-detection.c
10332 (scop_detection::stmt_has_simple_data_refs_p): Always use
10333 the full nest as region.
10334 (try_generate_gimple_bb): Likewise.
10335 * sese.c (scalar_evolution_in_region): Simplify now that
10336 SCEV can handle instantiation in regions.
10337 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10338 in the non-loop part of a function if requested.
10340 2017-10-17 Richard Biener <rguenther@suse.de>
10342 PR tree-optimization/82563
10343 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10345 (graphite_regenerate_ast_isl): Call it.
10346 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10348 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10350 PR tree-optimization/82549
10351 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10352 Formatting fixes. Instead of calling make_bit_field_ref with negative
10355 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10357 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10358 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10359 _mm_maskz_reduce_ss): New.
10360 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10361 __builtin_ia32_reducess_mask): Ditto..
10362 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10363 * config/i386/sse.md (reduces<mode>): Renamed to ...
10364 (reduces<mode><mask_scalar_name>): ... this.
10365 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10367 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10368 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10370 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10372 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10373 * selftest-run-tests.c (selftest::run_tests): Call
10374 selftest::unique_ptr_tests_cc_tests.
10375 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10376 * unique-ptr-tests.cc: New file.
10378 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10381 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10383 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10384 (make_hard_regno_born, make_hard_regno_dead): Update
10385 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10387 2017-10-16 Jeff Law <law@redhat.com>
10389 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10391 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10393 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10395 2017-10-16 Olivier Hainque <hainque@adacore.com>
10397 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10398 with_cpu if we were configured for an e500v2 target cpu name.
10400 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10402 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10403 * doc/invoke.texi: Document +nodsp as a valid extension for
10406 2017-10-16 Martin Liska <mliska@suse.cz>
10408 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10409 (test_set_range): Likewise.
10410 (test_range_functions): Rename to ...
10411 (test_bit_in_range): ... this.
10412 (sbitmap_c_tests): Add new test.
10414 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10416 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10418 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10419 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10421 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10423 * config/aarch64/aarch64-builtins.c
10424 (aarch64_types_quadopu_lane_qualifiers): New.
10425 (TYPES_QUADOPU_LANE): New.
10426 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10427 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10428 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10429 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10430 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10431 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10432 (Vdottype, DOTPROD): New.
10433 (sur): Add SDOT and UDOT.
10435 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10437 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10438 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10439 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10440 Add TARGET_DOTPROD.
10441 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10442 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10443 Enable TARGET_DOTPROD.
10444 (cortex-a75.cortex-a55): Likewise.
10445 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10447 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10449 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10450 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10452 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10454 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10455 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10456 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10457 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10458 * config/arm/types.md (neon_dot, neon_dot_q): New.
10459 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10461 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10463 * config/arm/arm.h (TARGET_DOTPROD): New.
10464 * config/arm/arm.c (arm_arch_dotprod): New.
10465 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10466 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10467 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10468 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10469 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10470 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10471 * doc/invoke.texi (armv8.2-a): Document dotprod
10473 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10475 * i386.c (ix86_vec_cost): New function.
10476 (ix86_rtx_costs): Handle vector operations better.
10477 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10478 * x86-tune-costs.h: Add new costs to all tables.
10480 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10482 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10484 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10485 divsd, sqrtss and sqrtsd
10486 * x86-tune-costs.h: Add new entries to all costs.
10487 (znver1_cost): Fix to match real instruction latencies.
10489 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10490 Michael Collison <michael.collison@arm.com>
10492 * compare-elim.c: Include emit-rtl.h.
10493 (can_merge_compare_into_arith): New function.
10494 (try_validate_parallel): Likewise.
10495 (try_merge_compare): Likewise.
10496 (try_eliminate_compare): Call the above when no previous clobber
10498 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10501 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10503 PR middle-end/62263
10504 PR middle-end/82498
10505 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10506 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10508 PR middle-end/62263
10509 PR middle-end/82498
10510 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10511 to be any operand_equal_p operands. For & (B - 1) require
10512 B to be power of 2. Recognize
10513 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10515 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10518 * optabs.c (expand_memory_blockage): Fix call of
10519 targetm.have_memory_blockage.
10521 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10524 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10525 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10526 objects to extra_objs instead of overwriting it.
10528 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10530 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10531 Use any_fp_register_operand as operand[3] predicate. Simplify
10532 equality test for operands[2] and operands[4] memory location.
10533 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10534 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10535 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10536 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10537 any_fp_register_operand as operand[1] predicate. Simplify
10538 equality test for operands[0] and operands[3] memory location.
10539 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10540 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10541 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10543 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10545 * target-insns.def: Add memory_blockage.
10546 * optabs.c (expand_memory_blockage): New function.
10547 (expand_asm_memory_barrier): Rename ...
10548 (expand_asm_memory_blockage): ... to this.
10549 (expand_mem_thread_fence): Call expand_memory_blockage
10550 instead of expand_asm_memory_barrier.
10551 (expand_mem_singnal_fence): Ditto.
10552 (expand_atomic_load): Ditto.
10553 (expand_atomic_store): Ditto.
10554 * doc/md.texi (Standard Pattern Names For Generation):
10555 Document memory_blockage instruction pattern.
10557 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10559 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10560 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10561 * config/rl78/rl78.md: New define_expand "adddi3".
10563 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10565 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10568 2017-10-13 Jeff Law <law@redhat.com>
10570 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10572 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10575 * internal-fn.c (expand_mul_overflow): If both operands have
10576 the same highpart of -1 or 0 and the topmost bit of lowpart
10577 is different, overflow is if res <= 0 rather than res < 0.
10579 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10581 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10582 TARGET_P9_VECTOR code for unaligned_load case.
10584 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10586 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10588 2017-10-13 Nathan Sidwell <nathan@acm.org>
10590 * tree-core.h (tree_contains_struct): Make bool.
10591 * tree.c (tree_contains_struct): Likewise.
10592 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10593 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10594 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10595 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10596 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10598 2017-10-13 Richard Biener <rguenther@suse.de>
10600 * graphite-isl-ast-to-gimple.c
10601 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10602 parameters and dominance check.
10603 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10604 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10605 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10606 Do not update SSA form here or do intermediate IL verification.
10607 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10608 (graphite_initialize): Remove check on the number of loops in
10609 the function and inline into graphite_transform_loops.
10610 (graphite_finalize): Inline into graphite_transform_loops.
10611 (graphite_transform_loops): Perform SSA update and IL verification
10613 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10615 2017-10-13 Richard Biener <rguenther@suse.de>
10617 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10618 graphite_expression_type_precision): Avoid global constructor
10620 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10621 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10622 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10623 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10624 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10625 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10627 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10630 * config/i386/i386.h (ix86_red_zone_size): New.
10631 * config/i386/i386.md (push peephole2s): Replace
10632 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10634 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10635 Alan Hayward <alan.hayward@arm.com>
10636 David Sherwood <david.sherwood@arm.com>
10638 * combine.c (can_change_dest_mode): Reject changes in
10639 REGMODE_NATURAL_SIZE.
10641 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10642 Alan Hayward <alan.hayward@arm.com>
10643 David Sherwood <david.sherwood@arm.com>
10645 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10646 (expand_debug_source_expr): Likewise.
10647 * combine.c (combine_simplify_rtx): Likewise.
10648 * cse.c (fold_rtx): Likewise.
10649 * fwprop.c (canonicalize_address): Likewise.
10650 * targhooks.c (default_shift_truncation_mask): Likewise.
10652 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10653 Alan Hayward <alan.hayward@arm.com>
10654 David Sherwood <david.sherwood@arm.com>
10656 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10657 (widened_mode): Likewise.
10658 (expand_unop): Likewise.
10659 * ree.c (transform_ifelse): Likewise.
10660 (merge_def_and_ext): Likewise.
10661 (combine_reaching_defs): Likewise.
10662 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10664 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10665 Alan Hayward <alan.hayward@arm.com>
10666 David Sherwood <david.sherwood@arm.com>
10668 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10669 * combine.c (gen_lowpart_for_combine): Likewise.
10670 * dwarf2out.c (rtl_for_decl_location): Likewise.
10671 * final.c (alter_subreg): Likewise.
10672 * rtlhooks.c (gen_lowpart_general): Likewise.
10673 (gen_lowpart_if_possible): Likewise.
10675 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10676 Alan Hayward <alan.hayward@arm.com>
10677 David Sherwood <david.sherwood@arm.com>
10679 * calls.c (expand_call): Use subreg_lowpart_offset.
10680 * cse.c (cse_insn): Likewise.
10681 * regcprop.c (copy_value): Likewise.
10682 (copyprop_hardreg_forward_1): Likewise.
10684 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10687 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10688 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10689 =Q constraints to +Q and into insn condition add check
10690 that operands[0] and operands[1] are equal.
10691 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10692 =Q constraints to +Q and into insn condition add check
10693 that operands[0] is equal to either operands[1] or operands[2].
10696 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10697 instead of handling MINUS_EXPR twice (once for each argument),
10698 canonicalize operand order and handle just once, use rtype where
10699 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10702 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10703 any values of __C while still being pattern recognizable as a simple
10704 rotate instruction.
10706 2017-10-13 Richard Biener <rguenther@suse.de>
10708 PR tree-optimization/82451
10710 2017-10-02 Richard Biener <rguenther@suse.de>
10712 PR tree-optimization/82355
10713 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10714 a mapping for the enclosing loop but avoid generating one for
10715 the loop tree root.
10716 (copy_bb_and_scalar_dependences): Remove premature codegen
10717 error on PHIs in blocks duplicated into multiple places.
10718 * graphite-scop-detection.c
10719 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10720 in the region use it as loop and nest to analyze the DR in.
10721 (try_generate_gimple_bb): Likewise.
10722 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10723 (add_loop_constraints): For blocks in a loop not in the region
10724 create a dimension with a single iteration.
10725 * sese.h (gbb_loop_at_index): Remove assert.
10727 * cfgloop.c (loop_preheader_edge): For the loop tree root
10728 return the single successor of the entry block.
10729 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10730 Reset the SCEV hashtable and niters.
10731 * graphite-scop-detection.c
10732 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10733 assert that we only have POLYNOMIAL_CHREC that vary in loops
10734 contained in the region.
10735 (scop_detection::graphite_can_represent_expr): Adjust.
10736 (scop_detection::stmt_has_simple_data_refs_p): For loops
10737 not in the region set loop to NULL. The nest is now the
10738 entry edge to the region.
10739 (try_generate_gimple_bb): Likewise.
10740 * sese.c (scalar_evolution_in_region): Adjust for
10741 instantiate_scev change.
10742 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10743 Make nest parameter the edge into the region.
10744 (create_data_ref): Likewise.
10745 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10746 entry edge into a region and adjust instantiate_scev calls.
10747 (create_data_ref): Likewise.
10748 (graphite_find_data_references_in_stmt): Likewise.
10749 (find_data_references_in_stmt): Pass the loop preheader edge
10750 from the nest argument.
10751 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10752 parameter the edge into the region.
10753 (instantiate_parameters): Use the loop preheader edge as entry.
10754 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10756 (get_instantiated_value_entry): Make instantiate_below parameter
10757 the edge into the region.
10758 (instantiate_scev_name): Likewise. Adjust dominance checks,
10759 when we cannot use loop-based instantiation instantiate by
10760 walking use-def chains.
10761 (instantiate_scev_poly): Adjust.
10762 (instantiate_scev_binary): Likewise.
10763 (instantiate_scev_convert): Likewise.
10764 (instantiate_scev_not): Likewise.
10765 (instantiate_array_ref): Remove.
10766 (instantiate_scev_3): Likewise.
10767 (instantiate_scev_2): Likewise.
10768 (instantiate_scev_1): Likewise.
10769 (instantiate_scev_r): Do not blindly handle N-operand trees.
10770 Do not instantiate array-refs. Handle all constants and invariants.
10771 (instantiate_scev): Make instantiate_below parameter
10772 the edge into the region.
10773 (resolve_mixers): Use the loop preheader edge for the region
10774 parameter to instantiate_scev_r.
10775 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10777 2017-10-13 Richard Biener <rguenther@suse.de>
10779 PR tree-optimization/82525
10780 * graphite-isl-ast-to-gimple.c
10781 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10783 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10784 Fail code generation when we cannot represent the isl integer.
10785 (binary_op_to_tree): Elide modulo operations that are no-ops
10786 in the type we code generate. Remove now superfluous code
10789 2017-10-13 Richard Biener <rguenther@suse.de>
10791 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10792 (scop_detection::harmful_loop_in_region): Remove premature
10793 IV type restriction.
10794 (scop_detection::graphite_can_represent_scev): We can handle
10795 pointer IVs just fine.
10797 2017-10-13 Alan Modra <amodra@gmail.com>
10799 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10800 "Clobbers and Scratch Registers". Add paragraph on
10801 alternative to clobbers for scratch registers and OpenBLAS
10804 2017-10-13 Alan Modra <amodra@gmail.com>
10806 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10807 example of a memory input for a string of known length. Move
10808 commentary out of table. Add a number of new examples
10809 covering array memory inputs.
10811 2017-10-12 Martin Liska <mliska@suse.cz>
10813 PR tree-optimization/82493
10814 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10815 (test_range_functions): New function.
10816 (sbitmap_c_tests): Likewise.
10817 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10818 * selftest.h (sbitmap_c_tests): New function.
10820 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10822 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10824 * config/rs6000/amo.h: Fix spacing issue.
10826 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10829 * config/i386/i386.md (*ashl<mode>3_mask_1,
10830 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10831 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10834 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10836 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10837 (profile_probability): Set max_probability
10838 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10840 * profile-count.c (profile_probability::differs_from_p): Do not
10841 rely on max_probaiblity == 10000
10843 2017-10-12 Jeff Law <law@redhat.com>
10845 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10848 2017-10-12 Martin Sebor <msebor@redhat.com>
10852 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10853 (handle_alias_pairs): Call it.
10854 * common.opt (-Wattribute-alias): New option.
10855 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10856 * doc/invoke.texi (-Wattribute-alias): Document.
10858 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10861 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10863 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10865 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10866 (make_hard_regno_born, make_hard_regno_dead): Update
10867 bb_killed_pseudos and bb_gen_pseudos.
10869 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10871 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10873 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10875 * config/alpha/alpha.c (alpha_split_conditional_move):
10876 Use std::swap instead of manually swapping.
10877 (alpha_stdarg_optimize_hook): Ditto.
10878 (alpha_canonicalize_comparison): Ditto.
10880 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10882 * tree-loop-distribution.c (struct builtin_info): New struct.
10883 (struct partition): Refactor fields into struct builtin_info.
10884 (partition_free): Free struct builtin_info.
10885 (build_size_arg_loc, build_addr_arg_loc): Delete.
10886 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10887 information from struct builtin_info.
10888 (find_single_drs): New function refactored from classify_partition.
10889 Also moved builtin validity checks to this function.
10890 (compute_access_range, alloc_builtin): New functions.
10891 (classify_builtin_st, classify_builtin_ldst): New functions.
10892 (classify_partition): Refactor code into functions find_single_drs,
10893 classify_builtin_st and classify_builtin_ldst.
10894 (distribute_loop): Don't do runtime alias check when distributing
10896 (find_seed_stmts_for_distribution): New function.
10897 (pass_loop_distribution::execute): Refactor code finding seed
10898 stmts into above function. Support distribution for the innermost
10899 two-level loop nest. Adjust dump information.
10901 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10903 * tree-loop-distribution.c: Adjust the general comment.
10904 (NUM_PARTITION_THRESHOLD): New macro.
10905 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10906 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10907 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10908 in call to build_partition_graph.
10909 (finalize_partitions): New parameter. Make loop distribution more
10910 conservative by fusing more partitions.
10911 (distribute_loop): Don't do runtime alias check in case of loop nest
10913 (find_seed_stmts_for_distribution): New function.
10914 (prepare_perfect_loop_nest): New function.
10915 (pass_loop_distribution::execute): Refactor code finding seed stmts
10916 and loop nest into above functions. Support loop nest distribution.
10917 Adjust dump information accordingly.
10919 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10921 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10922 and set PTYPE_SEQUENTIAL for merged partition.
10924 2017-10-12 Richard Biener <rguenther@suse.de>
10926 PR tree-optimization/69728
10928 2017-09-19 Richard Biener <rguenther@suse.de>
10930 PR tree-optimization/69728
10931 * graphite-sese-to-poly.c (schedule_error): New global.
10932 (add_loop_schedule): Handle empty domain by failing the
10934 (build_original_schedule): Handle schedule_error.
10936 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10937 domain by returning an unchanged schedule.
10939 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10941 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10942 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10944 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10946 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10949 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10952 * expr.c (store_field): Don't optimize away bitsize == 0 store
10953 from CALL_EXPR with addressable return type.
10955 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10957 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10958 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10959 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10960 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10961 TARGET_ISEL instead of TARGET_ISEL<sel>.
10963 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10965 * config/rs6000/rs6000.c
10966 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10968 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10970 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10972 (reg_or_cint_operand): Fix comment.
10973 (reg_or_zero_operand): New predicate.
10974 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10975 * config/rs6000/rs6000.c (output_isel): Delete.
10976 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10977 instead of reg_or_cint_operand. Output instruction directly (not via
10979 (isel_unsigned_<mode>): Ditto.
10980 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10981 gpc_reg_operand. Add an instruction alternative for this. Output
10982 instruction directly.
10983 (*isel_reversed_unsigned_<mode>): Ditto.
10985 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10987 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10988 (TARGET_CANONICALIZE_COMPARISON): Define.
10990 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10993 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10994 Check whether the dest is REG before adding REG_EQUIV note.
10996 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10999 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11001 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11002 (make_hard_regno_born, make_hard_regno_dead): Update
11003 bb_killed_pseudos and bb_gen_pseudos.
11005 2017-10-11 Nathan Sidwell <nathan@acm.org>
11007 * incpath.h (enum incpath_kind): Name enum, prefix values.
11008 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11009 * incpath.c (heads, tails): Use INC_MAX.
11010 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11011 (merge_include_chains, split_quote_chain,
11012 register_include_chains): Update incpath_kind names.
11013 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11014 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11016 (add_framework_path, darwin_register_objc_includes): Likewise.
11017 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11019 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11021 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11022 Do not use float_operator operator predicate.
11023 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11024 * config/i386/predicates.md (float_operator): Remove predicate.
11026 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11028 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11029 (*jccxf_i387): Ditto.
11030 (*jcc<mode>_i387): Ditto.
11031 (*jccu<mode>_i387): Ditto.
11032 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11033 (*jcc_*_i387 splitters): Remove.
11034 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11035 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11036 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11039 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11041 * profile-count.h (slow_safe_scale_64bit): New function.
11042 (safe_scale_64bit): New inline.
11043 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11044 * profile-count.c: Include wide-int.h
11045 (slow_safe_scale_64bit): New.
11047 2017-10-11 Nathan Sidwell <nathan@acm.org>
11049 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11050 HAS_DECL_ASSEMBLER_NAME_P.
11051 * gimple-expr.c (gimple_decl_printable_name: Check
11052 HAS_DECL_ASSEMBLER_NAME_P too.
11053 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11054 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11055 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11056 * passes.c (rest_of_decl_compilation): Check
11057 HAS_DECL_ASSEMBLER_NAME_P too.
11058 * recog.c (verify_changes): Likewise.
11059 * tree-pretty-print.c (dump_decl_name): Likewise.
11060 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11062 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11063 (SET_DECL_ASSEMBLER_NAME): Use it.
11064 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11065 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11066 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11068 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11070 * config.gcc (i386, x86_64): Add extra objects.
11071 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11072 (ix86_min_insn_size): Declare.
11073 (ix86_issue_rate): Declare.
11074 (ix86_adjust_cost): Declare.
11075 (ia32_multipass_dfa_lookahead): Declare.
11076 (ix86_macro_fusion_p): Declare.
11077 (ix86_macro_fusion_pair_p): Declare.
11078 (ix86_bd_has_dispatch): Declare.
11079 (ix86_bd_do_dispatch): Declare.
11080 (ix86_core2i7_init_hooks): Declare.
11081 (ix86_atom_sched_reorder): Declare.
11082 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11083 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11084 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11085 (rip_relative_addr_p): Rename to ...
11086 (ix86_rip_relative_addr_p): ... this one; export.
11087 (memory_address_length): Update.
11088 (ix86_issue_rate): Move to x86-tune-sched.c.
11089 (ix86_flags_dependent): Move to x86-tune-sched.c.
11090 (ix86_agi_dependent): Move to x86-tune-sched.c.
11091 (exact_dependency_1): Move to x86-tune-sched.c.
11092 (exact_store_load_dependency): Move to x86-tune-sched.c.
11093 (ix86_adjust_cost): Move to x86-tune-sched.c.
11094 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11095 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11096 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11097 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11098 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11099 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11100 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11101 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11102 (min_insn_size): Rename to ...
11103 (ix86_min_insn_size): ... this one; export.
11104 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11105 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11106 (core2i7_first_cycle_multipass_backtrack): Move to
11107 x86-tune-sched-core.c.
11108 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11109 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11110 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11111 (ix86_avoid_jump_mispredicts): Update.
11112 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11113 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11114 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11115 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11116 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11117 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11118 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11119 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11120 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11121 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11122 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11123 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11124 (BIG): Move to ix86-tune-sched-bd.c.
11125 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11126 (enum insn_path): Move to ix86-tune-sched-bd.c.
11127 (get_mem_group): Move to ix86-tune-sched-bd.c.
11128 (is_cmp): Move to ix86-tune-sched-bd.c.
11129 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11130 (is_branch): Move to ix86-tune-sched-bd.c.
11131 (is_prefetch): Move to ix86-tune-sched-bd.c.
11132 (init_window): Move to ix86-tune-sched-bd.c.
11133 (allocate_window): Move to ix86-tune-sched-bd.c.
11134 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11135 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11136 (process_end_window): Move to ix86-tune-sched-bd.c.
11137 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11138 (find_constant): Move to ix86-tune-sched-bd.c.
11139 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11140 (has_immediate): Move to ix86-tune-sched-bd.c.
11141 (get_insn_path): Move to ix86-tune-sched-bd.c.
11142 (get_insn_group): Move to ix86-tune-sched-bd.c.
11143 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11144 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11145 (add_insn_window): Move to ix86-tune-sched-bd.c.
11146 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11147 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11148 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11149 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11150 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11151 (do_dispatch): Move to ix86-tune-sched-bd.c.
11152 (has_dispatch): Move to ix86-tune-sched-bd.c.
11153 * config/i386/t-i386: Add new object files.
11154 * config/i386/x86-tune-costs.h: New file.
11155 * config/i386/x86-tune-sched-atom.c: New file.
11156 * config/i386/x86-tune-sched-bd.c: New file.
11157 * config/i386/x86-tune-sched-core.c: New file.
11158 * config/i386/x86-tune-sched.c: New file.
11160 2017-10-11 Liu Hao <lh_mouse@126.com>
11162 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11163 the generic version below instead.
11164 (should_colorize): Recognize Windows consoles as terminals
11166 * pretty-print.c [__MINGW32__] (write_all): New function.
11167 [__MINGW32__] (find_esc_head): Likewise.
11168 [__MINGW32__] (find_esc_terminator): Likewise.
11169 [__MINGW32__] (eat_esc_sequence): Likewise.
11170 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11172 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11175 2017-10-11 Richard Biener <rguenther@suse.de>
11177 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11178 Properly call analyze_scalar_evolution with the loop of the stmt.
11180 2017-10-11 Richard Biener <rguenther@suse.de>
11182 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11183 * tree-core.h (tree_base): Add chrec_var union member.
11184 * tree.h (CHREC_VAR): Remove.
11185 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11186 * tree-chrec.h (build_polynomial_chrec): Adjust.
11187 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11188 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11190 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11192 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11193 * match.pd: ... here.
11194 ((T) X == (T) Y): Relax condition.
11196 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11198 PR tree-optimization/82472
11199 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11201 (break_alias_scc_partitions): Update postorder number.
11203 2017-10-11 Martin Liska <mliska@suse.cz>
11206 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11208 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11210 2017-10-11 Martin Liska <mliska@suse.cz>
11212 * print-rtl.c (print_insn): Move declaration of idbuf
11213 to same scope as name.
11215 2017-10-11 Martin Liska <mliska@suse.cz>
11220 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11221 volatile arguments.
11223 2017-10-11 Martin Liska <mliska@suse.cz>
11226 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11227 volatile arguments.
11229 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11231 * config.gcc (default_gnu_indirect_function): Default to yes for
11232 arm*-*-linux* with glibc.
11234 2017-10-11 Richard Biener <rguenther@suse.de>
11236 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11237 default-defs and types we do not want to analyze.
11238 (interpret_loop_phi): Replace unreachable code with an assert.
11239 (compute_scalar_evolution_in_loop): Remove and inline ...
11240 (analyze_scalar_evolution_1): ... here, replacing condition with
11241 what makes the intent clearer. Remove handling of cases
11242 get_scalar_evolution now handles.
11244 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11246 PR rtl-optimization/81434
11247 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11248 comment for main loop. In sched_group_found if, also add checks for
11249 pass and min_cost_group.
11251 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11253 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11254 (rs6000_insn_cost): New function.
11255 * config/rs6000/rs6000.md (cost): New attribute.
11257 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11258 H.J. Lu <hongjiu.lu@intel.com>
11262 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11263 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11264 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11265 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11266 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11267 OPTION_MASK_ISA_MMX and some other bit set require both
11268 mmx and the ISAs without the mmx bit.
11269 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11270 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11271 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11272 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11273 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11274 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11275 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11276 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11277 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11278 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11279 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11280 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11281 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11282 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11283 Add OPTION_MASK_ISA_MMX.
11285 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11287 * config.gcc (armv7*-*-freebsd*): New target.
11288 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11290 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11292 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11293 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11296 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11298 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11299 a template parameter.
11300 (WIDE_INT_REF_FOR): Update accordingly.
11301 * tree.h (wi::int_traits <const_tree>): Delete.
11302 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11303 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11304 (wi::tree_to_wide_ref): New typedef.
11305 (wi::to_wide): New function.
11306 * calls.c (get_size_range): Use wi::to_wide when operating on
11307 trees as wide_ints.
11308 * cgraph.c (cgraph_node::create_thunk): Likewise.
11309 * config/i386/i386.c (ix86_data_alignment): Likewise.
11310 (ix86_local_alignment): Likewise.
11311 * dbxout.c (stabstr_O): Likewise.
11312 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11313 * expr.c (const_vector_from_tree): Likewise.
11314 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11315 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11316 (fold_negate_expr_1, int_const_binop_1, const_binop)
11317 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11318 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11319 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11320 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11321 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11322 (fold_not_const, round_up_loc): Likewise.
11323 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11324 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11325 (alloca_call_type): Likewise.
11326 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11327 * godump.c (go_output_typedef): Likewise.
11328 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11329 * internal-fn.c (get_min_precision): Likewise.
11330 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11331 * ipa-polymorphic-call.c
11332 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11333 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11334 (ipa_modify_call_arguments): Likewise.
11335 * match.pd: Likewise.
11336 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11337 * print-tree.c (print_node_brief, print_node): Likewise.
11338 * stmt.c (expand_case): Likewise.
11339 * stor-layout.c (layout_type): Likewise.
11340 * tree-affine.c (tree_to_aff_combination): Likewise.
11341 * tree-cfg.c (group_case_labels_stmt): Likewise.
11342 * tree-data-ref.c (dr_analyze_indices): Likewise.
11343 (prune_runtime_alias_test_list): Likewise.
11344 * tree-dump.c (dequeue_and_dump): Likewise.
11345 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11346 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11347 * tree-pretty-print.c (dump_generic_node): Likewise.
11348 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11349 (simple_iv_with_niters): Likewise.
11350 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11351 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11352 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11353 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11354 (refine_value_range_using_guard, number_of_iterations_ne_max)
11355 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11356 (get_cst_init_from_scev, record_nonwrapping_iv)
11357 (scev_var_range_cant_overflow): Likewise.
11358 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11359 * tree-ssa-pre.c (compute_avail): Likewise.
11360 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11361 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11362 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11363 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11364 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11365 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11366 (dump_case_nodes, try_switch_expansion): Likewise.
11367 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11368 (vect_do_peeling): Likewise.
11369 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11370 * tree-vect-stmts.c (vectorizable_load): Likewise.
11371 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11372 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11373 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11374 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11375 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11376 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11377 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11378 (evrp_dom_walker::before_dom_children): Likewise.
11379 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11380 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11381 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11382 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11383 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11384 (get_range_pos_neg): Likewise.
11385 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11386 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11387 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11388 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11389 * config/avr/avr.c (avr_fold_builtin): Likewise.
11390 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11391 * config/msp430/msp430.c (msp430_attr): Likewise.
11392 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11393 * config/powerpcspe/powerpcspe-c.c
11394 (altivec_resolve_overloaded_builtin): Likewise.
11395 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11396 (rs6000_expand_ternop_builtin): Likewise.
11397 * config/rs6000/rs6000-c.c
11398 (altivec_resolve_overloaded_builtin): Likewise.
11399 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11400 (rs6000_expand_ternop_builtin): Likewise.
11401 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11403 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11405 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11406 when copying loop nest with only one inner loop.
11408 2017-10-10 Richard Biener <rguenther@suse.de>
11410 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11411 blocks if SCEV is active.
11412 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11414 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11415 (scev_initialize): Assert we are not yet initialized.
11417 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11419 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11420 inner loop's exit stmt by making it always exit the loop, otherwise
11421 we would generate an infinite empty loop.
11423 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11425 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11426 renaming variables in new preheader if it's deleted.
11428 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11430 * tree-loop-distribution.c (struct partition): Remove unused field
11431 loops of the structure.
11432 (partition_alloc, partition_free): Ditto.
11433 (build_rdg_partition_for_vertex): Ditto.
11435 2017-10-09 Jeff Law <law@redhat.com>
11437 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11438 return type to match prototype and documentation.
11440 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11442 * config/rs6000/rs6000.c (processor_costs): Move to ...
11443 * config/rs6000/rs6000.h: ... here.
11444 (rs6000_cost): Declare.
11446 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11448 * except.c (setjmp_fn): New global variable.
11449 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11450 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11451 if DONT_USE_BUILTIN_SETJMP is defined.
11453 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11455 * target.def (insn_cost): New hook.
11456 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11457 * doc/tm.texi: Regenerate.
11458 * rtlanal.c (insn_cost): Use the new hook.
11460 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11462 * combine.c (combine_validate_cost): Compute the new insn_cost,
11463 not just pattern_cost.
11464 (try_combine): Adjust comment.
11466 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11468 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11470 * combine.c (uid_insn_cost): Adjust comment.
11471 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11473 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11474 * dse.c (find_shift_sequence): Ditto.
11475 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11476 (bb_valid_for_noce_process_p): Use pattern_cost.
11477 * rtl.h (insn_rtx_cost): Delete.
11478 (pattern_cost): New prototype.
11479 (insn_cost): New prototype.
11480 * rtlanal.c (insn_rtx_cost): Rename to...
11481 (pattern_cost): ... this.
11484 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11486 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11487 (*jcc<mode>_0_r_i387): Ditto.
11488 (*jccxf_r_i387): Ditto.
11489 (*jcc<mode>_r_i387): Ditto.
11490 (*jccu<mode>_r_i387): Ditto.
11491 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11492 (*jcc): Rename from *jcc_1.
11494 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11496 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11497 deferred rescans after the lvx/stvx recombination pre-pass.
11499 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11501 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11502 memory operation instruction support.
11503 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11504 (rs6000-ibm-aix[789]*): Likewise.
11505 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11506 Document new functions.
11508 2017-10-09 Richard Biener <rguenther@suse.de>
11510 PR tree-optimization/82397
11511 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11512 equality only for semantically equal trees.
11514 2017-10-09 Richard Biener <rguenther@suse.de>
11516 PR tree-optimization/82449
11517 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11518 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11519 allow constant addresses.
11520 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11523 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11525 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11528 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11531 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11534 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11537 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11539 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11542 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11543 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11545 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11547 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11548 (WI_BINARY_PREDICATE_RESULT): Likewise.
11549 (wi::binary_traits::operator_result): New type.
11550 (wi::binary_traits::predicate_result): Likewise.
11551 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11552 (generic_wide_int::operator==, generic_wide_int::operator!=)
11553 (generic_wide_int::operator&, generic_wide_int::and_not)
11554 (generic_wide_int::operator|, generic_wide_int::or_not)
11555 (generic_wide_int::operator^, generic_wide_int::operator+
11556 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11558 (operator~, unary operator-, operator==, operator!=, operator&)
11559 (operator|, operator^, operator+, binary operator-, operator*): New
11561 * expr.c (get_inner_reference): Use wi::bit_and_not.
11562 * fold-const.c (fold_binary_loc): Likewise.
11563 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11564 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11565 (bit_value_binop): Likewise.
11566 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11567 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11568 (extract_range_from_binary_expr_1): Likewise.
11569 (masked_increment): Likewise.
11570 (simplify_bit_ops_using_ranges): Likewise.
11572 2017-10-09 Martin Jambor <mjambor@suse.cz>
11575 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11576 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11577 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11579 (hsa_fixup_mov_insn_type): New function.
11580 (hsa_op_with_type::get_in_type): Use it.
11581 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11582 immediates in an assert.
11583 (hsa_op_with_type::extend_int_to_32bit): New method.
11584 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11585 types. Convert to dest type if necessary.
11586 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11587 (reg_for_gimple_ssa): Pass false as min32int to
11588 hsa_type_for_scalar_tree_type.
11589 (gen_hsa_addr): Fixup type when creating addresable temporary.
11590 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11591 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11592 necessary. Call hsa_fixup_mov_insn_type.
11593 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11594 extend operands and convert to dest type if necessary.
11595 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11596 to dest type if necessary.
11597 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11598 if conversion nt necessary and size matches.
11599 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11600 to dest type if necessary.
11601 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11602 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11604 (gen_hsa_clrsb): Likewise.
11605 (gen_hsa_ffs): Likewise.
11606 (gen_hsa_divmod): Extend operands and convert to dest type if
11608 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11610 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11612 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11613 Remove empty default arguments. Use a brace block as output
11615 (conditional return): Ditto.
11617 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11618 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11620 (group_ending_nop): Ditto.
11621 (doloop_end): Ditto.
11622 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11623 (splitters for those): Ditto.
11625 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11627 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11628 a conditional jump (and the compare for it) so that pc_rtx is the
11630 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11631 for the deleted and renamed ctr<mode>_internal[234] patterns.
11632 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11633 Delete second conditional return pattern.
11634 (ctr<mode>_internal2): Delete this second bdnz pattern.
11635 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11636 (ctr<mode>_internal4): Delete this second bdz pattern.
11638 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11640 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11641 (always_initialized_rtx_for_ssa_name_p): New predicate.
11642 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11643 (finish_out_of_ssa): Free new field of SA.
11644 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11645 * tree-ssa-coalesce.c: Include tree-ssa.h.
11646 (get_parm_default_def_partitions): Remove extern keyword.
11647 (get_undefined_value_partitions): New function.
11648 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11649 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11650 uninitialized bits.
11651 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11653 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11655 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11657 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11659 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11660 for targets that preffer 128bit.
11662 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11664 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11666 2017-10-08 Olivier Hainque <hainque@adacore.com>
11668 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11669 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11671 (thumb_set_return_address): Likewise.
11673 2017-10-08 Olivier Hainque <hainque@adacore.com>
11675 * common/config/arm/arm-common.c (arm_except_unwind_info):
11676 Handle DWARF2_UNWIND_INFO.
11678 2017-10-07 Michael Collison <michael.collison@arm.com>
11680 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11683 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11685 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11686 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11687 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11688 defined, force the creation of a new block for a dispatch label.
11690 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11692 * invoke.texi (Wsuggest-attribute=cold): Document.
11693 * common.opt (Wsuggest-attribute=cold): New
11694 * ipa-pure-const.c (warn_function_cold): New function.
11695 * predict.c (compute_function_frequency): Use it.
11696 * predict.h (warn_function_cold): Declare.
11698 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11700 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11703 2017-10-06 Martin Liska <mliska@suse.cz>
11705 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11706 keyword for member functions.
11707 (struct sanopt_tree_couple): New struct.
11708 (struct sanopt_tree_couple_hash): New function.
11709 (struct sanopt_ctx): Add new hash_map.
11710 (has_dominating_ubsan_ptr_check): New function.
11711 (record_ubsan_ptr_check_stmt): Likewise.
11712 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11713 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11714 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11716 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11719 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11720 aarch64_simd_valid_immediate on CONST_VECTORs.
11721 (aarch64_reg_or_bic_imm): Likewise.
11723 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11725 PR rtl-optimization/82396
11726 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11728 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11730 * graphite-dependences.c (scop_get_reads): Move code to...
11731 (scop_get_must_writes): Move code to...
11732 (scop_get_may_writes): Move code to...
11733 (scop_get_reads_and_writes): ... here.
11734 (scop_get_dependences): Call scop_get_reads_and_writes.
11736 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11738 PR tree-optimization/82434
11739 * fold-const.h (can_native_encode_type_p,
11740 can_native_encode_string_p): Remove.
11741 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11742 don't encode anything, just return what would be otherwise returned.
11743 (native_encode_fixed, native_encode_complex, native_encode_vector):
11745 (native_encode_string): Likewise. Inline by hand
11746 can_native_encode_string_p.
11747 (can_native_encode_type_p): Remove.
11748 (can_native_encode_string_p): Remove.
11749 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11750 STRING_CSTs using can_native_encode_string_p, test all
11751 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11752 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11753 argument from native_encode_expr.
11754 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11755 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11758 2017-10-06 Richard Biener <rguenther@suse.de>
11760 PR tree-optimization/82397
11761 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11762 operand_equal_p but rely on data_ref_compare_tree for detecting
11764 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11765 to match up with dr_group_sort_cmp.
11767 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11770 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11772 * config/s390/s390-builtin-types.def: Regenerate.
11774 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11777 * config/s390/s390-builtin-types.def: Regenerate.
11778 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11779 Change flag from B_VXE to B_VX.
11780 (s390_vec_min_dbl): Remove B_VXE flag.
11782 2017-10-06 Richard Biener <rguenther@suse.de>
11784 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11785 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11786 translate_isl_ast_to_gimple::is_valid_rename,
11787 translate_isl_ast_to_gimple::get_rename,
11788 translate_isl_ast_to_gimple::get_def_bb_for_const,
11789 translate_isl_ast_to_gimple::get_new_name,
11790 translate_isl_ast_to_gimple::collect_all_ssa_names,
11791 translate_isl_ast_to_gimple::copy_loop_phi_args,
11792 translate_isl_ast_to_gimple::collect_all_ssa_names,
11793 translate_isl_ast_to_gimple::copy_loop_phi_args,
11794 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11795 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11796 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11797 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11798 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11799 translate_isl_ast_to_gimple::copy_cond_phi_args,
11800 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11801 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11802 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11803 translate_isl_ast_to_gimple::rename_uses,
11804 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11805 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11806 (set_rename_for_each_def): Likewise.
11807 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11808 here. Handle rewriting SCEV analyzable uses here.
11809 (copy_bb_and_scalar_dependences): Generate code for PHI
11811 (graphite_regenerate_ast_isl): Adjust.
11812 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11813 (add_write, add_read): New functions.
11814 (build_cross_bb_scalars_def): Use it and simplify.
11815 (build_cross_bb_scalars_use): Likewise.
11816 (graphite_find_cross_bb_scalar_vars): Inline into...
11817 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11818 simulating out-of-SSA. Compute liveout and add dependencies.
11819 (build_scops): Force an empty entry block.
11820 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11822 (sese_build_liveouts): Declare.
11823 (sese_trivially_empty_bb_p): Likewise.
11824 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11825 compute liveout and debug_liveout.
11826 (sese_bad_liveouts_use): Remove.
11827 (sese_reset_debug_liveouts_bb): Likewise.
11828 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11829 (sese_build_liveouts): Build liveout and debug_liveout and store
11831 (new_sese_info): Adjust.
11832 (free_sese_info): Likewise.
11833 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11834 do not build liveout here.
11835 (move_sese_in_condition): Adjust region entry.
11836 (scev_analyzable_p): Match up with chrec_apply requirements.
11837 (sese_trivially_empty_bb_p): New.
11838 * tree-into-ssa.c (get_reaching_def): Properly support generating
11839 default-defs for incremental rewrite of anonymous names.
11841 2017-10-06 Richard Biener <rguenther@suse.de>
11843 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11844 precision do not perform modulo reduction.
11846 2017-10-06 Richard Biener <rguenther@suse.de>
11848 PR tree-optimization/82436
11849 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11850 conservatively choose the vectorization factor when checking
11851 whether we can perform the required load permutation.
11852 (vect_transform_slp_perm_load): Assert when we may not fail.
11854 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11856 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11857 message for incompatible -msdata=* and -mcall-* options.
11859 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11861 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11862 rate for post-reload scheduling.
11864 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11866 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11868 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11870 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11871 to improve monte carlo in scimark.
11873 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11875 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11876 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11877 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11878 pentium4_cost, nocona_cost): Set reassociation width to 1.
11879 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11880 width to 2 for fp operations and 1 otherwise.
11881 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11883 (atom_cost): Set reassociation width to 2.
11884 (slm_cost, generic_cost): Set fp reassociation width
11885 to 2 and 1 otherwise.
11886 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11887 (core_cost): Set fp reassociation width to 4 and vector to 2.
11888 (ix86_reassociation_width): Rewrite using cost table; special case
11889 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11890 and TARGET_AVX128_OPTIMAL.
11891 * config/i386/i386.h (processor_costs): Add
11892 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11893 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11894 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11895 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11896 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11897 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11899 2017-10-05 Nathan Sidwell <nathan@acm.org>
11901 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11903 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11905 * config/arm/arm.c (arm_test_fpu_data): New.
11906 (arm_run_selftests): Call arm_test_fpu_data.
11908 2017-10-04 Nathan Sidwell <nathan@acm.org>
11910 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11912 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11914 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11916 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11917 check type for aarch64_simd_valid_immediate.
11918 (aarch64_output_simd_mov_immediate): Update prototype.
11919 (aarch64_simd_valid_immediate): Update prototype.
11920 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11921 support for ORR-immediate.
11922 (and<mode>3): modified pattern to add support for BIC-immediate.
11923 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11924 now checks for valid immediate for BIC and ORR based on new enum
11926 (aarch64_output_simd_mov_immediate): Function now used to output
11927 BIC/ORR imm as well based on new enum argument.
11928 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11930 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11931 (aarch64_reg_or_bic_imm): Likewise.
11933 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11935 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11936 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11938 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11942 PR rtl-optimization/82396
11943 * haifa-sched.c (autopref_multipass_init): Simplify
11945 (autopref_rank_data): Simplify sort order.
11946 * sched-int.h (autopref_multipass_data_): Remove
11947 multi_mem_insn_p, min_offset and max_offset.
11949 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11951 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11953 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11955 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11956 vect_doubleint_cvt.
11958 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11960 * doc/sourcebuild.texi: Document vect_long_mult.
11962 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11964 PR tree-optimization/82413
11965 * fold-const.c (build_range_check): Use widest_int when comparing
11966 the maximum ETYPE value with HIGH.
11968 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11970 PR rtl-optimization/82396
11971 * haifa-sched.c (autopref_multipass_init): Simplify
11973 (autopref_rank_data): Simplify sort order.
11974 * sched-int.h (autopref_multipass_data_): Remove
11975 multi_mem_insn_p, min_offset and max_offset.
11977 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11979 PR tree-optimization/82381
11980 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11981 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11984 PR tree-optimization/82374
11985 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11986 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11987 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11988 current_function_decl to the new decl.
11990 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11992 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11993 helper macro for IEEE float128 hardware built-in functions.
11994 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11996 (TRUNCF128_ODD): Likewise.
11997 (ADDF128_ODD): Likewise.
11998 (SUBF128_ODD): Likewise.
11999 (MULF128_ODD): Likewise.
12000 (DIVF128_ODD): Likewise.
12001 (FMAF128_ODD): Likewise.
12002 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12003 UNSPEC_TRUNC_ROUND_TO_ODD.
12004 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12005 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12006 floating point round to odd instructions.
12007 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12008 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12009 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12010 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12011 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12012 (trunc<mode>sf2_hw): Change the truncate with round to odd
12013 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12014 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12015 to odd hardware instructions.
12016 (sub<mode>3_odd): Likewise.
12017 (mul<mode>3_odd): Likewise.
12018 (div<mode>3_odd): Likewise.
12019 (sqrt<mode>2_odd): Likewise.
12020 (fma<mode>4_odd): Likewise.
12021 (fms<mode>4_odd): Likewise.
12022 (nfma<mode>4_odd): Likewise.
12023 (nfms<mode>4_odd): Likewise.
12024 (trunc<mode>df2_odd): Change the truncate with round to odd
12025 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12027 * doc/extend.texi (PowerPC built-in functions): Update documentation
12028 for existing IEEE float128-bit built-in functions. Add built-in
12029 functions that generate the IEEE 128-bit floating point round to
12032 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12034 PR rtl-optimization/77729
12035 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12036 to (X&(C1&~C2))|C2 transformations.
12038 2017-10-03 Martin Jambor <mjambor@suse.cz>
12040 PR tree-optimization/82363
12041 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12042 mismatch, mark lacc written regardless of racc.
12044 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12046 PR tree-optimization/82381
12047 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12048 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12049 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12050 fallthrough into reassoc_stmt_dominates_stmt_p.
12053 * combine.c (combine_instructions): Don't combine in unreachable
12056 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12059 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12060 function to not use the have_cpu variable. Do not set cpu_index,
12061 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12062 or the default cpu.
12063 (rs6000_valid_attribute_p): Remove duplicate initializations of
12064 old_optimize and func_optimize.
12065 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12066 (rs6000_activate_target_options): Make global.
12067 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12070 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12072 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12073 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12074 Set *poffset to 0 and *psize and *pmax_size to -1 if
12075 *poffset + *psize overflows in HOST_WIDE_INT.
12077 PR tree-optimization/82387
12078 PR tree-optimization/82388
12079 PR tree-optimization/82389
12080 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12081 instead of live_bytes non-NULL.
12083 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12086 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12089 2017-10-02 Richard Biener <rguenther@suse.de>
12091 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12092 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12093 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12095 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12097 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12098 requested precision matches the type's.
12099 * calls.c (alloc_max_size): Calculate the new candidate size as
12100 a widest_int and use wi::to_widest when comparing it with the
12101 current candidate size.
12102 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12103 zero rather than integer_zero_node.
12104 * match.pd: Check for a no-op conversion before using wi::add
12105 rather than after. Use tree_to_uhwi when summing small shift
12106 counts into an unsigned int.
12108 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12109 Alan Hayward <alan.hayward@arm.com>
12110 David Sherwood <david.sherwood@arm.com>
12113 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12114 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12115 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12116 POINTER_AND_FP_REGS.
12118 2017-10-02 Richard Biener <rguenther@suse.de>
12120 PR tree-optimization/82355
12121 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12122 a mapping for the enclosing loop but avoid generating one for
12123 the loop tree root.
12124 (copy_bb_and_scalar_dependences): Remove premature codegen
12125 error on PHIs in blocks duplicated into multiple places.
12126 * graphite-scop-detection.c
12127 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12128 in the region use it as loop and nest to analyze the DR in.
12129 (try_generate_gimple_bb): Likewise.
12130 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12131 (add_loop_constraints): For blocks in a loop not in the region
12132 create a dimension with a single iteration.
12133 * sese.h (gbb_loop_at_index): Remove assert.
12135 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12137 * omp-expand.c (adjust_context_scope): New function.
12138 (expand_parallel_call): Call adjust_context_scope.
12140 2017-10-01 Jeff Law <law@redhat.com>
12142 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12143 dom_opt_dom_walker class with direct access to private members.
12144 Add comments. Call test_for_singularity.
12145 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12146 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12147 m_dummy_cond anymore.
12148 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12150 (pass_dominator:execute): Build the dummy_cond here and pass it
12151 to the dom_opt_dom_walker ctor.
12152 (test_for_singularity): New function.
12154 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12155 Maya Rashish <coypu@sdf.org>
12157 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12158 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12159 (alpha*-*-netbsd*) Use nbsd_tm_file.
12160 (arm*-*-netbsdelf*) Likewise.
12161 (i[34567]86-*-netbsdelf*) Likewise.
12162 (x86_64-*-netbsd*) Likewise.
12163 (mips*-*-netbsd*) Likewise.
12164 (powerpc-*-netbsd*) Likewise.
12165 (sh*-*-netbsd*) Likewise.
12166 (sparc-*-netbsdelf*) Likewise.
12167 (sparc64-*-netbsd*) Likewise.
12168 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12170 (vax-*-netbsdelf*) Likewise.
12171 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12172 (UINT_FAST8_TYPE) Likewise.
12173 (INT_FAST16_TYPE) Check CHAR_FAST16.
12174 (UINT_FAST16_TYPE) Likewise.
12176 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12179 * config/i386/i386.md
12180 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12181 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12182 *divmodsi4_zext_2): New define_insn_and_split.
12183 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12184 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12185 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12186 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12187 New define_insn_and_split.
12188 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12189 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12190 operands[1] having DImode when mode is SImode.
12192 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12193 always SImode for DIV and MOD in REG_EQUAL notes.
12195 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12197 PR middle-end/82319
12198 * match.pd: Fix handling of NaNs in pattern.
12200 2017-09-29 Jeff Law <law@redhat.com>
12202 * sbitmap.c (bitmap_bit_in_range_p): New function.
12203 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12204 * tree-ssa-dse.c (live_bytes_read): New function.
12205 (dse_classify_store): Ignore reads of dead bytes.
12207 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12208 typos and whitespace errors.
12209 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12210 * config/s390/s390.c (s390_emit_prologue): Likewise.
12212 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12215 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12216 with a symbol for LRA.
12218 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12220 PR rtl-optimization/82338
12221 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12223 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12225 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12226 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12227 (qsort_chk): Declare.
12228 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12229 (qsort_chk): New function.
12231 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12233 PR tree-optimization/82337
12234 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12235 phi definition if the PHI result appears in an abnormal PHI.
12236 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12237 basis appears in an abnormal PHI.
12239 2017-09-29 Richard Biener <rguenther@suse.de>
12241 * graphite-isl-ast-to-gimple.c
12242 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12243 (binary_op_to_tree): Use it.
12244 (get_rename_from_scev): Likewise.
12245 (copy_loop_phi_nodes): Likewise.
12246 (copy_bb_and_scalar_dependences): Likewise.
12247 (translate_pending_phi_nodes): Likewise.
12249 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12252 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12253 for movabsq $(i32 << shift), r64.
12255 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12257 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12258 index when encoding %esp as %rsp to avoid 0x67 prefix.
12260 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12262 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12263 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12265 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12267 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12268 Extensions with more than 16 double VFP registers.
12269 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12270 to_clear_mask and all code related to it. Replace the remaining
12271 entry by a sbitmap and adapt code accordingly.
12273 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12275 * brig-builtins.def: Change pure attributes to const.
12277 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12279 * config.gcc (default_gnu_indirect_function): Default to yes for
12280 sparc*-*-linux* with glibc.
12282 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12284 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12285 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12286 when creating .init_array and .fini_array sections with priority
12289 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12292 * config/aarch64/aarch64.c
12293 (aarch64_builtin_support_vector_misalignment): Always return false
12294 when misalignment is unknown.
12296 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12298 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12299 this function to return false if the definition used by the swap
12300 instruction is artificial, or if the memory address from which the
12301 constant value is loaded is not represented by a base address held
12302 in a register or if the base address register is a frame or stack
12303 pointer. Additionally, return false if the base address of the
12304 loaded constant is a SYMBOL_REF but is not considered to be a
12306 (replace_swapped_load_constant): New function.
12307 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12308 loaded constant vector with a load of a swapped constant vector.
12310 2017-09-27 Carl Love <cel@us.ibm.com>
12312 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12313 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12314 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12317 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12319 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12320 first, always call autopref_rank_data otherwise.
12322 2017-09-27 Richard Biener <rguenther@suse.de>
12324 * graphite-scop-detection.c (find_scop_parameters): Move
12325 loop bound handling ...
12326 (gather_bbs::before_dom_children): ... here, avoiding the need
12327 to build scop_info->loop_nest.
12328 (record_loop_in_sese): Remove.
12329 * sese.h (sese_info_t::loop_nest): Remove.
12330 * sese.c (new_sese_info): Do not allocate loop_nest.
12331 (free_sese_info): Do not free loop_nest.
12333 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12336 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12337 lhs from calls if the lhs has addressable type.
12339 2017-09-27 Richard Biener <rguenther@suse.de>
12341 * graphite.h (scop::max_alias_set): New member.
12342 * graphite-scop-detection.c: Remove references to non-existing
12343 --param in comments.
12344 (build_alias_sets): Record the maximum alias set used for drs.
12345 (build_scops): Support zero as unlimited for
12346 --param graphite-max-arrays-per-scop.
12347 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12348 and inline into ...
12349 (build_poly_sr_1): ... here. Compute alias set based on the
12350 maximum alias set used for drs rather than
12351 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12353 2017-09-27 Richard Biener <rguenther@suse.de>
12355 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12356 --param loop-block-tile-size=0 to disable tiling.
12358 2017-09-27 Richard Biener <rguenther@suse.de>
12360 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12361 (graphite-max-nb-scop-params): Document special value zero.
12362 * domwalk.h (dom_walker::STOP): New symbolical constant.
12363 (dom_walker::dom_walker): Add optional parameter for bb to
12365 (dom_walker::~dom_walker): Declare.
12366 (dom_walker::before_dom_children): Document STOP return value.
12367 (dom_walker::m_user_bb_to_rpo): New member.
12368 (dom_walker::m_bb_to_rpo): Likewise.
12369 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12370 mapping here if not provided by the user.
12371 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12372 provided by the user.
12373 (dom_walker::STOP): Define.
12374 (dom_walker::walk): Do not compute bb to RPO mapping here.
12375 Support STOP return value from before_dom_children to stop
12377 * graphite-optimize-isl.c (optimize_isl): If the schedule
12378 is the same still generate code if -fgraphite-identity
12379 or -floop-parallelize-all are given.
12380 * graphite-scop-detection.c: Include cfganal.h.
12381 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12383 (gather_bbs::before_dom_children): Return STOP for BBs
12385 (build_scops): Compute bb to RPO mapping and pass it to
12386 the domwalk. Treat --param graphite-max-nb-scop-params=0
12387 as not limiting the number of params.
12388 * graphite.c (graphite_initialize): Remove limit on the
12389 number of basic-blocks in a function.
12390 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12391 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12392 default value of 10.
12394 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12396 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12397 Adjust code to eliminate needing to do the shift right 32-bits
12398 operation after XSCVDPSPN.
12400 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12402 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12403 ((X / Y) != 0 -> X >= Y): Likewise.
12405 2017-09-26 Carl Love <cel@us.ibm.com>
12407 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12408 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12409 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12410 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12411 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12412 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12413 definitions and overloading.
12414 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12415 statement for P9V_BUILTIN_XST_LEN_R.
12416 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12417 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12418 define_expand and define_insn for the instructions and builtins.
12419 * doc/extend.texi: Update the built-in documentation file for the new
12420 built-in functions.
12421 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12422 define_insn for the instructions
12424 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12427 * gcc/config/netbsd-protos.h: New file.
12428 * gcc/config/netbsd.c: New file.
12429 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12430 * gcc/config/t-netbsd: New file.
12431 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12432 (tmake_file) Add t-netbsd.
12433 (extra_objs) Add netbsd.o.
12435 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12439 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12441 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12443 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12444 sign extension from a vector register to a GPR by doing a 32-bit
12445 direct move and then an EXTSW.
12446 (extendsi<mode>2 splitter): Likewise.
12447 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12448 right or vector extract after doing XSCVDPSPN. Use
12449 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12451 (movdi_from_sf_zero_ext): Likewise.
12452 (reload_gpr_from_vsxsf): Likewise.
12453 (p8_mfvsrd_4_disf): Delete, no longer used.
12454 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12455 then needing to move the SFmode to a GPR to use the XSCVDPSP
12456 instruction instead of FRSP and XSCVDPSPN.
12457 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12458 it is adjacent to the other XSCVSPDP insns.
12459 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12460 SFmode to be in traditional Altivec registers.
12461 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12462 (vsx_xscvspdpn): Likewise.
12463 (vsx_xscvspdpn_scalar): Likewise.
12465 2017-09-26 Martin Jambor <mjambor@suse.cz>
12467 * tree-sra.c (compare_access_positions): Put integral types first,
12468 stabilize sorting of integral types, remove conditions putting
12469 non-full-precision integers last.
12470 (sort_and_splice_var_accesses): Disable scalarization if a
12471 non-integert would be represented by a non-full-precision integer.
12473 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12475 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12476 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12477 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12478 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12479 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12480 conditionals inside the function instead of around it. Call
12481 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12482 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12484 2017-09-26 Richard Biener <rguenther@suse.de>
12486 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12488 (scop_detection::build_scop_breadth): ... this. Removed.
12489 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12490 (scop_detection::harmful_stmt_in_bb): Likewise.
12491 (scop_detection::graphite_can_represent_stmt): Likewise.
12492 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12493 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12494 (scop_detection::can_represent_loop_1): ... this. Removed.
12495 (scop_detection::harmful_loop_in_region): Simplify after inlining
12496 the above and remove more quadraticness.
12497 (build_scops): Adjust.
12498 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12501 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12504 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12505 REGNO (base) == SP_REG if base is a REG.
12507 PR middle-end/35691
12508 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12509 if it is different SSA_NAME.
12510 (optimize_range_tests_cmp_bitwise): New function.
12511 (optimize_range_tests): Call it.
12513 2017-09-26 Richard Biener <rguenther@suse.de>
12515 PR tree-optimization/82321
12516 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12517 for the def being inside the loop.
12519 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12521 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12523 * config/s390/s390-builtins.def: Fix constraint on op4.
12525 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12527 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12528 independent expanders.
12529 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12530 ("vec_ordered", "vec_unordered"): New expanders.
12532 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12534 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12537 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12539 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12540 vec_unpacks_lo_v16qi.
12541 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12543 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12545 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12546 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12547 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12549 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12551 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12553 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12555 ("*vec_slb<mode>"): New insn pattern.
12556 ("vec_shr_<mode>"): New expander.
12557 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12558 and force the shift count operand to V16QImode.
12559 ("vec_srb<mode>"): Set shift count mode to V16QI.
12561 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12563 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12564 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12565 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12567 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12570 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12572 2017-09-26 Richard Biener <rguenther@suse.de>
12574 PR tree-optimization/82320
12575 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12578 2017-09-25 Jeff Law <law@redhat.com>
12580 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12581 prototype for new argument.
12582 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12583 mostly extracted from rs6000_emit_allocate_stack.
12584 (rs6000_emit_probe_stack_range_stack_clash): New function.
12585 (rs6000_emit_allocate_stack): Call
12586 rs6000_emit_probe_stack_range_stack_clash as needed.
12587 (rs6000_emit_probe_stack_range): Add additional argument
12588 to call to gen_probe_stack_range{si,di}.
12589 (output_probe_stack_range): New.
12590 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12591 (output_probe_stack_range_stack_clash): New.
12592 (rs6000_emit_prologue): Emit notes into dump file as requested.
12593 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12594 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12595 Add additional operand and pass it to output_probe_stack_range.
12597 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12599 PR tree-optimization/82163
12600 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12601 (checking_verify_loop_closed_ssa): New parameter.
12602 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12603 (check_loop_closed_ssa_stmt): Delete.
12604 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12605 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12606 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12609 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12611 * brig-builtins.def: Treat HSAIL barrier builtins as
12612 setjmp/longjump style functions.
12614 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12616 * target.def (constant_alignment): New hook.
12617 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12618 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12619 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12620 * doc/tm.texi: Regenerate.
12621 * targhooks.h (default_constant_alignment): Declare.
12622 (constant_alignment_word_strings): Likewise.
12623 * targhooks.c (default_constant_alignment): New function.
12624 (constant_alignment_word_strings): Likewise.
12625 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12626 instead of CONSTANT_ALIGNMENT.
12627 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12628 (force_const_mem): Likewise.
12629 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12630 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12631 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12632 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12633 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12635 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12636 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12637 constant_alignment_word_strings.
12638 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12639 (CONSTANT_ALIGNMENT): Likewise.
12640 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12641 (arm_constant_alignment): New function.
12642 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12643 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12644 constant_alignment_word_strings.
12645 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12646 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12647 constant_alignment_word_strings.
12648 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12649 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12650 (cris_constant_alignment): New function.
12651 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12652 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12653 (epiphany_constant_alignment): New function.
12654 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12655 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12656 constant_alignment_word_strings.
12657 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12658 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12659 constant_alignment_word_strings.
12660 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12661 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12662 constant_alignment_word_strings.
12663 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12664 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12665 * config/i386/i386.c (ix86_constant_alignment): Make static.
12666 Use the same interface as the target hook.
12667 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12668 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12669 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12670 constant_alignment_word_strings.
12671 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12672 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12673 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12674 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12675 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12676 constant_alignment_word_strings.
12677 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12678 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12679 constant_alignment_word_strings.
12680 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12681 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12682 constant_alignment_word_strings.
12683 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12684 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12686 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12687 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12688 * config/mips/mips.c (mips_constant_alignment): New function.
12689 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12690 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12691 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12692 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12693 (mmix_constant_alignment): Make static. Use the same interface
12694 as the target hook.
12695 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12696 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12697 constant_alignment_word_strings.
12698 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12699 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12700 constant_alignment_word_strings.
12701 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12702 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12703 constant_alignment_word_strings.
12704 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12705 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12706 (rs6000_constant_alignment): New function.
12707 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12708 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12709 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12710 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12711 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12712 (rs6000_constant_alignment): New function.
12713 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12714 * config/s390/s390.c (s390_constant_alignment): New function.
12715 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12716 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12717 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12718 constant_alignment_word_strings.
12719 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12720 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12721 (sparc_constant_alignment): New function.
12722 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12723 * config/spu/spu.c (spu_constant_alignment): New function.
12724 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12725 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12726 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12727 constant_alignment_word_strings.
12728 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12729 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12730 constant_alignment_word_strings.
12731 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12732 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12733 constant_alignment_word_strings.
12734 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12735 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12736 (visium_constant_alignment): New function.
12737 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12738 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12739 (xtensa_constant_alignment): New function.
12740 * system.h (CONSTANT_ALIGNMENT): Poison.
12742 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12744 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12745 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12746 (rs6000_builtin_valid_without_lhs): New helper function.
12747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12748 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12750 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12752 * target.h (vec_perm_indices): Use unsigned short rather than
12754 (auto_vec_perm_indices): Likewise.
12755 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12756 Use unsigned int rather than unsigned char.
12757 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12759 2017-09-25 Richard Biener <rguenther@suse.de>
12761 * cfgloop.h (sort_sibling_loops): Declare.
12762 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12763 (sort_sibling_loops): New function sorting the sibling loop list
12765 * graphite.c (graphite_transform_loops): Sort sibling loops.
12767 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12769 * target.def (vec_perm_const_ok): Change sel parameter to
12771 * optabs-query.c (can_vec_perm_p): Update accordingly.
12772 * doc/tm.texi: Regenerate.
12773 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12774 auto_vec_perm_indices and remove separate nelt field.
12775 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12776 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12777 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12778 (aarch64_expand_vec_perm_const): Update accordingly.
12779 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12780 to vec_perm_indices.
12781 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12782 auto_vec_perm_indices and remove separate nelt field.
12783 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12784 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12785 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12787 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12788 to vec_perm_indices.
12789 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12790 sel to vec_perm_indices.
12791 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12792 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12793 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12795 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12798 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12801 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12802 on the FUNCTION_DECL function context if it has a DIE that is a
12805 2017-09-25 Richard Biener <rguenther@suse.de>
12807 PR tree-optimization/82285
12808 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12811 2017-09-25 Tom de Vries <tom@codesourcery.com>
12815 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12818 2017-09-25 Richard Biener <rguenther@suse.de>
12820 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12821 ISL errors other than isl_error_quota happen. Dump if the
12822 schedule is the same.
12823 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12824 errors instead of aborting inside ISL.
12826 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12829 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12830 of libgcc_eh for m64.
12831 * config/i386/darwin64.h: Likewise.
12833 2017-09-25 Richard Biener <rguenther@suse.de>
12835 PR middle-end/82144
12836 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12837 attribute for incomplete types nor twice for complete ones.
12839 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12842 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12843 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12846 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12849 * config/i386/i386.opt (mprefer-avx256): Use
12850 ix86_target_flags variable.
12851 * config/i386/i386.c (ix86_target_string): Move
12852 -mprefer-avx256 to flag2_opts.
12854 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12856 PR middle-end/35691
12857 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12858 and x != -1 | y != -1 into (x & y) != -1.
12860 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12862 * config.gcc: Add new case statement to set
12863 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12864 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12865 s390x-*-linux* case statements. Added aarch64 to the list of
12866 supported architectures.
12868 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12870 PR tree-optimization/82289
12871 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12872 STMT_VINFO_RELEVANT_P.
12874 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12875 Alan Hayward <alan.hayward@arm.com>
12876 David Sherwood <david.sherwood@arm.com>
12878 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12879 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12880 (extract_range_from_binary_expr_1): Don't call
12881 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12883 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12884 Alan Hayward <alan.hayward@arm.com>
12885 David Sherwood <david.sherwood@arm.com>
12887 * target.def (preferred_vector_alignment): New hook.
12888 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12890 * doc/tm.texi: Regenerate.
12891 * targhooks.h (default_preferred_vector_alignment): Declare.
12892 * targhooks.c (default_preferred_vector_alignment): New function.
12893 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12895 (DR_TARGET_ALIGNMENT): New macro.
12896 (aligned_access_p): Update commentary.
12897 (vect_known_alignment_in_bytes): New function.
12898 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12900 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12901 Calculate the misalignment based on the target alignment rather than
12903 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12904 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12905 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12906 the target alignment, rather than masking the element misalignment
12907 with the number of elements in a vector. Also use the target
12908 alignment when calculating the maximum number of peels.
12909 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12910 instead of TYPE_ALIGN_UNIT.
12911 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12912 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12913 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12914 (vect_create_data_ref_ptr): Likewise.
12915 (vect_setup_realignment): Realign by ANDing with
12916 -DR_TARGET_MISALIGNMENT.
12917 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12918 the number of peels based on DR_TARGET_ALIGNMENT.
12919 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12920 with the guaranteed alignment boundary when deciding whether
12922 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12923 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12924 (ensure_base_align): Remove stmt_info parameter. Get the
12925 target base alignment from DR_TARGET_ALIGNMENT.
12926 (vectorizable_store): Update call accordingly. Interpret
12927 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12929 (vectorizable_load): Likewise.
12931 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12932 Alan Hayward <alan.hayward@arm.com>
12933 David Sherwood <david.sherwood@arm.com>
12935 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12936 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12937 (vect_enhance_data_refs_alignment): Likewise.
12939 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12941 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12942 error. Only quit immediately if parsing is complete.
12943 (BEGIN): Initialize fatal_err and parse_done.
12944 (begin fpu, end fpu): Check number of arguments.
12945 (begin arch, end arch): Likewise.
12946 (begin cpu, end cpu): Likewise.
12947 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12948 (optalias): Likewise.
12950 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12952 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12953 * config/arm/arm-isa.h: Delete. Move definitions to ...
12954 * arm-cpus.in: ... here. Use new feature and fgroup values.
12955 * config/arm/arm.c (arm_option_override): Use lower case for feature
12957 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12958 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12959 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12961 (print_isa_bits_for): New function.
12962 (gen_isa): New function.
12963 (gen_comm_data): Use print_isa_bits_for.
12964 (define feature): New keyword.
12965 (define fgroup): New keyword.
12966 * config/arm/t-arm (TM_H): Remove.
12967 (GTM_H): Add arm-isa.h.
12968 (arm-isa.h): Add rule to generate file.
12969 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12970 case for feature bit names.
12972 2017-09-22 Richard Biener <rguenther@suse.de>
12974 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12976 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12977 print of no dependency loops ...
12978 * graphite.c (graphite_transform_loops): ... here.
12979 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12981 (same_close_phi_node, remove_duplicate_close_phi,
12982 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12983 (canonicalize_loop_closed_ssa): ... here and simplify.
12984 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12985 (optimize_isl): Use dump_printf_loc to tell when we stopped
12986 optimizing because of an ISL timeout.
12988 2017-09-22 Richard Biener <rguenther@suse.de>
12990 PR tree-optimization/82291
12991 * tree-if-conv.c (predicate_mem_writes): Make sure to
12992 remove writes in blocks predicated with false.
12994 2017-09-22 Richard Biener <rguenther@suse.de>
12996 * sese.c: Include cfganal.h.
12997 (if_region_set_false_region): Remove.
12998 (create_if_region_on_edge): Likewise.
12999 (move_sese_in_condition): Re-implement without destroying
13002 2017-09-22 Richard Biener <rguenther@suse.de>
13004 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13005 Verify both BBs contain loop PHI nodes before dispatching to
13006 copy_loop_phi_args.
13007 (graphite_regenerate_ast_isl): Do not recompute dominators,
13008 do not verify three times. Restructure for clarity.
13009 * graphite-scop-detection.c (same_close_phi_node,
13010 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13011 defined_in_loop_p, canonicalize_loop_closed_ssa,
13012 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13013 checking and SSA rewrite, move to ...
13014 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13015 (graphite_initialize): Do not pass in ctx, do not reset the
13016 SCEV cache, compute only dominators.
13017 (graphite_transform_loops): Allocate ISL ctx after
13018 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13019 Maintain post-dominators only around build_scops.
13020 * sese.c (if_region_set_false_region): Make static. Free
13021 and recompute dominators.
13022 (move_sese_in_condition): Assert we don't get called with
13023 post-dominators computed.
13024 * sese.h (if_region_set_false_region): Remove.
13026 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13028 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13029 mode attribute for TARGET_AVX512VL.
13031 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13033 * config/i386/i386.opt (mprefer-avx256): New option.
13034 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13036 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13037 for TARGET_PREFER_AVX256.
13038 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13040 2017-09-21 Jeff Law <law@redhat.com>
13042 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13043 Fix dump output if the only stack space is for pushed registers.
13045 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13047 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13050 2017-09-21 Martin Sebor <msebor@redhat.com>
13053 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13054 code (in C++) or code that triggers warnings.
13056 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13058 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13060 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13062 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13063 * sched-rgn.c: Ditto.
13064 * sel-sched-ir.c: Ditto.
13066 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13068 * toplev.h (set_random_seed): Adjust return type.
13069 * toplev.c (init_local_tick): Move eager initialization of random_seed
13070 to get_random_seed. Adjust comment.
13071 (init_random_seed): Inline to get_random_seed, delete.
13072 (get_random_seed): Initialize random_seed lazily.
13073 (set_random_seed): Do not return previous value.
13074 (print_switch_value): Do not call get_random_seed.
13076 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13078 * cgraph.c (delete_function_version): New, broken out from...
13079 (cgraph_node::delete_function_version): ...here. Rename to
13080 cgraph_node::delete_function_version_by_decl. Update all uses.
13081 (cgraph_node::remove): Call delete_function_version.
13083 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13086 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13087 VAR_DECLs to which addressable non-volatile parameters are mapped
13088 and for id->retvar after the return value assignment. Clear
13089 id->retval and id->retbnd after inlining.
13091 2017-09-21 Richard Biener <rguenther@suse.de>
13093 PR tree-optimization/82276
13094 PR tree-optimization/82244
13095 * tree-vrp.c (build_assert_expr_for): Set
13096 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13098 (remove_range_assertions): Revert earlier change.
13100 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13103 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13105 2017-09-21 Richard Biener <rguenther@suse.de>
13107 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13108 Restore valid IL after code generation errors.
13109 * graphite.c (graphite_transform_loops): Diagnose code
13110 generation issues as MSG_MISSED_OPTIMIZATION and continue
13111 with processing SCOPs.
13113 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13114 Alan Hayward <alan.hayward@arm.com>
13115 David Sherwood <david.sherwood@arm.com>
13117 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13118 rather than choosing between plus_constant and gen_rtx_<CODE>.
13119 * expr.c (emit_push_insn): Likewise.
13120 (expand_expr_real_2): Likewise.
13122 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13123 Alan Hayward <alan.hayward@arm.com>
13124 David Sherwood <david.sherwood@arm.com>
13126 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13128 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13129 Alan Hayward <alan.hayward@arm.com>
13130 David Sherwood <david.sherwood@arm.com>
13132 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13133 calling tree_to_uhwi.
13135 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13136 Alan Hayward <alan.hayward@arm.com>
13137 David Sherwood <david.sherwood@arm.com>
13139 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13140 INTEGER_CST rather than a negative test for ADDR_EXPR.
13142 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13143 Alan Hayward <alan.hayward@arm.com>
13144 David Sherwood <david.sherwood@arm.com>
13146 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13147 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13149 2017-09-21 Richard Biener <rguenther@suse.de>
13151 PR tree-optimization/71351
13152 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13153 graphite_create_new_loop_guard): Remove, fold remaining parts
13155 (translate_isl_ast_node_for): ... here and simplify.
13157 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13160 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13161 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13162 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13163 alternative always use QI mode, for -Os imov (=R,R) alternative
13164 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13167 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13168 Jeff Law <law@redhat.com>
13170 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13171 (allocate_stack_space): New function, partially extracted from
13172 s390_emit_prologue.
13173 (s390_emit_prologue): Track offset to most recent stack probe.
13174 Code to allocate space moved into allocate_stack_space.
13175 Dump actions when no stack is allocated.
13176 (s390_prologue_plus_offset): New function.
13177 (s390_emit_stack_probe): Likewise.
13179 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13181 * common.opt (Wa, Wl, Wp, g, gz=): Add
13183 (gno-column-info): Remove.
13184 (gcolumn-info): Drop RejectNegative.
13185 (gno-): New prefix.
13186 (gno-record-gcc-switches): Remove.
13187 (grecord-gcc-switches): Drop RejectNegative.
13188 (gno-split-dwarf): Remove.
13189 (gsplit-dwarf): Drop RejectNegative.
13190 (gno-strict-dwarf): Remove.
13191 (gstrict-dwarf): Drop RejectNegative.
13192 * config/darwin.opt (gfull, gused): Add RejectNegative.
13193 * dwarf2out.c (gen_producer_string): Drop
13194 gno-record-gcc-switches handler.
13195 * optc-gen.awk: Add g to prefixes with negative forms.
13196 * opts-common.c (remapping_prefix_p): New.
13197 (find_opt): Check it.
13198 (generate_canonical_option): Test g prefix.
13199 (option_map): Add -gno- mapping.
13200 (add_misspelling_candidates): Check remapping_prefix_p.
13202 2017-09-20 Jeff Law <law@redhat.com>
13204 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13205 thinko in stack clash protection support.
13207 * explow.c (compute_stack_clash_protection_loop_data): Use
13208 CONST_INT_P instead of explicit test. Verify object is a
13209 CONST_INT_P before looking at INTVAL.
13210 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13211 instead of explicit test.
13213 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13216 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13217 address instead of to r1 and r11.
13219 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13221 * config.gcc: Support "knm".
13222 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13223 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13225 * config/i386/i386.c (m_KNM): Define.
13226 (processor_target_table): Add "knm".
13228 (ix86_option_override_internal): Add "knm".
13229 (ix86_issue_rate): Add PROCESSOR_KNM.
13230 (ix86_adjust_cost): Ditto.
13231 (ia32_multipass_dfa_lookahead): Ditto.
13232 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13233 (fold_builtin_cpu): Add M_INTEL_KNM.
13234 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13235 (processor_type): Add PROCESSOR_KNM.
13236 * config/i386/x86-tune.def: Add m_KNM.
13237 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13239 2017-09-20 Richard Biener <rguenther@suse.de>
13241 PR tree-optimization/80213
13242 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13243 are allowed in empty BBs as well.
13244 (canonicalize_loop_closed_ssa): Also look for other complex
13246 (scop_detection::get_sese): Include the loop-closed PHI block
13248 (scop_detection::merge_sese): Remove code adding extra blocks.
13249 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13250 (build_scops): Assert the final returned scop is invalid.
13252 2017-09-20 Richard Biener <rguenther@suse.de>
13254 PR tree-optimization/82264
13255 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13257 (vn_phi_lookup): Likewise.
13258 (vn_phi_insert): Likewise.
13260 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13262 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13263 that fits into uhwi or shwi, add DW_AT_const_value regardless
13264 of early_dwarf without going through RTL, using add_AT_unsigned
13267 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13268 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13269 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13270 (DEBUG_MACRO_SECTION): Likewise.
13271 (DEBUG_LTO_MACRO_SECTION): Likewise.
13272 (DEBUG_STR_DWO_SECTION): Likewise.
13273 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13274 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13275 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13276 (DEBUG_LTO_LINE_STR_SECTION): Define.
13277 (init_sections_and_labels): Initialize debug_line_str_section
13278 variable. Initialize debug_loc_section for -gdwarf-5 to
13279 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13281 2017-09-20 Richard Biener <rguenther@suse.de>
13283 * graphite-sese-to-poly.c (extract_affine): Properly handle
13284 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13286 2017-09-20 Richard Biener <rguenther@suse.de>
13288 PR tree-optimization/81373
13289 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13290 Force SESE live-out defs to be handled even if they are
13293 2017-09-19 Jeff Law <law@redhat.com>
13295 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13296 nothing for stack adjustments with REG_STACK_CHECK.
13297 * sched-deps.c (parse_add_or_inc): Reject insns with
13298 REG_STACK_CHECK from dependency breaking.
13299 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13300 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13301 * reg-notes.def (STACK_CHECK): New note.
13303 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13304 (ix86_expand_prologue): Dump stack clash info as needed.
13305 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13307 * function.c (dump_stack_clash_frame_info): New function.
13308 * function.h (dump_stack_clash_frame_info): Prototype.
13309 (enum stack_clash_probes): New enum.
13311 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13312 flag_stack_clash_protection.
13313 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13314 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13315 (arm_frame_pointer_required): Likewise.
13316 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13317 (ia64_expand_prologue): Likewise.
13318 * config/mips/mips.c (mips_expand_prologue): Likewise.
13319 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13320 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13321 (sparc_flat_expand_prologue): Likewise.
13322 * config/spu/spu.c (spu_expand_prologue): Likewise.
13324 * explow.c: Include "params.h".
13325 (anti_adjust_stack_and_probe_stack_clash): New function.
13326 (get_stack_check_protect): Likewise.
13327 (compute_stack_clash_protection_loop_data): Likewise.
13328 (emit_stack_clash_protection_loop_start): Likewise.
13329 (emit_stack_clash_protection_loop_end): Likewise.
13330 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13331 Use anti_adjust_stack_and_probe_stack_clash.
13332 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13333 (emit_stack_clash_protection_loop_start): Likewise.
13334 (emit_stack_clash_protection_loop_end): Likewise.
13335 * rtl.h (get_stack_check_protect): Prototype.
13336 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13337 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13338 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13340 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13342 * doc/tm.texi: Rebuilt.
13343 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13344 get_stack_check_protect.
13345 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13346 * config/arm/arm.c (arm_expand_prologue): Likewise.
13347 (arm_frame_pointer_required): Likewise.
13348 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13349 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13350 * config/mips/mips.c (mips_expand_prologue): Likewise.
13351 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13352 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13353 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13354 (sparc_flat_expand_prologue): Likewise.
13356 * common.opt (-fstack-clash-protection): New option.
13357 * flag-types.h (enum stack_check_type): Note difference between
13358 -fstack-check= and -fstack-clash-protection.
13359 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13360 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13361 * toplev.c (process_options): Issue warnings/errors for cases
13362 not handled with -fstack-clash-protection.
13363 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13364 (-fstack-check): Note additional problem with -fstack-check=generic.
13365 Note that -fstack-check is primarily for Ada and refer users
13366 to -fstack-clash-protection for stack-clash-protection.
13367 Document new params for stack clash protection.
13369 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13371 * config/i386/i386.c (ix86_split_long_move): Do not handle
13372 address used for LEA in a special way.
13374 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13376 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13378 2017-09-19 Martin Sebor <msebor@redhat.com>
13381 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13382 of incompatible types.
13384 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13387 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13388 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13389 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13391 2017-09-19 Richard Biener <rguenther@suse.de>
13393 PR tree-optimization/82244
13394 * tree-vrp.c (remove_range_assertions): Do not propagate
13395 a constant to abnormals but replace the assert with a copy.
13397 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13399 PR rtl-optimization/57878
13400 PR rtl-optimization/68988
13401 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13402 avoidance test involving non_reload_pseudos. Move frequency test
13403 below the general fragmentation avoidance test.
13405 2017-09-19 Richard Biener <rguenther@suse.de>
13407 PR tree-optimization/69728
13408 * graphite-sese-to-poly.c (schedule_error): New global.
13409 (add_loop_schedule): Handle empty domain by failing the
13411 (build_original_schedule): Handle schedule_error.
13413 2017-09-19 Richard Biener <rguenther@suse.de>
13415 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13416 Do not iterate to sibling loops but only to siblings of inner
13419 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13422 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13423 that the registers are different.
13425 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13427 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13428 to processor_model and "amdfam17h" to arch_names_table.
13429 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13431 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13434 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13436 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13437 Alan Hayward <alan.hayward@arm.com>
13438 David Sherwood <david.sherwood@arm.com>
13440 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13442 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13444 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13445 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13446 vect_schedule_slp_instance.
13447 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13448 Update call to vect_slp_analyze_node_operations. Simplify return
13450 (vect_slp_analyze_bb_1): Update call accordingly.
13451 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13452 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13453 (vect_schedule_slp): Update call accordingly.
13455 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13456 Alan Hayward <alan.hayward@arm.com>
13457 David Sherwood <david.sherwood@arm.com>
13459 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13460 with types that aren't in fact scalar.
13462 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13464 * tree-vect-slp.c (vect_record_max_nunits): New function,
13466 (vect_build_slp_tree_1): ...here.
13467 (vect_build_slp_tree_2): Call it for phis too.
13469 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13471 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13472 to vect_get_vec_def_for_operand when getting the mask operand.
13474 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13475 Alan Hayward <alan.hayward@arm.com>
13476 David Sherwood <david.sherwood@arm.com>
13478 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13481 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13482 Alan Hayward <alan.hayward@arm.com>
13483 David Sherwood <david.sherwood@arm.com>
13485 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13486 calculation for vector booleans.
13488 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13489 Alan Hayward <alan.hayward@arm.com>
13490 David Sherwood <david.sherwood@arm.com>
13492 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13494 (vect_transform_stmt): ...here.
13495 (vect_analyze_stmt): Use it instead of calling
13496 vectorizable_live_operation directly.
13498 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13500 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13501 non-SIMT targets in acc vector loops.
13503 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13505 * configure.ac: Add arc and check if assembler supports gdwarf2.
13506 * configure: Regenerate.
13508 2017-09-18 Richard Biener <rguenther@suse.de>
13510 PR tree-optimization/82220
13511 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13512 epilogue niters from the min_profitable_iters compute.
13514 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13517 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13518 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13519 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13521 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13524 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13525 switching to a new text section.
13527 2017-09-18 Richard Biener <rguenther@suse.de>
13529 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13531 (build_alias_set): Reject aliases with no access function.
13533 2017-09-18 Richard Biener <rguenther@suse.de>
13535 PR tree-optimization/79622
13536 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13538 (build_cross_bb_scalars_use): Likewise.
13540 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13542 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13544 2017-09-18 Alan Modra <amodra@gmail.com>
13547 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13548 stack_pointer_rtx for count 0. Update comments. Break up
13549 large rtl expression.
13551 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13553 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13554 Increase to 20 bytes.
13555 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13556 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13557 or avx version of the stub.
13559 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13562 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13563 compute the minimum stack alignment. Also update preferred stack
13564 boundary for leaf functions.
13566 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13568 PR tree-optimization/82228
13569 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13572 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13574 * common/config/nds32/nds32-common.c
13575 (nds32_option_optimization_table): Refine formatting.
13576 (nds32_option_optimization_table): Use -fsched-pressure and
13577 -fomit-frame-pointer for specific optimization level.
13579 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13581 * config/nds32/nds32.c: Refine formatting and comments.
13582 * config/nds32/nds32.h: Likewise.
13583 * config/nds32/nds32.md: Likewise.
13584 * config/nds32/nds32-cost.c: Likewise.
13585 * config/nds32/nds32-isr.c: Likewise.
13586 * config/nds32/nds32-md-auxiliary.c: Likewise.
13587 * config/nds32/nds32-multiple.md: Likewise.
13588 * config/nds32/nds32-predicates.c: Likewise.
13590 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13591 Jakub Jelinek <jakub@redhat.com>
13593 Add support for -std=c++2a.
13594 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13596 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13598 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13601 * doc/extend.texi (Common Function Attributes): Add
13602 references to ARM, AArch64, and S/390 specific attributes.
13603 (Function Specific Option Pragmas): Add AArch64 and S/390
13604 to list of back ends that support the target pragma.
13606 2017-09-15 Nathan Sidwell <nathan@acm.org>
13608 * doc/standards.texi: Fix C++17 description. Update URLs for
13611 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13613 * common.opt (Wcast-align=strict): New warning option.
13614 * doc/invoke.texi: Document -Wcast-align=strict.
13616 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13618 * cgraph.h (cgraph_thunk_info): Add comments.
13619 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13620 assert for VIRTUAL_* arguments stricter.
13622 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13624 PR tree-optimization/71026
13625 * match.pd: Move RDIV patterns from fold-const.c
13626 * fold-const.c (distribute_real_division): Removed.
13627 (fold_binary_loc): Remove calls to distribute_real_divison.
13629 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13631 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13632 c++1z and gnu++1z as deprecated. Change other references to
13633 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13634 Change -Wc++1z-compat to -Wc++17-compat.
13635 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13636 * dwarf2out.c (highest_c_language): Handle C++17.
13637 (gen_compile_unit_die): Likewise.
13639 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13641 PR rtl-optimization/82192
13642 * combine.c (make_extraction): Don't look through non-paradoxical
13643 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13646 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13647 Alan Hayward <alan.hayward@arm.com>
13648 David Sherwood <david.sherwood@arm.com>
13650 * target.def (function_arg_offset): New hook.
13651 * targhooks.h (default_function_arg_offset): Declare.
13652 * targhooks.c (default_function_arg_offset): New function.
13653 * function.c (locate_and_pad_parm): Use
13654 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13655 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13656 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13657 * doc/tm.texi: Regenerate.
13658 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13659 * config/spu/spu.c (spu_function_arg_offset): New function.
13660 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13661 * system.h (FUNCTION_ARG_OFFSET): Poison.
13663 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13664 Alan Hayard <alan.hayward@arm.com>
13665 David Sherwood <david.sherwood@arm.com>
13667 * target.def (truly_noop_truncation): New hook.
13668 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13669 than TRULY_NOOP_TRUNCATION.
13670 * hooks.h (hook_bool_uint_uint_true): Declare.
13671 * hooks.c (hook_bool_uint_uint_true): New function.
13672 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13673 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13674 * doc/tm.texi: Regenerate.
13675 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13676 rather than TRULY_NOOP_TRUNCATION in comments.
13677 (simplify_comparison): Likewise.
13678 (record_truncated_value): Likewise.
13679 * expmed.c (extract_bit_field_1): Likewise.
13680 (extract_split_bit_field): Likewise.
13681 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13682 instead of TRULY_NOOP_TRUNCATION.
13683 * function.c (assign_parm_setup_block): Likewise.
13684 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13685 * rtlhooks.c: Include target.h.
13686 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13687 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13688 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13689 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13690 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13691 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13692 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13693 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13694 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13695 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13696 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13697 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13698 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13699 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13700 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13701 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13702 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13703 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13704 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13705 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13706 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13707 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13708 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13709 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13710 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13711 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13712 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13713 rather than TRULY_NOOP_TRUNCATION in comments.
13714 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13715 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13716 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13717 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13718 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13719 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13720 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13721 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13722 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13723 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13724 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13725 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13726 rather than TRULY_NOOP_TRUNCATION in comments.
13727 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13728 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13729 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13730 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13731 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13732 TRULY_NOOP_TRUNCATION condition.
13733 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13734 (TRULY_NOOP_TRUNCATION): Delete.
13735 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13736 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13737 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13738 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13739 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13740 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13741 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13742 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13743 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13744 rather than TRULY_NOOP_TRUNCATION in comments.
13745 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13746 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13747 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13748 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13749 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13750 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13752 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13755 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13756 (*cmp_ior): Likewise.
13757 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13758 (*ior_scc_scc_cmp): Likewise.
13759 (*and_scc_scc): Likewise.
13760 (*and_scc_scc_cmp): Likewise.
13762 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13763 Alan Hayard <alan.hayward@arm.com>
13764 David Sherwood <david.sherwood@arm.com>
13766 * target.def (can_change_mode_class): New hook.
13767 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13768 (hard_regno_nregs): Likewise.
13769 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13770 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13771 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13772 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13773 (LOAD_EXTEND_OP): Update accordingly.
13774 * doc/tm.texi: Regenerate.
13775 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13776 CANNOT_CHANGE_MODE_CLASS.
13777 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13778 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13779 * combine.c (simplify_set): Update accordingly.
13780 * emit-rtl.c (validate_subreg): Likewise.
13781 * recog.c (general_operand): Likewise.
13782 * regcprop.c (mode_change_ok): Likewise.
13783 * reload1.c (choose_reload_regs): Likewise.
13784 (inherit_piecemeal_p): Likewise.
13785 * rtlanal.c (simplify_subreg_regno): Likewise.
13786 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13787 instead of CANNOT_CHANGE_MODE_CLASS.
13788 (reload_cse_simplify_operands): Likewise.
13789 * reload.c (push_reload): Use targetm.can_change_mode_class
13790 instead of CANNOT_CHANGE_MODE_CLASS.
13791 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13792 REG_CANNOT_CHANGE_MODE_P.
13793 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13794 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13795 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13796 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13797 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13798 (arm_can_change_mode_class): New function.
13799 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13800 than CANNOT_CHANGE_MODE_CLASS in comments.
13801 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13802 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13803 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13804 (ix86_can_change_mode_class): ...this new function, inverting the
13805 sense of the return value.
13806 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13807 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13808 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13809 (ia64_can_change_mode_class): New function.
13810 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13811 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13812 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13813 (m32c_can_change_mode_class): ...this new function, inverting the
13814 sense of the return value.
13815 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13816 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13817 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13818 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13819 (mips_can_change_mode_class): ...this new function, inverting the
13820 sense of the return value.
13821 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13822 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13823 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13824 (msp430_can_change_mode_class): New function.
13825 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13826 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13827 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13828 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13829 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13830 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13831 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13832 (pa_cannot_change_mode_class): Replace with...
13833 (pa_can_change_mode_class): ...this new function, inverting the
13834 sense of the return value.
13835 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13836 than CANNOT_CHANGE_MODE_CLASS in comments.
13837 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13838 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13839 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13840 (pdp11_cannot_change_mode_class): Replace with...
13841 (pdp11_can_change_mode_class): ...this new function, inverting the
13842 sense of the return value.
13843 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13844 * config/powerpcspe/powerpcspe-protos.h
13845 (rs6000_cannot_change_mode_class_ptr): Delete.
13846 * config/powerpcspe/powerpcspe.c
13847 (rs6000_cannot_change_mode_class_ptr): Delete.
13848 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13849 (rs6000_option_override_internal): Assign to
13850 targetm.can_change_mode_class instead of
13851 rs6000_cannot_change_mode_class_ptr.
13852 (rs6000_cannot_change_mode_class): Replace with...
13853 (rs6000_can_change_mode_class): ...this new function, inverting the
13854 sense of the return value.
13855 (rs6000_debug_cannot_change_mode_class): Replace with...
13856 (rs6000_debug_can_change_mode_class): ...this new function.
13857 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13858 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13859 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13860 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13861 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13863 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13864 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13865 (rs6000_option_override_internal): Assign to
13866 targetm.can_change_mode_class instead of
13867 rs6000_cannot_change_mode_class_ptr.
13868 (rs6000_cannot_change_mode_class): Replace with...
13869 (rs6000_can_change_mode_class): ...this new function, inverting the
13870 sense of the return value.
13871 (rs6000_debug_cannot_change_mode_class): Replace with...
13872 (rs6000_debug_can_change_mode_class): ...this new function.
13873 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13874 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13875 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13876 (s390_can_change_mode_class): ...this new function, inverting the
13877 sense of the return value.
13878 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13879 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13880 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13881 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13882 (sh_cannot_change_mode_class): Replace with...
13883 (sh_can_change_mode_class): ...this new function, inverting the
13884 sense of the return value.
13885 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13886 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13887 (sparc_can_change_mode_class): New function.
13888 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13889 * config/spu/spu.c (spu_can_change_mode_class): New function.
13890 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13891 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13892 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13893 (visium_can_change_mode_class): New function.
13894 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13896 2017-09-15 Richard Biener <rguenther@suse.de>
13898 PR tree-optimization/82217
13899 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13900 but not undefined case.
13902 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13905 * postreload.c (reload_cse_simplify_operands): Skip
13906 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13908 2017-09-15 Richard Biener <rguenther@suse.de>
13910 PR tree-optimization/68823
13911 * graphite-scop-detection.c (build_alias_set): If we have a
13912 possible dependence check whether we can handle them by just
13913 looking at the DRs DR_ACCESS_FNs.
13914 (build_scops): If build_alias_set fails, fail the SCOP.
13916 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13918 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13919 to support float128 built-in functions that require the ISA 3.0
13921 (BU_FLOAT128_3_HW): Likewise.
13922 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13923 built-in functions.
13924 (FMAF128): Likewise.
13926 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13927 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13928 floating point instructions.
13929 (rs6000_invalid_builtin): Likewise.
13930 (rs6000_builtin_mask_names): Likewise.
13931 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13932 (RS6000_BTM_FLOAT128_HW): Likewise.
13933 (RS6000_BTM_COMMON): Likewise.
13934 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13936 * doc/extend.texi (RS/6000 built-in functions): Document the
13937 IEEE 128-bit floating point square root and fused multiply-add
13938 built-in functions.
13940 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13942 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13943 reg (r2) isn't in the set of registers defined in the prologue.
13945 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13946 Alan Hayward <alan.hayward@arm.com>
13947 David Sherwood <david.sherwood@arm.com>
13949 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13950 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13951 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13952 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13953 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13955 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13956 max_vectorization_factor.
13957 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13959 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13960 Alan Hayward <alan.hayward@arm.com>
13961 David Sherwood <david.sherwood@arm.com>
13963 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13964 (vect_worthwhile_without_simd_p): Declare.
13965 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13966 (vectorizable_reduction): Use it.
13967 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13968 (vectorizable_operation): Likewise.
13970 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13971 Alan Hayward <alan.hayward@arm.com>
13972 David Sherwood <david.sherwood@arm.com>
13974 * tree-vectorizer.h (vect_get_num_copies): New function.
13975 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13976 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13977 (vectorizable_induction): Likewise.
13978 (vectorizable_live_operation): Likewise.
13979 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13980 (vectorizable_bswap): Likewise.
13981 (vectorizable_call): Likewise.
13982 (vectorizable_conversion): Likewise.
13983 (vectorizable_assignment): Likewise.
13984 (vectorizable_shift): Likewise.
13985 (vectorizable_operation): Likewise.
13986 (vectorizable_store): Likewise.
13987 (vectorizable_load): Likewise.
13988 (vectorizable_condition): Likewise.
13989 (vectorizable_comparison): Likewise.
13990 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13992 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13993 Alan Hayward <alan.hayward@arm.com>
13994 David Sherwood <david.sherwood@arm.com>
13996 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13997 of vect_init_vector.
13999 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14000 Alan Hayward <alan.hayward@arm.com>
14001 David Sherwood <david.sherwood@arm.com>
14003 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14004 an inline wrapper that provides a location.
14005 (gimple_build_vector): Likewise.
14006 * gimple-fold.c (gimple_build_vector_from_val): New function.
14007 (gimple_build_vector): Likewise.
14008 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14009 functions to build the initial value. Always return a gimple value.
14010 (get_initial_defs_for_reduction): Likewise. Only compute
14012 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14013 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14014 (vectorizable_induction): Use gimple_build_vector rather than
14017 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14018 Alan Hayward <alan.hayward@arm.com>
14019 David Sherwood <david.sherwood@arm.com>
14021 * target.h (vec_perm_indices): New typedef.
14022 (auto_vec_perm_indices): Likewise.
14023 * optabs-query.h: Include target.h
14024 (can_vec_perm_p): Take a vec_perm_indices *.
14025 * optabs-query.c (can_vec_perm_p): Likewise.
14026 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14027 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14028 * tree-vect-generic.c (lower_vec_perm): Likewise.
14029 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14030 (vect_grouped_load_supported): Likewise.
14031 (vect_shift_permute_load_chain): Likewise.
14032 (vect_permute_store_chain): Use auto_vec_perm_indices.
14033 (vect_permute_load_chain): Likewise.
14034 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14035 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14036 Update uses of can_vec_perm_p.
14037 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14038 mode with a number of elements. Take a vec_perm_indices *.
14039 (vect_create_epilog_for_reduction): Update accordingly.
14040 Use auto_vec_perm_indices.
14041 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14042 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14043 (vect_transform_slp_perm_load): Likewise.
14044 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14045 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14046 (vect_gen_perm_mask_checked): Likewise.
14047 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14048 (vect_gen_perm_mask_checked): Likewise.
14049 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14050 (vectorizable_store): Likewise.
14051 (vectorizable_load): Likewise.
14052 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14053 (vectorizable_bswap): Likewise.
14055 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14056 Alan Hayward <alan.hayward@arm.com>
14057 David Sherwood <david.sherwood@arm.com>
14059 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14060 * tree.c (build_vector): Likewise.
14061 (build_vector_from_ctor): Update accordingly.
14062 (build_vector_from_val): Likewise.
14063 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14064 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14065 * tree-vect-generic.c (add_rshift): Likewise.
14066 (expand_vector_divmod): Likewise.
14067 (optimize_vector_constructor): Likewise.
14068 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14069 (vect_transform_slp_perm_load): Likewise.
14070 (vect_schedule_slp_instance): Likewise.
14071 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14072 (vectorizable_call): Likewise.
14073 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14074 * expmed.c (make_tree): Likewise.
14075 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14076 a vector passed to build_vector.
14077 (fold_convert_const): Likewise.
14078 (exact_inverse): Likewise.
14079 (fold_ternary_loc): Likewise.
14080 (fold_relational_const): Likewise.
14081 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14082 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14083 (const_unop): Likewise. Store the reduction accumulator in a
14084 variable rather than an array.
14085 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14086 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14087 the new vector, rather than constructing it after the input arrays.
14088 (native_interpret_vector): Use auto_vec<tree> when building
14089 a vector passed to build_vector. Add elements in order.
14090 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14091 auto_vec<tree> when building a vector passed to build_vector.
14092 (vect_create_epilog_for_reduction): Likewise.
14093 (vectorizable_induction): Likewise.
14094 (get_initial_def_for_reduction): Likewise. Fix indentation of
14096 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14098 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14099 passed to build_vector.
14101 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14102 Alan Hayward <alan.hayward@arm.com>
14103 David Sherwood <david.sherwood@arm.com>
14105 * tree-core.h (tree_base::u): Add an "nelts" field.
14106 (tree_vector): Use VECTOR_CST_NELTS as the length.
14107 * tree.c (tree_size): Likewise.
14108 (make_vector): Initialize VECTOR_CST_NELTS.
14109 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14110 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14111 TYPE_VECTOR_SUBPARTS.
14112 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14113 as the number of units, setting it from VECTOR_CST_NELTS.
14114 (const_vector_from_tree): Likewise.
14115 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14116 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14117 (fold_negate_expr_1): Likewise.
14118 (fold_convert_const): Likewise.
14119 (const_binop): Likewise. Differentiate the number of output and
14121 (const_unop): Likewise.
14122 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14123 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14124 in cases that did the opposite.
14126 2017-09-14 Richard Biener <rguenther@suse.de>
14128 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14131 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14133 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14135 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14138 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14139 if and where to split a bb, except for splitting before debug insn
14140 sequences followed by non-label real insn. Delete debug insns
14141 in between basic blocks.
14143 * combine.c (make_compound_operation_int): Formatting fixes.
14145 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14146 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14147 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14148 * config/sol2.h (LINK_EH_SPEC): Likewise.
14149 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14150 * config/s390/linux.h (LINK_SPEC): Likewise.
14151 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14152 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14153 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14154 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14155 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14156 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14157 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14158 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14160 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14162 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14164 (ENDFILE_LINUX_SPEC): Likewise.
14165 (LINK_EH_SPEC): Likewise.
14166 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14167 (LINK_OS_LINUX_SPEC32): Likewise.
14168 (LINK_OS_LINUX_SPEC64): Likewise.
14169 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14170 (LINK_OS_LINUX_SPEC): Likewise.
14172 2017-09-13 Martin Liska <mliska@suse.cz>
14174 PR middle-end/82154
14175 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14176 CASE_HIGH is NULL_TREE.
14178 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14179 Alan Hayward <alan.hayward@arm.com>
14180 David Sherwood <david.sherwood@arm.com>
14182 * target.def (secondary_memory_needed): New hook.
14183 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14184 instead of SECONDARY_MEMORY_NEEDED.
14185 (secondary_memory_needed_mode): Likewise.
14186 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14187 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14188 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14189 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14190 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14191 * doc/tm.texi: Regenerate.
14192 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14193 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14194 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14195 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14196 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14197 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14198 mode argument first and change the reg_class arguments to reg_class_t.
14199 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14200 Make static. Update the call to inline_secondary_memory_needed.
14201 (ix86_register_move_cost): Update the call to
14202 inline_secondary_memory_needed.
14203 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14204 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14206 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14207 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14209 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14210 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14211 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14212 and match hook interface. Add comment from mips.h.
14213 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14214 * config/mmix/mmix.md (truncdfsf2): Refer to
14215 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14217 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14218 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14219 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14220 (pa_secondary_memory_needed): New function.
14221 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14222 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14223 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14224 (pdp11_secondary_memory_needed): Make static and match hook interface.
14225 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14226 * config/powerpcspe/powerpcspe-protos.h
14227 (rs6000_secondary_memory_needed_ptr): Delete.
14228 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14230 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14231 (rs6000_option_override_internal): Assign to
14232 targetm.secondary_memory_needed rather than
14233 rs6000_secondary_memory_needed_ptr.
14234 (rs6000_secondary_memory_needed): Match hook interface.
14235 (rs6000_debug_secondary_memory_needed): Likewise.
14236 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14237 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14238 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14239 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14240 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14241 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14243 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14244 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14245 (rs6000_option_override_internal): Assign to
14246 targetm.secondary_memory_needed rather than
14247 rs6000_secondary_memory_needed_ptr.
14248 (rs6000_secondary_memory_needed): Match hook interface.
14249 (rs6000_debug_secondary_memory_needed): Likewise.
14250 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14251 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14252 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14253 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14254 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14255 (sparc_secondary_memory_needed): New function.
14256 * lra-constraints.c (check_and_process_move): Refer to
14257 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14259 (curr_insn_transform): Likewise.
14260 (process_alt_operands): Use targetm.secondary_memory_needed
14261 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14262 (check_secondary_memory_needed_p): Likewise.
14263 (choose_split_class): Likewise.
14264 * reload.c: Unconditionally include code that was previously
14265 conditional on SECONDARY_MEMORY_NEEDED.
14266 (push_secondary_reload): Use targetm.secondary_memory_needed
14267 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14268 (push_reload): Likewise.
14269 * reload1.c: Unconditionally include code that was previously
14270 conditional on SECONDARY_MEMORY_NEEDED.
14271 (choose_reload_regs): Use targetm.secondary_memory_needed
14272 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14273 (gen_reload): Likewise.
14274 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14276 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14277 Alan Hayward <alan.hayward@arm.com>
14278 David Sherwood <david.sherwood@arm.com>
14280 * target.def (secondary_memory_needed_mode): New hook:
14281 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14282 * targhooks.h (default_secondary_memory_needed_mode): New function.
14283 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14284 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14285 * doc/tm.texi: Regenerate.
14286 * lra-constraints.c (check_and_process_move): Use
14287 targetm.secondary_memory_needed_mode instead of
14288 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14289 (curr_insn_transform): Likewise.
14290 * reload.c (get_secondary_mem): Likewise.
14291 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14292 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14294 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14295 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14296 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14297 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14298 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14300 * config/powerpcspe/powerpcspe-protos.h
14301 (rs6000_secondary_memory_needed_mode): Delete.
14302 * config/powerpcspe/powerpcspe.c
14303 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14304 (rs6000_secondary_memory_needed_mode): Make static.
14305 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14306 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14308 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14310 (rs6000_secondary_memory_needed_mode): Make static.
14311 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14312 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14313 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14314 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14315 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14317 (sparc_secondary_memory_needed_mode): New function.
14318 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14320 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14322 * config/aarch64/constraints.md (Umq): New constraint.
14323 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14325 (mov<mode>): Update condition.
14327 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14329 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14330 when bitposition is the same.
14332 2017-09-13 Richard Biener <rguenther@suse.de>
14334 * dwarf2out.c (output_die_symbol): Remove.
14335 (output_die): Do not output a DIEs symbol.
14337 2017-09-13 Richard Biener <rguenther@suse.de>
14339 PR middle-end/82128
14340 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14341 default-def to avoid breaking iterator update with the weird
14342 interaction with cgraph_update_edges_for_call_stmt_node.
14344 2017-09-13 Richard Biener <rguenther@suse.de>
14346 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14347 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14348 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14349 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14351 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14353 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14354 Disable pc relative literal load irrespective of
14355 TARGET_FIX_ERR_A53_84341 for default.
14357 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14359 * config/sparc/sparc.c (output_return): Output the source location of
14360 the insn in the delay slot, if any.
14361 (output_sibcall): Likewise.
14363 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14366 * common.opt (-static-pie): New alias.
14367 (shared): Negate static-pie.
14368 (-no-pie): Update help text.
14370 (static-pie): New option.
14371 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14372 -static-pie support.
14373 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14374 (LINK_EH_SPEC): Likewise.
14375 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14376 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14377 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14378 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14379 (init_gcc_specs): Likewise.
14380 (init_spec): Likewise.
14381 (display_help): Update help message for -pie.
14382 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14385 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14387 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14388 (movdi_aarch64): Likewise.
14389 (movti_aarch64): Likewise.
14391 2017-09-12 Simon Wright <simon@pushface.org>
14394 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14395 calculation of the minor version, always output as 0.
14397 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14400 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14401 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14402 on it early, rather than manual conversion late. For
14403 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14404 instead of performing manual conversion.
14406 2017-09-12 Carl Love <cel@us.ibm.com>
14408 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14409 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14411 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14412 VMULOSW): Add definitions.
14413 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14414 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14415 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14416 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14417 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14419 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14421 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14423 (movti_aarch64): Likewise.
14424 (movdf_aarch64): Likewise.
14425 (movtf_aarch64): Likewise.
14426 (load_pairdi): Likewise.
14427 (store_pairdi): Likewise.
14428 (load_pairdf): Likewise.
14429 (store_pairdf): Likewise.
14430 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14431 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14432 (ldr_got_small_<mode>): Likewise.
14433 (ldr_got_small_28k_<mode>): Likewise.
14434 (ldr_got_tiny): Likewise.
14435 * config/aarch64/iterators.md (ldst_sz): New.
14436 (ldpstp_sz): Likewise.
14437 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14439 (thunderx_load): Split load_8 to load_16.
14440 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14442 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14443 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14444 (xgene1_store_pair): Split store_8 to store_16.
14445 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14446 (falkor_st_0_st_sd): Split store_8 to store_16.
14448 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14450 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14451 and store1/2/3/4 to store_4/8/12/16.
14452 * config/aarch64/aarch64.md: Update for rename.
14453 * config/arm/arm.md: Likewise.: Likewise.
14454 * config/arm/arm.c: Likewise.
14455 * config/arm/thumb1.md: Likewise.
14456 * config/arm/thumb2.md: Likewise.
14457 * config/arm/vfp.md: Likewise.
14458 * config/arm/arm-generic.md: Likewise.
14459 * config/arm/arm1020e.md: Likewise.
14460 * config/arm/arm1026ejs.md: Likewise.
14461 * config/arm/arm1136jfs.md: Likewise.
14462 * config/arm/arm926ejs.md: Likewise.
14463 * config/arm/cortex-a15.md: Likewise.
14464 * config/arm/cortex-a17.md: Likewise.
14465 * config/arm/cortex-a5.md: Likewise.
14466 * config/arm/cortex-a53.md: Likewise.
14467 * config/arm/cortex-a57.md: Likewise.
14468 * config/arm/cortex-a7.md: Likewise.
14469 * config/arm/cortex-a8.md: Likewise.
14470 * config/arm/cortex-a9.md: Likewise.
14471 * config/arm/cortex-m4.md: Likewise.
14472 * config/arm/cortex-m7.md: Likewise.
14473 * config/arm/cortex-r4.md: Likewise.
14474 * config/arm/exynos-m1.md: Likewise.
14475 * config/arm/fa526.md: Likewise.
14476 * config/arm/fa606te.md: Likewise.
14477 * config/arm/fa626te.md: Likewise.
14478 * config/arm/fa726te.md: Likewise.
14479 * config/arm/fmp626.md: Likewise.
14480 * config/arm/iwmmxt.md: Likewise.
14481 * config/arm/ldmstm.md: Likewise.
14482 * config/arm/marvell-pj4.md: Likewise.
14483 * config/arm/xgene1.md: Likewise.
14484 * config/aarch64/thunderx.md: Likewise.
14485 * config/aarch64/thunderx2t99.md: Likewise.
14486 * config/aarch64/falkor.md: Likewise.
14488 2017-09-12 Martin Liska <mliska@suse.cz>
14490 * attribs.c (private_lookup_attribute): New function.
14491 * attribs.h (private_lookup_attribute): Declared here.
14492 (lookup_attribute): Called from this place.
14494 2017-09-12 Richard Biener <rguenther@suse.de>
14496 PR tree-optimization/82157
14497 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14498 stmts with side-effects.
14500 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14501 Alan Hayward <alan.hayward@arm.com>
14502 David Sherwood <david.sherwood@arm.com>
14504 * target.def (hard_regno_nregs): New hook.
14505 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14506 * targhooks.h (default_hard_regno_nregs): Declare.
14507 * targhooks.c (default_hard_regno_nregs): New function.
14508 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14509 (TARGET_HARD_REGNO_NREGS): ...this hook.
14510 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14511 (CLASS_MAX_NREGS): Likewise.
14512 * doc/tm.texi: Regenerate.
14513 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14514 instead of HARD_REGNO_NREGS.
14515 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14516 HARD_REGNO_NREGS in the comment.
14517 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14518 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14519 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14520 Return an unsigned int.
14521 (TARGET_HARD_REGNO_NREGS): Redefine.
14522 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14523 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14524 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14525 (arc_hard_regno_nregs): New function.
14526 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14527 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14528 (arm_hard_regno_nregs): New function.
14529 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14530 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14531 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14532 (TARGET_HARD_REGNO_NREGS): Redefine.
14533 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14534 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14535 (HARD_REGNO_NREGS): Delete.
14536 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14537 (cr16_hard_regno_nregs): New function.
14538 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14539 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14540 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14541 (cris_hard_regno_nregs): New function.
14542 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14543 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14544 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14545 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14546 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14547 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14548 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14549 (frv_hard_regno_nregs): Make static. Take and return an
14551 (frv_class_max_nregs): Remove outdated copy of documentation.
14552 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14553 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14554 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14555 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14556 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14557 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14558 (TARGET_HARD_REGNO_NREGS): Redefine.
14559 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14560 (CLASS_MAX_NREGS): Update comment.
14561 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14562 (ia64_hard_regno_nregs): New function.
14563 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14564 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14565 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14566 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14567 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14569 (m32c_hard_regno_nregs): Likewise. Make static.
14570 (TARGET_HARD_REGNO_NREGS): Redefine.
14571 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14572 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14573 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14574 (m68k_hard_regno_nregs): New function.
14575 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14576 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14577 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14578 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14579 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14580 Take and return an unsigned int.
14581 (TARGET_HARD_REGNO_NREGS): Redefine.
14582 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14583 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14584 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14585 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14586 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14587 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14588 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14589 (msp430_hard_regno_nregs): Make static. Take and return an
14591 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14592 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14593 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14594 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14595 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14596 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14597 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14598 (TARGET_HARD_REGNO_NREGS): Redefine.
14599 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14600 (PA_HARD_REGNO_NREGS): ...this.
14601 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14602 (PA_HARD_REGNO_NREGS): ...this.
14603 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14604 (pa_hard_regno_nregs): New function.
14605 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14606 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14607 (pdp11_hard_regno_nregs): New function.
14608 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14609 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14610 (rs6000_hard_regno_nregs_hook): New function.
14611 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14612 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14613 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14614 Take and return an unsigned int. Move earlier in file.
14615 (TARGET_HARD_REGNO_NREGS): Redefine.
14616 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14617 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14618 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14619 (rl78_hard_regno_nregs): Make static. Take and return an
14621 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14622 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14623 (rs6000_hard_regno_nregs_hook): New function.
14624 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14625 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14626 (TARGET_HARD_REGNO_NREGS): Redefine.
14627 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14628 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14629 instead of HARD_REGNO_NREGS.
14630 (s390_hard_regno_nregs): New function.
14631 (s390_hard_regno_mode_ok): Add comment from s390.h.
14632 (TARGET_HARD_REGNO_NREGS): Redefine.
14633 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14634 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14635 (sh_hard_regno_nregs): New function.
14636 (sh_pass_in_reg_p): Use it.
14637 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14638 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14639 (sparc_hard_regno_nregs): New function.
14640 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14641 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14642 (spu_function_arg_advance): Use it, supplying a valid register number.
14643 (TARGET_HARD_REGNO_NREGS): Redefine.
14644 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14645 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14646 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14647 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14648 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14649 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14650 (CLASS_MAX_NREGS): Remove copy of old documentation.
14651 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14652 (visium_hard_regno_nregs): New function.
14653 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14654 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14655 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14656 xtensa_hard_regno_nregs): New function.
14657 * system.h (HARD_REGNO_NREGS): Poison.
14659 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14661 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14662 hard_regno_nregs instead of HARD_REGNO_NREGS.
14663 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14664 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14665 (c6x_expand_epilogue): Likewise.
14666 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14667 (frv_read_iacc_argument): Likewise.
14668 * config/sh/sh.c: Include regs.h.
14669 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14670 (regs_used): Likewise.
14671 (output_stack_adjust): Likewise.
14672 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14673 * expmed.c: Include regs.h.
14674 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14675 * ree.c: Include regs.h.
14676 (combine_reaching_defs): Use hard_regno_nregs instead of
14678 (add_removable_extension): Likewise.
14680 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14682 * regs.h (hard_regno_nregs): Turn into a function.
14683 (end_hard_regno): Update accordingly.
14684 * caller-save.c (setup_save_areas): Likewise.
14685 (save_call_clobbered_regs): Likewise.
14686 (replace_reg_with_saved_mem): Likewise.
14687 (insert_restore): Likewise.
14688 (insert_save): Likewise.
14689 * combine.c (can_change_dest_mode): Likewise.
14690 (move_deaths): Likewise.
14691 (distribute_notes): Likewise.
14692 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14693 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14694 (rs6000_split_multireg_move): Likewise.
14695 (rs6000_register_move_cost): Likewise.
14696 (rs6000_memory_move_cost): Likewise.
14697 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14698 (rs6000_split_multireg_move): Likewise.
14699 (rs6000_register_move_cost): Likewise.
14700 (rs6000_memory_move_cost): Likewise.
14701 * cselib.c (cselib_reset_table): Likewise.
14702 (cselib_lookup_1): Likewise.
14703 * emit-rtl.c (set_mode_and_regno): Likewise.
14704 * function.c (aggregate_value_p): Likewise.
14705 * ira-color.c (setup_profitable_hard_regs): Likewise.
14706 (check_hard_reg_p): Likewise.
14707 (calculate_saved_nregs): Likewise.
14708 (assign_hard_reg): Likewise.
14709 (improve_allocation): Likewise.
14710 (calculate_spill_cost): Likewise.
14711 * ira-emit.c (modify_move_list): Likewise.
14712 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14713 (ira_hard_reg_in_set_p): Likewise.
14714 * ira.c (setup_reg_mode_hard_regset): Likewise.
14715 (clarify_prohibited_class_mode_regs): Likewise.
14716 (check_allocation): Likewise.
14717 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14718 (lra_setup_reg_renumber): Likewise.
14719 (setup_try_hard_regno_pseudos): Likewise.
14720 (spill_for): Likewise.
14721 (assign_hard_regno): Likewise.
14722 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14723 * lra-constraints.c (in_class_p): Likewise.
14724 (lra_constraint_offset): Likewise.
14725 (simplify_operand_subreg): Likewise.
14726 (lra_constraints): Likewise.
14727 (split_reg): Likewise.
14728 (split_if_necessary): Likewise.
14729 (invariant_p): Likewise.
14730 (inherit_in_ebb): Likewise.
14731 * lra-lives.c (process_bb_lives): Likewise.
14732 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14733 (get_hard_regs): Likewise.
14734 (do_remat): Likewise.
14735 * lra-spills.c (assign_spill_hard_regs): Likewise.
14736 * mode-switching.c (create_pre_exit): Likewise.
14737 * postreload.c (reload_combine_recognize_pattern): Likewise.
14738 * recog.c (peep2_find_free_register): Likewise.
14739 * regcprop.c (kill_value_regno): Likewise.
14740 (set_value_regno): Likewise.
14741 (copy_value): Likewise.
14742 (maybe_mode_change): Likewise.
14743 (find_oldest_value_reg): Likewise.
14744 (copyprop_hardreg_forward_1): Likewise.
14745 * regrename.c (check_new_reg_p): Likewise.
14746 (regrename_do_replace): Likewise.
14747 * reload.c (push_reload): Likewise.
14748 (combine_reloads): Likewise.
14749 (find_dummy_reload): Likewise.
14750 (operands_match_p): Likewise.
14751 (find_reloads): Likewise.
14752 (find_equiv_reg): Likewise.
14753 (reload_adjust_reg_for_mode): Likewise.
14754 * reload1.c (count_pseudo): Likewise.
14755 (count_spilled_pseudo): Likewise.
14756 (find_reg): Likewise.
14757 (clear_reload_reg_in_use): Likewise.
14758 (free_for_value_p): Likewise.
14759 (allocate_reload_reg): Likewise.
14760 (choose_reload_regs): Likewise.
14761 (reload_adjust_reg_for_temp): Likewise.
14762 (emit_reload_insns): Likewise.
14763 (delete_output_reload): Likewise.
14764 * rtlanal.c (subreg_get_info): Likewise.
14765 * sched-deps.c (sched_analyze_reg): Likewise.
14766 * sel-sched.c (init_regs_for_mode): Likewise.
14767 (mark_unavailable_hard_regs): Likewise.
14768 (choose_best_reg_1): Likewise.
14769 (verify_target_availability): Likewise.
14770 * valtrack.c (dead_debug_insert_temp): Likewise.
14771 * var-tracking.c (track_loc_p): Likewise.
14772 (emit_note_insn_var_location): Likewise.
14773 * varasm.c (make_decl_rtl): Likewise.
14774 * reginfo.c (choose_hard_reg_mode): Likewise.
14775 (init_reg_modes_target): Refer directly to
14776 this_target_regs->x_hard_regno_nregs.
14778 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14780 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14781 instead of hard_regno_nregs.
14783 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14785 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14786 end_hard_regno instead of hard_regno_nregs.
14787 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14788 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14789 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14790 * ira-color.c (improve_allocation): Likewise.
14791 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14792 * lra-lives.c (mark_regno_live): Likewise.
14793 (mark_regno_dead): Likewise.
14794 * lra-remat.c (operand_to_remat): Likewise.
14795 * lra.c (collect_non_operand_hard_regs): Likewise.
14796 * postreload.c (reload_combine_note_store): Likewise.
14797 (move2add_valid_value_p): Likewise.
14798 * reload.c (regno_clobbered_p): Likewise.
14800 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14802 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14804 * config/v850/v850.c (v850_reorg): Likewise.
14805 * reload.c (refers_to_regno_for_reload_p): Likewise.
14806 (find_equiv_reg): Likewise.
14807 * reload1.c (reload_reg_reaches_end_p): Likewise.
14809 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14811 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14813 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14814 * config/arm/arm.c (output_move_neon): Likewise.
14815 (arm_attr_length_move_neon): Likewise.
14816 (neon_split_vcombine): Likewise.
14817 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14818 (c6x_mark_reg_written): Likewise.
14819 (c6x_dwarf_register_span): Likewise.
14820 * config/i386/i386.c (ix86_save_reg): Likewise.
14821 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14822 (rws_access_reg): Likewise.
14823 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14824 * mode-switching.c (create_pre_exit): Likewise.
14825 * ree.c (combine_reaching_defs): Likewise.
14826 (add_removable_extension): Likewise.
14827 * regcprop.c (find_oldest_value_reg): Likewise.
14828 (copyprop_hardreg_forward_1): Likewise.
14829 * reload.c (reload_inner_reg_of_subreg): Likewise.
14830 (push_reload): Likewise.
14831 (combine_reloads): Likewise.
14832 (find_dummy_reload): Likewise.
14833 (reload_adjust_reg_for_mode): Likewise.
14834 * reload1.c (find_reload_regs): Likewise.
14835 (forget_old_reloads_1): Likewise.
14836 (reload_reg_free_for_value_p): Likewise.
14837 (reload_adjust_reg_for_temp): Likewise.
14838 (emit_reload_insns): Likewise.
14839 (delete_output_reload): Likewise.
14840 * sel-sched.c (choose_best_reg_1): Likewise.
14841 (choose_best_pseudo_reg): Likewise.
14843 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14844 Alan Hayward <alan.hayward@arm.com>
14845 David Sherwood <david.sherwood@arm.com>
14847 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14848 * target.def (slow_unaligned_access): New hook.
14849 * targhooks.h (default_slow_unaligned_access): Declare.
14850 * targhooks.c (default_slow_unaligned_access): New function.
14851 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14852 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14853 * doc/tm.texi: Regenerate.
14854 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14855 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14856 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14858 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14859 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14861 (rs6000_slow_unaligned_access): New function.
14862 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14863 (expand_block_compare): Likewise.
14864 (expand_strn_compare): Likewise.
14865 (rs6000_rtx_costs): Likewise.
14866 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14867 (riscv_slow_unaligned_access): Likewise.
14868 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14869 (riscv_slow_unaligned_access_p): ...this and make static.
14870 (riscv_option_override): Update accordingly.
14871 (riscv_slow_unaligned_access): New function.
14872 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14873 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14874 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14875 (rs6000_slow_unaligned_access): New function.
14876 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14877 (rs6000_rtx_costs): Likewise.
14878 * config/rs6000/rs6000-string.c (expand_block_compare)
14879 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14880 of SLOW_UNALIGNED_ACCESS.
14881 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14882 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14883 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14884 of SLOW_UNALIGNED_ACCESS.
14885 * expmed.c (simple_mem_bitfield_p): Likewise.
14886 * expr.c (alignment_for_piecewise_move): Likewise.
14887 (emit_group_load_1): Likewise.
14888 (emit_group_store): Likewise.
14889 (copy_blkmode_from_reg): Likewise.
14890 (emit_push_insn): Likewise.
14891 (expand_assignment): Likewise.
14892 (store_field): Likewise.
14893 (expand_expr_real_1): Likewise.
14894 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14895 * lra-constraints.c (simplify_operand_subreg): Likewise.
14896 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14897 * gimple-ssa-store-merging.c: Likewise in block comment at start
14899 * tree-ssa-strlen.c: Include target.h.
14900 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14901 of SLOW_UNALIGNED_ACCESS.
14902 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14904 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14906 PR rtl-optimization/82185
14907 * expmed.c (emit_store_flag_int): Only test tem if it has been
14910 2017-09-12 Richard Biener <rguenther@suse.de>
14912 PR middle-end/82149
14913 * match.pd ((FTYPE) N CMP CST): Fix typo.
14915 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14917 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14919 (mips_near_type_p): Add 'short_call' attribute as a synonym
14921 * doc/extend.texi (short_call): Document new function attribute.
14923 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14926 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14927 assertion check that in the condition.
14928 (get_atomic_generic_size): Likewise. Before testing if parameter
14929 has pointer type, if it has array type, call for C++
14930 default_conversion to perform array-to-pointer conversion.
14932 2017-09-12 Richard Biener <rguenther@suse.de>
14934 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14935 for operations we cannot scalarize.
14937 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14939 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14940 vectors heap vectors. Clean up comments.
14941 Make visited_bbs a reference.
14942 (profitable_jump_thread_path): Make GC
14943 vectors heap vectors. Clean up comments.
14945 (convert_and_register_jump_thread_path): Make GC vectors heap
14947 (check_subpath_and_update_thread_path): Same. Clean up comments.
14948 Make visited_bbs a reference.
14949 (handle_phi): Abstract common code to to
14950 register_jump_thread_path_if_profitable.
14951 Rename VAR_BB to DEF_BB.
14953 Make GC vectors heap vectors.
14954 Make visited_bbs a reference.
14955 (handle_assignment): Same.
14956 (register_jump_thread_path_if_profitable): New.
14957 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14959 Make GC vectors heap vectors. Clean up comments.
14960 Make visited_bbs a reference.
14961 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14962 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14965 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14968 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14969 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14971 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14973 Revert r251800 and r251799.
14975 2017-09-11 Martin Jambor <mjambor@suse.cz>
14978 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14979 arguments in advance.
14980 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14981 use it to find predecessor edges.
14982 (naive_outof_ssa): Collect vector of predecessors.
14984 2017-09-08 Jason Merrill <jason@redhat.com>
14986 PR c++/70029 - ICE with ref-qualifier and -flto
14987 * langhooks.h (struct lang_hooks_for_types): Add
14988 copy_lang_qualifiers.
14989 * attribs.c (build_type_attribute_qual_variant): Use it.
14990 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14992 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14993 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14995 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14998 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14999 (*mulsi3_sp64): New instruction.
15000 (mulsi3): New expander.
15002 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15004 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15006 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15008 * sancov.c: Include memmodel.h.
15010 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15013 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15016 2017-09-07 Carl Love <cel@us.ibm.com>
15018 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15019 the sldi instruction.
15021 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15023 * sancov.c: Include tm_p.h.
15025 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15028 * output.h (switch_to_other_text_partition): New declaration.
15029 * varasm.c (switch_to_other_text_partition): New function.
15030 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15031 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15032 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15033 to the other text partition before emitting LCL label and switch back
15034 after emitting the word after it.
15036 2017-09-07 Richard Biener <rguenther@suse.de>
15038 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15039 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15040 critical edges here, after loop init.
15041 (pass_data_pre): Remove PROP_no_crit_edges flags.
15042 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15043 for valueization of call args to avoid leaking VN_TOP.
15044 (visit_use): Assert we do not visit default defs.
15045 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15046 Use error_mark_node to more easily detect leaking VN_TOP.
15047 All default-defs are varying, not VN_TOP. Mark them visited.
15048 (run_scc_vn): Make code match comment.
15050 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15052 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15053 OPTION_MASK_FLOAT128_KEYWORD.
15054 (POWERPC_MASKS): Likewise.
15055 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15056 support for the -mfloat128-type option, and make -mfloat128
15057 default on PowerPC Linux systems. Define or undefine
15058 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15059 Define __float128 to be __ieee128 if IEEE 128-bit support is
15060 enabled, or undefine it.
15061 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15062 Delete defining __FLOAT128_TYPE__.
15063 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15064 -mfloat128-type option and make -mfloat128 default on PowerPC
15066 (TARGET_FLOAT128_TYPE): Likewise.
15067 (-mfloat128-type): Likewise.
15068 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15069 Delete the -mfloat128-type option and make -mfloat128 default on
15070 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15071 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15072 128-bit floating point is enabled. Change tests from using
15073 -mfloat128-type to -mfloat128.
15074 (rs6000_mangle_type): Use the correct mangling for the __float128
15075 type even if normal long double is restricted to 64-bits.
15076 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15078 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15079 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15080 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15082 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15083 documentation for -mfloat128.
15085 2017-09-06 Olivier Hainque <hainque@adacore.com>
15087 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15089 2017-09-06 Wish Wu <wishwu007@gmail.com>
15090 Jakub Jelinek <jakub@redhat.com>
15092 * asan.c (initialize_sanitizer_builtins): Add
15093 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15094 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15095 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15096 BT_FN_VOID_UINT64_PTR variables.
15097 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15098 (BT_FN_VOID_UINT16_UINT16): Likewise.
15099 (BT_FN_VOID_UINT32_UINT32): Likewise.
15100 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15101 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15102 (BT_FN_VOID_UINT64_PTR): Likewise.
15103 * common.opt (flag_sanitize_coverage): New variable.
15104 (fsanitize-coverage=trace-pc): Remove.
15105 (fsanitize-coverage=): Add.
15106 * flag-types.h (enum sanitize_coverage_code): New enum.
15107 * fold-const.c (fold_range_test): Disable non-short-circuit
15108 optimization if flag_sanitize_coverage.
15109 (fold_truth_andor): Likewise.
15110 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15111 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15112 (coverage_sanitizer_opts): New array.
15113 (get_closest_sanitizer_option): Add OPTS argument, handle also
15114 OPT_fsanitize_coverage_.
15115 (parse_sanitizer_options): Adjusted to also handle
15116 OPT_fsanitize_coverage_.
15117 (common_handle_option): Add OPT_fsanitize_coverage_.
15118 * sancov.c (instrument_comparison, instrument_switch): New function.
15119 (sancov_pass): Add trace-cmp support.
15120 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15121 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15122 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15123 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15124 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15125 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15126 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15127 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15128 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15129 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15131 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15133 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15134 error. Only quit immediately if parsing is complete.
15135 (BEGIN): Initialize fatal_err and parse_done.
15136 (begin fpu, end fpu): Check number of arguments.
15137 (begin arch, end arch): Likewise.
15138 (begin cpu, end cpu): Likewise.
15139 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15140 (optalias): Likewise.
15142 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15144 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15145 * config/arm/arm-isa.h: Delete. Move definitions to ...
15146 * arm-cpus.in: ... here. Use new feature and fgroup values.
15147 * config/arm/arm.c (arm_option_override): Use lower case for feature
15149 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15150 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15151 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15153 (print_isa_bits_for): New function.
15154 (gen_isa): New function.
15155 (gen_comm_data): Use print_isa_bits_for.
15156 (define feature): New keyword.
15157 (define fgroup): New keyword.
15158 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15159 (arm-isa.h): Add rule to generate file.
15160 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15161 case for feature bit names.
15163 2017-09-06 Richard Biener <rguenther@suse.de>
15165 * tree-ssa-pre.c (NECESSARY): Remove.
15166 (create_expression_by_pieces): Do not touch pass-local flags.
15167 (insert_into_preds_of_block): Likewise.
15168 (do_pre_regular_insertion): Likewise.
15169 (eliminate_insert): Likewise.
15170 (eliminate_dom_walker::before_dom_children): Likewise.
15171 (fini_eliminate): Do not look at inserted_exprs.
15172 (mark_operand_necessary): Remove.
15173 (remove_dead_inserted_code): Replace with simple work-list
15174 algorithm based on inserted_exprs and SSA uses.
15175 (pass_pre::execute): Re-order fini_eliminate and
15176 remove_dead_inserted_code.
15178 2017-09-06 Olivier Hainque <hainque@adacore.com>
15180 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15181 for VxWorks 7. Adjust surrounding comments.
15183 2017-09-06 Richard Biener <rguenther@suse.de>
15185 * gimple-ssa-strength-reduction.c
15186 (find_candidates_dom_walker::before_dom_children): Also allow
15189 2017-09-06 Richard Biener <rguenther@suse.de>
15191 PR tree-optimization/82108
15192 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15193 for gap in the non-permutation SLP case.
15195 2017-09-06 Martin Jambor <mjambor@suse.cz>
15197 PR tree-optimization/82078
15198 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15199 add_access_to_work_queue...
15200 (build_accesses_from_assign): ...here.
15201 (propagate_all_subaccesses): Make sure racc is the group
15202 representative, if there is one.
15204 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15206 PR middle-end/82095
15207 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15210 2017-09-06 Richard Biener <rguenther@suse.de>
15212 * gimple-ssa-strength-reduction.c
15213 (find_candidates_dom_walker::before_doom_children): Use a
15214 type and not a mode check.
15216 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15219 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15221 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15223 2017-09-05 Jeff Law <law@redhat.com>
15225 PR tree-optimization/64910
15226 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15227 cases where we have 3 or more operands.
15229 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15231 PR middle-end/81768
15232 * omp-low.c (lower_omp_for): Recompute tree invariant if
15233 gimple_omp_for_initial/final is ADDR_EXPR.
15235 PR middle-end/81768
15236 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15237 into gimple val before gimplification fo the COND_EXPR.
15239 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15241 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15242 REGION_COPY argument.
15243 (thread_through_all_blocks): Remove unused argument to
15244 duplicate_thread_path.
15246 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15247 Alan Hayward <alan.hayward@arm.com>
15248 David Sherwood <david.sherwood@arm.com>
15250 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15251 Take a scalar_mode rather than a machine_mode.
15252 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15253 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15254 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15255 (aarch64_gen_adjusted_ldpstp): Likewise.
15256 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15259 Alan Hayward <alan.hayward@arm.com>
15260 David Sherwood <david.sherwood@arm.com>
15262 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15263 Take a scalar_int_mode instead of a machine_mode.
15264 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15265 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15266 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15267 (aarch64_simd_attr_length_rglist): Delete.
15268 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15269 a scalar_int_mode instead of a machine_mode.
15270 (aarch64_add_offset): Likewise.
15271 (aarch64_internal_mov_immediate): Likewise
15272 (aarch64_add_constant_internal): Likewise.
15273 (aarch64_add_constant): Likewise.
15274 (aarch64_movw_imm): Likewise.
15275 (aarch64_rtx_arith_op_extract_p): Likewise.
15276 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15277 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15278 Remove assert that the mode isn't a vector.
15279 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15280 (aarch64_expand_mov_immediate): Update calls after above changes.
15281 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15282 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15283 (aarch64_move_imm): Likewise.
15284 (aarch64_can_const_movi_rtx_p): Likewise.
15285 (aarch64_strip_extend): Likewise.
15286 (aarch64_extr_rtx_p): Likewise.
15287 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15288 a CONST_INT when the mode parameter is VOIDmode.
15289 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15291 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15293 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15294 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15295 (bitwise_type_for_mode): Update accordingly.
15297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15299 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15300 * stor-layout.c (mode_for_size_tree): Likewise.
15301 (mode_for_array): Update accordingly.
15302 (layout_decl): Likewise.
15303 (compute_record_mode): Likewise. Only set the mode once.
15305 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15307 * target.def (get_mask_mode): Change return type to opt_mode.
15309 * doc/tm.texi: Regenerate.
15310 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15311 * targhooks.c (default_get_mask_mode): Likewise.
15312 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15313 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15314 targetm.get_mask_mode.
15315 * tree.c (build_truth_vector_type): Likewise.
15317 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15319 * machmode.h (mode_for_vector): Return an opt_mode.
15320 * stor-layout.c (mode_for_vector): Likewise.
15321 (mode_for_int_vector): Update accordingly.
15322 (layout_type): Likewise.
15323 * config/i386/i386.c (emit_memmov): Likewise.
15324 (ix86_expand_set_or_movmem): Likewise.
15325 (ix86_expand_vector_init): Likewise.
15326 (ix86_get_mask_mode): Likewise.
15327 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15329 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15330 * expmed.c (extract_bit_field_1): Likewise.
15331 * expr.c (expand_expr_real_2): Likewise.
15332 * optabs-query.c (can_vec_perm_p): Likewise.
15333 (can_vec_mask_load_store_p): Likewise.
15334 * optabs.c (expand_vec_perm): Likewise.
15335 * targhooks.c (default_get_mask_mode): Likewise.
15336 * tree-vect-stmts.c (vectorizable_store): Likewise.
15337 (vectorizable_load): Likewise.
15338 (get_vectype_for_scalar_type_and_size): Likewise.
15340 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15342 * machmode.h (mode_for_int_vector): New function.
15343 * stor-layout.c (mode_for_int_vector): Likewise.
15344 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15345 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15346 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15347 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15348 (s390_expand_vcond): Likewise.
15350 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15352 * machmode.h (opt_machine_mode): New type.
15353 (opt_mode<T>): Allow construction from anything that can be
15355 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15356 (mode_for_size): Return an opt_machine_mode.
15357 * stor-layout.c (mode_for_size): Likewise.
15358 (mode_for_size_tree): Update call accordingly.
15359 (bitwise_mode_for_mode): Likewise.
15360 (make_fract_type): Likewise.
15361 (make_accum_type): Likewise.
15362 * caller-save.c (replace_reg_with_saved_mem): Update call
15364 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15365 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15366 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15367 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15368 * expmed.c (extract_bit_field_1): Likewise.
15369 * reload.c (get_secondary_mem): Likewise.
15370 * varasm.c (assemble_integer): Likewise.
15371 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15374 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15376 * machmode.h (decimal_float_mode_for_size): New function.
15377 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15378 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15379 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15380 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15381 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15382 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15384 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15386 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15387 (get_builtin_sync_mode): Likewise.
15388 (expand_ifn_atomic_compare_exchange): Likewise.
15389 (expand_builtin_atomic_clear): Likewise.
15390 (expand_builtin_atomic_test_and_set): Likewise.
15391 (fold_builtin_atomic_always_lock_free): Likewise.
15392 * calls.c (compute_argument_addresses): Likewise.
15393 (emit_library_call_value_1): Likewise.
15394 (store_one_arg): Likewise.
15395 * combine.c (combine_instructions): Likewise.
15396 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15397 * config/arm/arm.c (arm_function_value): Likewise.
15398 (aapcs_allocate_return_reg): Likewise.
15399 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15400 * config/i386/i386.c (construct_container): Likewise.
15401 (ix86_gimplify_va_arg): Likewise.
15402 (ix86_expand_sse_cmp): Likewise.
15403 (emit_memmov): Likewise.
15404 (emit_memset): Likewise.
15405 (expand_small_movmem_or_setmem): Likewise.
15406 (ix86_expand_pextr): Likewise.
15407 (ix86_expand_pinsr): Likewise.
15408 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15409 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15411 * config/mips/mips.c (mips_function_value_1) Likewise.
15412 (mips_block_move_straight): Likewise.
15413 (mips_expand_ins_as_unaligned_store): Likewise.
15414 * config/powerpcspe/powerpcspe.c
15415 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15416 (rs6000_darwin64_record_arg_flush): Likewise.
15417 * config/rs6000/rs6000.c
15418 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15419 (rs6000_darwin64_record_arg_flush): Likewise.
15420 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15421 (sparc_function_value_1): Likewise.
15422 * config/spu/spu.c (adjust_operand): Likewise.
15423 (spu_emit_branch_or_set): Likewise.
15424 (arith_immediate_p): Likewise.
15425 * emit-rtl.c (gen_lowpart_common): Likewise.
15426 * expr.c (expand_expr_real_1): Likewise.
15427 * function.c (assign_parm_setup_block): Likewise.
15428 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15429 * reload1.c (alter_reg): Likewise.
15430 * stor-layout.c (mode_for_vector): Likewise.
15431 (layout_type): Likewise.
15433 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15435 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15436 (spu_convert_move): Likewise.
15437 * lower-subreg.c (resolve_simple_move): Likewise.
15439 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15442 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15443 define_insn to a define_expand.
15444 (altivec_vsum2sws_direct): New define_insn.
15445 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15447 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15449 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15452 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15457 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15458 realign stack if stack alignment needed is less than incoming
15461 2017-09-05 Marek Polacek <polacek@redhat.com>
15464 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15467 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15469 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15471 2017-09-05 Richard Biener <rguenther@suse.de>
15473 PR tree-optimization/82084
15474 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15476 2017-09-05 Richard Biener <rguenther@suse.de>
15478 PR tree-optimization/82102
15479 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15481 2017-09-05 Martin Liska <mliska@suse.cz>
15483 PR tree-optimization/82032
15484 * tree-cfg.c (generate_range_test): New function.
15485 * tree-cfg.h (generate_range_test): Declared here.
15486 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15487 (cleanup_control_expr_graph): Use it.
15488 * tree-switch-conversion.c (try_switch_expansion): Remove
15490 (emit_case_nodes): Use generate_range_test.
15492 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15495 * config/i386/i386.md (*<btsc><mode>_mask): Add
15496 TARGET_USE_BT to insn constraint.
15497 (*btr<mode>_mask): Ditto.
15499 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15501 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15502 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15504 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15507 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15508 TARGET_NEON and TARGET_IWMMXT.
15509 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15510 TARGET_NEON and TARGET_IWMMXT.
15511 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15513 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15515 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15516 (ix86_rewrite_tls_address): Ditto.
15517 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15518 (ix86_rewrite_tls_address_1): Ditto.
15519 (ix86_rewrite_tls_address): Ditto.
15520 * config/i386/predicates.md (tls_address_pattern): New predicate.
15521 * config/i386/i386.md (TLS address splitter): New splitter.
15523 2017-09-04 Richard Biener <rguenther@suse.de>
15525 PR tree-optimization/82084
15526 * fold-const.h (can_native_encode_string_p): Declare.
15527 * fold-const.c (can_native_encode_string_p): Factor out from ...
15528 (native_encode_string): ... here.
15529 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15530 vectorizing stores from constants we later cannot handle.
15532 2017-09-04 Marek Polacek <polacek@redhat.com>
15535 * doc/invoke.texi: Update -Wtautological-compare documentation.
15537 2017-09-04 Jeff Law <law@redhat.com>
15539 PR tree-optimization/64910
15540 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15541 swap the first and last operand if the last is a constant.
15543 2017-09-04 Marek Polacek <polacek@redhat.com>
15546 * convert.c (do_narrow): When sanitizing signed integer overflows,
15547 bail out for signed types.
15548 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15550 2017-09-04 Richard Biener <rguenther@suse.de>
15552 PR tree-optimization/82060
15553 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15554 Move devirtualization after stmt folding and before EH/AB/noreturn
15555 cleanup to get the stmt refs canonicalized. Use a bool instead
15556 of gimple_modified_p since that doesn't work for NOPs. Schedule
15557 NOPs generated by folding for removal.
15559 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15560 Alan Hayward <alan.hayward@arm.com>
15561 David Sherwood <david.sherwood@arm.com>
15563 * coretypes.h (pad_direction): New enum.
15564 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15565 (FUNCTION_ARG_PADDING): Likewise.
15566 * target.def (function_arg_padding): New hook.
15567 * targhooks.h (default_function_arg_padding): Declare.
15568 * targhooks.c (default_function_arg_padding): New function.
15569 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15570 (TARGET_FUNCTION_ARG_PADDING): ...this.
15571 * doc/tm.texi: Regenerate.
15572 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15573 instead of direction.
15574 (compute_argument_addresses): Likewise.
15575 (load_register_parameters): Likewise.
15576 (emit_library_call_value_1): Likewise.
15577 (store_one_arg): Use targetm.calls.function_arg_padding instead
15578 of FUNCTION_ARG_PADDING.
15579 (must_pass_in_stack_var_size_or_pad): Likewise.
15580 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15581 (emit_group_store): Likewise.
15582 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15583 instead of FUNCTION_ARG_PADDING.
15584 (emit_push_insn): Likewise, and propagate enum change throughout
15586 * function.h (direction): Delete.
15587 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15589 * function.c (assign_parm_find_stack_rtl): Likewise.
15590 (assign_parm_setup_block_p): Likewise.
15591 (assign_parm_setup_block): Likewise.
15592 (gimplify_parameters): Likewise.
15593 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15594 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15596 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15597 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15598 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15599 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15600 (aarch64_function_arg_padding): ...this new function.
15601 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15602 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15603 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15604 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15605 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15606 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15607 (arm_pad_arg_upward): Replace with...
15608 (arm_function_arg_padding): ...this new function.
15609 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15611 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15612 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15613 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15614 (ia64_hpux_function_arg_padding): Replace with...
15615 (ia64_function_arg_padding): ...this new function. Use pad_direction
15616 instead of direction. Check for TARGET_HPUX.
15617 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15618 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15619 (iq2000_function_arg_padding): New function.
15620 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15621 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15622 (mips_function_arg_padding): ...this new function.
15623 (mips_pad_reg_upward): Update accordingly.
15624 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15625 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15626 targetm.calls.function_arg_padding.
15627 (FUNCTION_ARG_PADDING): Delete.
15628 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15629 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15630 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15631 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15632 (nios2_block_reg_padding): Return pad_direction instead of direction.
15633 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15634 instead of direction.
15635 (nios2_function_arg_padding): Likewise. Make static.
15636 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15637 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15638 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15639 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15640 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15641 (pa_function_arg_padding): Make static. Return pad_direction instead
15643 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15644 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15645 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15646 instead of direction. Use targetm.calls.function_arg_padding.
15647 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15648 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15649 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15650 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15651 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15653 (function_arg_padding): Rename to...
15654 (rs6000_function_arg_padding): ...this. Make static. Return
15655 pad_direction instead of direction.
15656 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15657 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15658 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15659 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15660 instead of direction. Use targetm.calls.function_arg_padding.
15661 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15662 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15663 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15664 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15665 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15666 (function_arg_padding): Rename to...
15667 (rs6000_function_arg_padding): ...this. Make static. Return
15668 pad_direction instead of direction.
15669 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15670 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15671 * config/s390/s390.c (s390_function_arg_padding): New function.
15672 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15673 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15674 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15675 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15676 (function_arg_padding): Rename to...
15677 (sparc_function_arg_padding): ...this. Make static. Return
15678 pad_direction instead of direction.
15679 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15680 * config/spu/spu.c (spu_function_arg_padding): New function.
15681 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15682 * system.h (FUNCTION_ARG_PADDING): Poison.
15684 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15685 Alan Hayward <alan.hayward@arm.com>
15686 David Sherwood <david.sherwood@arm.com>
15688 * target.def (modes_tieable_p): New hook.
15689 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15690 (TARGET_MODES_TIEABLE_P): ...this.
15691 * doc/tm.texi.in: Regenerate.
15692 * hooks.h (hook_bool_mode_mode_true): Declare.
15693 * hooks.c (hook_bool_mode_mode_true): New function.
15694 * combine.c (subst): Use targetm.modes_tieable_p instead of
15696 * dse.c (find_shift_sequence): Likewise.
15697 * expmed.c (extract_low_bits): Likewise.
15698 * lower-subreg.c: Include target.h.
15699 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15701 * rtlanal.c (rtx_cost): Likewise.
15702 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15703 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15704 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15705 (TARGET_MODES_TIEABLE_P): Redefine.
15706 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15707 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15708 (TARGET_MODES_TIEABLE_P): Redefine.
15709 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15710 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15711 (arc_modes_tieable_p): New function.
15712 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15713 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15714 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15715 (arm_modes_tieable_p): Make static.
15716 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15717 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15718 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15719 (TARGET_MODES_TIEABLE_P): Redefine.
15720 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15721 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15722 (TARGET_MODES_TIEABLE_P): Redefine.
15723 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15724 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15725 (cr16_modes_tieable_p): New function.
15726 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15727 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15728 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15729 (TRULY_NOOP_TRUNCATION): Update comment.
15730 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15731 (TRULY_NOOP_TRUNCATION): Update comment.
15732 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15733 (frv_modes_tieable_p): New function.
15734 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15735 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15736 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15737 (TARGET_MODES_TIEABLE_P): Redefine.
15738 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15739 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15740 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15741 (TARGET_MODES_TIEABLE_P): Redefine.
15742 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15743 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15744 (ia64_modes_tieable_p): New function.
15745 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15746 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15747 (iq2000_modes_tieable_p): New function.
15748 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15749 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15750 (lm32_modes_tieable_p): New function.
15751 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15752 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15753 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15754 (TARGET_MODES_TIEABLE_P): Redefine.
15755 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15756 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15757 (m32r_modes_tieable_p): New function.
15758 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15759 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15760 (m68k_modes_tieable_p): New function.
15761 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15762 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15763 (mcore_modes_tieable_p): New function.
15764 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15765 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15767 (TARGET_MODES_TIEABLE_P): Redefine.
15768 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15769 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15770 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15771 (TARGET_MODES_TIEABLE_P): Redefine.
15772 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15773 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15774 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15775 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15776 (mn10300_modes_tieable_p): ...this and make static.
15777 (TARGET_MODES_TIEABLE_P): Redefine.
15778 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15779 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15780 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15781 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15782 (msp430_modes_tieable_p): Make static.
15783 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15784 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15785 (TARGET_MODES_TIEABLE_P): Redefine.
15786 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15787 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15788 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15789 (TARGET_MODES_TIEABLE_P): Redefine.
15790 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15791 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15792 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15793 (TARGET_MODES_TIEABLE_P): Redefine.
15794 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15795 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15796 (pdp11_modes_tieable_p): New function.
15797 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15798 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15799 (rs6000_modes_tieable_p): New function.
15800 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15801 * config/powerpcspe/powerpcspe.md: Update comment.
15802 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15803 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15804 (TARGET_MODES_TIEABLE_P): Redefine.
15805 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15806 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15807 (rl78_modes_tieable_p): New function.
15808 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15809 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15810 (rs6000_modes_tieable_p): New function.
15811 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15812 * config/rs6000/rs6000.md: Update comment.
15813 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15814 * config/rx/rx.c (rx_modes_tieable_p): New function.
15815 (TARGET_MODES_TIEABLE_P): Redefine.
15816 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15817 * config/s390/s390.c (s390_modes_tieable_p): New function.
15818 (TARGET_MODES_TIEABLE_P): Redefine.
15819 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15820 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15821 (sh_modes_tieable_p): New function.
15822 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15823 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15824 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15825 (sparc_modes_tieable_p): Make static.
15826 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15827 * config/spu/spu.c (spu_modes_tieable_p): New function.
15828 (TARGET_MODES_TIEABLE_P): Redefine.
15829 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15830 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15831 (TARGET_MODES_TIEABLE_P): Redefine.
15832 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15833 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15834 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15835 * config/v850/v850.c (v850_modes_tieable_p): New function.
15836 (TARGET_MODES_TIEABLE_P): Redefine.
15837 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15838 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15839 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15840 (visium_modes_tieable_p): New function.
15841 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15842 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15843 (xtensa_modes_tieable_p): New function.
15844 * system.h (MODES_TIEABLE_P): Poison.
15846 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15847 Alan Hayward <alan.hayward@arm.com>
15848 David Sherwood <david.sherwood@arm.com>
15850 * target.def (hard_regno_mode_ok): New hook.
15851 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15852 (TARGET_HARD_REGNO_MODE_OK): ...this.
15853 * doc/tm.texi.in: Regenerate.
15854 * hooks.h (hook_bool_uint_mode_true): Declare.
15855 * hooks.c (hook_bool_uint_mode_true): New function.
15856 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15857 HARD_REGNO_MODE_OK.
15858 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15859 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15860 instead of HARD_REGNO_MODE_OK.
15861 * caller-save.c: Include target.h.
15862 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15863 HARD_REGNO_MODE_OK.
15864 * combine.c (can_combine_p): Likewise.
15865 (combinable_i3pat): Likewise.
15866 (can_change_dest_mode): Likewise.
15867 * expr.c (init_expr_target): Likewise.
15868 (convert_move): Likewise.
15869 (convert_modes): Likewise.
15870 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15871 (setup_prohibited_mode_move_regs): Likewise.
15872 * ira.h (target_ira): Likewise.
15873 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15874 * lra-constraints.c (process_alt_operands): Likewise.
15875 (split_reg): Likewise.
15876 * recog.c (peep2_find_free_register): Likewise.
15877 * ree.c (combine_reaching_defs): Likewise.
15878 * regcprop.c (maybe_mode_change): Likewise.
15879 * reginfo.c (init_reg_sets_1): Likewise.
15880 (choose_hard_reg_mode): Likewise.
15881 (simplifiable_subregs): Likewise.
15882 * regrename.c (check_new_reg_p): Likewise.
15883 * reload.c (find_valid_class): Likewise.
15884 (find_valid_class_1): Likewise.
15885 (reload_inner_reg_of_subreg): Likewise.
15886 (push_reload): Likewise.
15887 (combine_reloads): Likewise.
15888 (find_dummy_reload): Likewise.
15889 (find_reloads): Likewise.
15890 * reload1.c (find_reg): Likewise.
15891 (set_reload_reg): Likewise.
15892 (allocate_reload_reg): Likewise.
15893 (choose_reload_regs): Likewise.
15894 (reload_adjust_reg_for_temp): Likewise.
15895 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15896 (simplify_subreg_regno): Likewise.
15897 * sel-sched.c (init_regs_for_mode): Likewise.
15898 * varasm.c (make_decl_rtl): Likewise.
15899 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15900 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15901 HARD_REGNO_MODE_OK.
15902 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15903 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15904 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15905 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15906 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15908 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15909 (arc_mode_class): Delete.
15910 (HARD_REGNO_MODE_OK): Delete.
15911 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15912 (arc_hard_regno_mode_ok): Rename old array to...
15913 (arc_hard_regno_mode_ok_modes): ...this.
15914 (arc_conditional_register_usage): Update accordingly.
15915 (arc_mode_class): Make static.
15916 (arc_hard_regno_mode_ok): New function.
15917 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15918 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15919 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15920 (arm_hard_regno_mode_ok): Make static.
15921 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15922 HARD_REGNO_MODE_OK.
15923 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15924 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15925 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15928 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15929 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15930 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15931 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15932 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15933 * config/bfin/predicates.md (valid_reg_operand): Use
15934 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15935 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15936 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15937 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15938 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15939 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15940 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15941 (cr16_hard_regno_mode_ok): Make static and return a bool.
15942 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15943 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15944 (cris_hard_regno_mode_ok): New function.
15945 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15946 (epiphany_mode_class): Delete.
15947 (HARD_REGNO_MODE_OK): Delete.
15948 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15949 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15950 (hard_regno_mode_ok): Rename to...
15951 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15952 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15953 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15954 HARD_REGNO_MODE_OK.
15955 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15956 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15957 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15958 (frv_hard_regno_mode_ok): Make static and return a bool.
15959 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15960 HARD_REGNO_MODE_OK.
15961 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15962 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15963 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15964 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15967 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15968 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15969 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15972 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15973 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15974 (ia64_hard_regno_mode_ok): New function.
15975 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15976 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15977 (iq2000_hard_regno_mode_ok): New function.
15978 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15979 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15980 (lm32_hard_regno_mode_ok): New function.
15981 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15982 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15983 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15984 instead of HARD_REGNO_MODE_OK.
15985 (m32c_hard_regno_ok): Rename to...
15986 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15987 (m32c_cannot_change_mode_class): Update accordingly.
15988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15989 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15990 (m32r_mode_class): Delete.
15991 (HARD_REGNO_MODE_OK): Delete.
15992 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15993 (m32r_hard_regno_mode_ok): Rename to...
15994 (m32r_hard_regno_modes): ...this.
15995 (m32r_mode_class): Make static.
15996 (m32r_hard_regno_mode_ok): New function.
15997 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15998 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15999 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16000 (m68k_hard_regno_mode_ok): Make static.
16001 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16002 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16003 (mcore_hard_regno_mode_ok): New function.
16004 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16005 (HARD_REGNO_MODE_OK): Delete.
16006 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16008 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16009 (microblaze_hard_regno_mode_ok): New function.
16010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16011 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16012 (mips_hard_regno_mode_ok): Delete.
16013 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16014 (mips_hard_regno_mode_ok_p): ...this and make static.
16015 (mips_hard_regno_mode_ok_p): Rename to...
16016 (mips_hard_regno_mode_ok_uncached): ...this.
16017 (mips_hard_regno_mode_ok): New function.
16018 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16019 of HARD_REGNO_MODE_OK.
16020 (mips_option_override): Update after above name changes.
16021 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16022 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16023 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16024 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16025 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16026 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16027 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16028 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16029 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16030 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16031 (msp430_hard_regno_mode_ok): Make static and return a bool.
16032 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16033 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16034 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16036 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16037 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16038 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16039 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16040 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16041 (PA_HARD_REGNO_MODE_OK): ...this
16042 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16043 (PA_HARD_REGNO_MODE_OK): ...this.
16044 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16045 (pa_hard_regno_mode_ok): New function.
16046 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16047 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16048 (pdp11_hard_regno_mode_ok): New function.
16049 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16050 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16052 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16054 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16055 (rs6000_hard_regno_mode_ok): Rename to...
16056 (rs6000_hard_regno_mode_ok_uncached): ...this.
16057 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16058 (rs6000_hard_regno_mode_ok): New function.
16059 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16060 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16061 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16062 (riscv_hard_regno_mode_ok): ...this and make static.
16063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16064 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16065 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16066 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16067 (rl78_hard_regno_mode_ok): Make static and return bool.
16068 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16069 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16071 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16072 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16073 (rs6000_hard_regno_mode_ok): Rename to...
16074 (rs6000_hard_regno_mode_ok_uncached): ...this.
16075 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16076 (rs6000_hard_regno_mode_ok): New function.
16077 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16078 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16079 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16080 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16081 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16082 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16083 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16084 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16085 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16086 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16087 (sh_hard_regno_mode_ok): Make static.
16088 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16089 instead of HARD_REGNO_MODE_OK.
16090 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16091 (sparc_mode_class): Delete.
16092 (HARD_REGNO_MODE_OK): Delete.
16093 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16094 (hard_regno_mode_classes): Make static.
16095 (sparc_mode_class): Likewise.
16096 (sparc_hard_regno_mode_ok): New function.
16097 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16098 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16099 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16102 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16103 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16104 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16105 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16106 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16107 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16108 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16109 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16110 (visium_hard_regno_mode_ok): New function.
16111 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16112 instead of HARD_REGNO_MODE_OK.
16113 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16114 (HARD_REGNO_MODE_OK): Delete.
16115 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16116 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16117 (xtensa_option_override): Update accordingly.
16118 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16119 (xtensa_hard_regno_mode_ok): New function.
16120 * system.h (HARD_REGNO_MODE_OK): Poison.
16122 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16123 Alan Hayward <alan.hayward@arm.com>
16124 David Sherwood <david.sherwood@arm.com>
16126 * target.def (hard_regno_call_part_clobbered): New hook.
16127 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16128 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16129 * doc/tm.texi: Regenerate.
16130 * hooks.h (hook_bool_uint_mode_false): Declare.
16131 * hooks.c (hook_bool_uint_mode_false): New function.
16132 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16133 * cselib.c (cselib_process_insn): Use
16134 targetm.hard_regno_call_part_clobbered instead of
16135 HARD_REGNO_CALL_PART_CLOBBERED.
16136 * ira-conflicts.c (ira_build_conflicts): Likewise.
16137 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16138 * lra-constraints.c (need_for_call_save_p): Likewise.
16139 * lra-lives.c: Include target.h.
16140 (check_pseudos_live_through_calls): Use
16141 targetm.hard_regno_call_part_clobbered instead of
16142 HARD_REGNO_CALL_PART_CLOBBERED.
16143 * regcprop.c: Include target.h.
16144 (copyprop_hardreg_forward_1): Use
16145 targetm.hard_regno_call_part_clobbered instead of
16146 HARD_REGNO_CALL_PART_CLOBBERED.
16147 * reginfo.c (choose_hard_reg_mode): Likewise.
16148 * regrename.c (check_new_reg_p): Likewise.
16149 * reload.c (find_equiv_reg): Likewise.
16150 * reload1.c (emit_reload_insns): Likewise.
16151 * sched-deps.c (deps_analyze_insn): Likewise.
16152 * sel-sched.c (init_regs_for_mode): Likewise.
16153 (mark_unavailable_hard_regs): Likewise.
16154 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16155 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16156 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16158 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16159 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16160 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16162 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16164 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16165 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16166 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16168 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16169 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16170 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16172 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16173 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16175 * config/powerpcspe/powerpcspe.c
16176 (rs6000_hard_regno_call_part_clobbered): New function.
16177 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16178 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16179 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16181 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16182 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16183 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16185 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16186 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16187 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16189 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16190 Alan Hayward <alan.hayward@arm.com>
16191 David Sherwood <david.sherwood@arm.com>
16193 * rtl.h (subreg_memory_offset): Declare.
16194 * emit-rtl.c (subreg_memory_offset): New function.
16195 * expmed.c (store_bit_field_1): Use it.
16196 * expr.c (undefined_operand_subword_p): Likewise.
16197 * simplify-rtx.c (simplify_subreg): Likewise.
16199 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16201 PR rtl-optimization/57448
16204 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16205 using atomic_load pattern.
16206 (expand_atomic_store): Likewise.
16208 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16211 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16212 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16213 handling. Use replace_call_with_value with NULL instead of
16214 gsi_replace, unlink_stmt_vdef and release_defs.
16216 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16219 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16221 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16224 * rtl.h (emit_library_call_value_1): Declare.
16225 (emit_library_call): Replace declaration with a series of overloads.
16226 Remove the parameter count argument.
16227 (emit_library_call_value): Likewise.
16228 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16229 with an "rtx_mode_t *".
16230 (emit_library_call_value): Delete.
16231 (emit_library_call): Likewise.
16232 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16233 (asan_emit_allocas_unpoison): Likewise.
16234 * builtins.c (expand_builtin_powi): Likewise.
16235 (expand_asan_emit_allocas_unpoison): Likewise.
16236 * cfgexpand.c (expand_main_function): Likewise.
16237 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16238 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16239 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16240 * config/arm/arm.c (arm_trampoline_init): Likewise.
16241 (arm_call_tls_get_addr): Likewise.
16242 (arm_expand_divmod_libfunc): Likewise.
16243 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16244 (smulsi3_highpart): Likewise.
16245 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16246 (c6x_expand_compare): Likewise.
16247 (c6x_expand_movmem): Likewise.
16248 * config/frv/frv.c (frv_trampoline_init): Likewise.
16249 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16250 (ix86_expand_divmod_libfunc): Likewise.
16251 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16252 (ia64_expand_compare): Likewise.
16253 (ia64_profile_hook): Likewise.
16254 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16255 (nonlocal_goto): Likewise.
16256 (restore_stack_nonlocal): Likewise.
16257 * config/m32r/m32r.c (block_move_call): Likewise.
16258 (m32r_trampoline_init): Likewise.
16259 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16260 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16261 (m68k_call_m68k_read_tp): Likewise.
16262 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16263 (microblaze_expand_divide): Likewise.
16264 * config/mips/mips.h (mips_args): Likewise.
16265 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16266 (MIPS_ICACHE_SYNC): Likewise.
16267 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16268 (nios2_trampoline_init): Likewise.
16269 * config/pa/pa.c (hppa_tls_call): Likewise.
16270 (pa_trampoline_init): Likewise.
16271 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16272 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16273 (expand_strn_compare): Likewise.
16274 (rs6000_generate_compare): Likewise.
16275 (rs6000_expand_float128_convert): Likewise.
16276 (output_profile_hook): Likewise.
16277 (rs6000_trampoline_init): Likewise.
16278 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16279 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16280 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16281 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16282 (rs6000_generate_compare): Likewise.
16283 (rs6000_expand_float128_convert): Likewise.
16284 (output_profile_hook): Likewise.
16285 (rs6000_trampoline_init): Likewise.
16286 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16287 * config/sh/sh.c (sh_trampoline_init): Likewise.
16288 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16289 (sparc_emit_float_lib_cmp): Likewise.
16290 (sparc32_initialize_trampoline): Likewise.
16291 (sparc64_initialize_trampoline): Likewise.
16292 (sparc_profile_hook): Likewise.
16293 * config/spu/spu.c (ea_load_store): Likewise.
16294 * config/spu/spu.md (floatunssidf2): Likewise.
16295 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16296 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16297 * config/visium/visium.c (expand_block_move_4): Likewise.
16298 (expand_block_move_2): Likewise.
16299 (expand_block_move_1): Likewise.
16300 (expand_block_set_4): Likewise.
16301 (expand_block_set_2): Likewise.
16302 (expand_block_set_1): Likewise.
16303 (visium_trampoline_init): Likewise.
16304 (visium_profile_hook): Likewise.
16305 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16306 (xtensa_setup_frame_addresses): Likewise.
16307 (xtensa_trampoline_init): Likewise.
16308 * except.c (sjlj_emit_function_enter): Likewise.
16309 (sjlj_emit_function_exit): Likewise.
16310 * explow.c (allocate_dynamic_stack_space): Likewise.
16311 (probe_stack_range): Likewise.
16312 * expr.c (convert_mode_scalar): Likewise.
16313 * optabs.c (expand_binop): Likewise.
16314 (expand_twoval_binop_libfunc): Likewise.
16315 (expand_unop): Likewise.
16316 (prepare_cmp_insn): Likewise.
16317 (prepare_float_lib_cmp): Likewise.
16318 (expand_float): Likewise.
16319 (expand_fix): Likewise.
16320 (expand_fixed_convert): Likewise.
16321 (maybe_emit_sync_lock_test_and_set): Likewise.
16322 (expand_atomic_compare_and_swap): Likewise.
16323 (expand_mem_thread_fence): Likewise.
16324 (expand_atomic_fetch_op): Likewise.
16326 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16328 * doc/generic.texi (OpenACC): Adjust URL.
16329 * doc/invoke.texi (C Dialect Options): Ditto.
16331 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16333 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16334 predicate for operand 1. Add (m,<S>) constraint.
16335 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16336 Prevent memory operand 1 with register operand 2.
16338 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16340 PR rtl-optimization/82024
16341 * combine.c (try_combine): If the combination result is a PARALLEL,
16342 and we only need to retain the SET in there that would be placed
16343 at I2, check that we can place that at I3 instead, before doing so.
16345 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16348 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16350 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16351 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16354 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16355 Jeff Law <law@redhat.com>
16357 * varasm.c (bss_initializer_p): Do not put constants into .bss
16358 (categorize_decl_for_section): Handle bss_initializer_p returning
16359 false when DECL_INITIAL is NULL.
16361 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16364 * config/s390/s390.c (s390_can_inline_p): New function.
16366 2017-09-01 Jeff Law <law@redhat.com>
16368 PR tree-optimization/82052
16369 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16370 Always initialize the returned slot after a hash table miss
16371 when INSERT is true.
16373 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16375 * config/s390/s390.md (mem_signal_fence): Remove.
16376 * doc/md.texi (mem_signal_fence): Remove.
16377 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16379 * target-insns.def (mem_signal_fence): Remove.
16381 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16384 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16387 * asan.c (create_odr_indicator): Strip name encoding from assembler
16388 name before appending it after __odr_asan_.
16390 2017-09-01 Martin Liska <mliska@suse.cz>
16392 PR tree-optimization/82059
16393 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16394 frequency only when an edge is redirected.
16396 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16398 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16399 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16400 (arc_conditional_register_usage): Remove ARC600 lp_count
16402 (arc_file_start): Emit Tag_ARC_CPU_variation.
16403 (arc_can_use_doloop_p): New conditions to use ZOLs.
16404 (hwloop_fail): New function.
16405 (hwloop_optimize): Likewise.
16406 (hwloop_pattern_reg): Likewise.
16407 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16408 (arc_reorg_loops): New function, calls reorg_loops.
16409 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16410 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16412 (arc_loop_hazard): Remove function, functionality moved into
16414 (arc_hazard): Remove arc_loop_hazard call.
16415 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16416 into hwloop_optimize.
16417 (arc_label_align): Remove ZOL handling.
16418 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16419 * config/arc/arc.md (doloop_begin): Remove pattern.
16420 (doloop_begin_i): Likewise.
16421 (doloop_end_i): Likewise.
16422 (doloop_fallback): Likewise.
16423 (doloop_fallback_m): Likewise.
16424 (doloop_end): Reimplement expand.
16425 (arc_lp): New pattern for LP instruction.
16426 (loop_end): New pattern.
16427 (loop_fail): Likewise.
16428 (decrement_and_branch_until_zero): Likewise.
16429 * config/arc/arc.opt (mlpc-width): New option.
16430 * doc/invoke.texi (mlpc-width): Document option.
16432 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16434 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16435 (arc_ccfsm_advance): Fix checking for delay slots.
16436 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16438 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16440 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16442 (movhi_insn): Update store instruction constraint which are saving
16443 6-bit short immediates.
16444 (movsi_insn): Consider also short scaled load operations.
16445 (zero_extendhisi2_i): Use Usd constraint instead of T.
16446 (extendhisi2_i): Add q constraint.
16447 (arc_clzsi2): Add type and length attributes.
16448 (arc_ctzsi2): Likewise.
16449 * config/arc/constraints.md (Usc): Update constraint, the
16450 assembler can parse two relocations for a single instruction.
16452 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16454 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16455 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16457 2017-08-31 Olivier Hainque <hainque@adacore.com>
16459 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16460 match as powerpc-wrs-vxworks*.
16462 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16464 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16465 register constraint for by-element operand.
16466 (aarch64_mls_elt_merge<mode>): Likewise.
16468 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16470 * config/arc/arc.c (arc_can_follow_jump): Check for short
16473 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16475 * config.gcc: Use g.opt for arc.
16476 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16477 functionality moved to ...
16478 (legitimate_scaled_address_p): New function, ...here.
16479 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16480 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16481 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16483 (arc_override_options): Handle G option.
16484 (arc_output_pic_addr_const): Correct function definition.
16485 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16486 (arc_decl_anon_ns_mem_p): Delete.
16487 (arc_in_small_data_p): Overhaul this function to take into
16488 consideration the value given via G option.
16489 (arc_rewrite_small_data_1): Renamed and corrected old
16490 arc_rewrite_small_data function.
16491 (arc_rewrite_small_data): New function.
16492 (small_data_pattern): Don't use pic_offset_table_rtx.
16493 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16494 * config/arc/simdext.md (movmisalignv2hi): Use
16495 prepare_move_operands function.
16497 (movmisalign*): Likewise.
16498 * doc/invoke.texi (ARC options): Document -G option.
16500 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16502 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16504 * config/arc/arc.c (arc_print_operand): Output scalled address for
16505 sdata whenever is possible.
16506 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16507 load/stores are available.
16508 (compact_sda_memory_operand): Check for the alignment required by
16509 code density instructions.
16510 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16512 * config/arc/constraints.md (Usd): Update constraint.
16513 (Us0): New constraint.
16514 (Usc): Update constraint.
16516 2017-08-31 Richard Biener <rguenther@suse.de>
16518 PR middle-end/82054
16519 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16520 function only once.
16522 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16524 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16525 Resize type_signature.
16527 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16528 Alan Hayward <alan.hayward@arm.com>
16529 David Sherwood <david.sherwood@arm.com>
16531 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16532 subregs whose inner modes can be stored in GPRs.
16533 (aarch64_classify_index): Likewise.
16535 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16536 Alan Hayward <alan.hayward@arm.com>
16537 David Sherwood <david.sherwood@arm.com>
16539 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16540 (V_INT_EQUIV): ...this.
16541 (v_cmp_result): Rename to...
16542 (v_int_equiv): ...this.
16543 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16544 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16545 (copysign<mode>3): Likewise.
16546 (aarch64_simd_bsl<mode>_internal): Likewise.
16547 (aarch64_simd_bsl<mode>): Likewise.
16548 (vec_cmp<mode><mode>): Likewise.
16549 (vcond<mode><mode>): Likewise.
16550 (vcond<v_cmp_mixed><mode>): Likewise.
16551 (vcondu<mode><v_cmp_mixed>): Likewise.
16552 (aarch64_cm<optab><mode>): Likewise.
16553 (aarch64_cmtst<mode>): Likewise.
16554 (aarch64_fac<optab><mode>): Likewise.
16555 (vec_perm_const<mode>): Likewise.
16556 (vcond_mask_<mode><v_cmp_result>): Rename to...
16557 (vcond_mask_<mode><v_int_equiv>): ...this.
16558 (vec_cmp<mode><v_cmp_result>): Rename to...
16559 (vec_cmp<mode><v_int_equiv>): ...this.
16561 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16562 Alan Hayward <alan.hayward@arm.com>
16563 David Sherwood <david.sherwood@arm.com>
16565 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16567 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16568 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16569 (UNSPEC_LD4_DREG): New unspecs.
16570 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16571 (aarch64_ld2<mode>_dreg_be): Replace with...
16572 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16574 (aarch64_ld3<mode>_dreg_le)
16575 (aarch64_ld3<mode>_dreg_be): Replace with...
16576 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16578 (aarch64_ld4<mode>_dreg_le)
16579 (aarch64_ld4<mode>_dreg_be): Replace with...
16580 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16583 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16585 PR tree-optimization/81987
16586 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16587 insert an initializer in a location not dominated by the stride
16590 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16592 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16593 on the entire header of the finally block in the fallthru case.
16595 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16597 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16599 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16601 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16602 rs6000_emit_move_from_cr and call renamed function.
16603 (rs6000_emit_prologue): Call renamed functions.
16604 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16605 movesi_from_cr, remove volatile CRs.
16607 2017-08-30 Jon Beniston <jon@beniston.com>
16608 Richard Biener <rguenther@suse.de>
16610 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16611 of VECTOR_MODE_P check.
16612 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16613 element vector types.
16615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16617 * df.h (df_read_modify_subreg_p): Remove in favor of...
16618 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16619 const_rtx instead of an rtx.
16620 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16621 * df-problems.c (df_word_lr_mark_ref): Likewise.
16622 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16623 (mark_pseudo_reg_dead): Likewise.
16624 (mark_ref_dead): Likewise.
16625 * reginfo.c (init_subregs_of_mode): Likewise.
16626 * sched-deps.c (sched_analyze_1): Likewise.
16627 * df-scan.c (df_def_record_1): Likewise.
16628 (df_uses_record): Likewise.
16629 (df_read_modify_subreg_p): Remove in favor of...
16630 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16631 const_rtx instead of an rtx.
16633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16634 Alan Hayward <alan.hayward@arm.com>
16635 David Sherwood <david.sherwood@arm.com>
16637 * rtl.h (partial_subreg_p): New function.
16638 * caller-save.c (save_call_clobbered_regs): Use it.
16639 * calls.c (expand_call): Likewise.
16640 * combine.c (combinable_i3pat): Likewise.
16641 (simplify_set): Likewise.
16642 (make_extraction): Likewise.
16643 (make_compound_operation_int): Likewise.
16644 (gen_lowpart_or_truncate): Likewise.
16645 (force_to_mode): Likewise.
16646 (make_field_assignment): Likewise.
16647 (reg_truncated_to_mode): Likewise.
16648 (record_truncated_value): Likewise.
16649 (move_deaths): Likewise.
16650 * cse.c (record_jump_cond): Likewise.
16651 (cse_insn): Likewise.
16652 * cselib.c (cselib_lookup_1): Likewise.
16653 * expmed.c (extract_bit_field_using_extv): Likewise.
16654 * function.c (assign_parm_setup_reg): Likewise.
16655 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16656 * ira-build.c (create_insn_allocnos): Likewise.
16657 * lra-coalesce.c (merge_pseudos): Likewise.
16658 * lra-constraints.c (match_reload): Likewise.
16659 (simplify_operand_subreg): Likewise.
16660 (curr_insn_transform): Likewise.
16661 * lra-lives.c (process_bb_lives): Likewise.
16662 * lra.c (new_insn_reg): Likewise.
16663 (lra_substitute_pseudo): Likewise.
16664 * regcprop.c (mode_change_ok): Likewise.
16665 (maybe_mode_change): Likewise.
16666 (copyprop_hardreg_forward_1): Likewise.
16667 * reload.c (push_reload): Likewise.
16668 (find_reloads): Likewise.
16669 (find_reloads_subreg_address): Likewise.
16670 * reload1.c (alter_reg): Likewise.
16671 (eliminate_regs_1): Likewise.
16672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16674 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16676 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16677 back to if statements, including unpack.
16679 2017-08-30 Martin Liska <mliska@suse.cz>
16681 PR inline-asm/82001
16682 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16684 (func_checker::compare_asm_inputs_outputs): ... this function.
16685 (func_checker::compare_gimple_asm): Use the function to compare
16686 also ASM constrains.
16687 * ipa-icf-gimple.h: Rename the function.
16689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16690 Alan Hayward <alan.hayward@arm.com>
16691 David Sherwood <david.sherwood@arm.com>
16693 * coretypes.h (complex_mode): New type.
16694 * gdbhooks.py (build_pretty_printer): Handle it.
16695 * machmode.h (complex_mode): New class.
16696 (complex_mode::includes_p): New function.
16697 (is_complex_int_mode): Likewise.
16698 (is_complex_float_mode): Likewise.
16699 * genmodes.c (get_mode_class): Handle complex mode classes.
16700 * function.c (expand_function_end): Use is_complex_int_mode.
16702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16703 Alan Hayward <alan.hayward@arm.com>
16704 David Sherwood <david.sherwood@arm.com>
16706 * coretypes.h (scalar_mode_pod): New typedef.
16707 * gdbhooks.py (build_pretty_printer): Handle it.
16708 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16709 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16710 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16711 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16712 as_a <scalar_mode>.
16714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16715 Alan Hayward <alan.hayward@arm.com>
16716 David Sherwood <david.sherwood@arm.com>
16718 * machmode.h (mode_for_vector): Take a scalar_mode instead
16720 * stor-layout.c (mode_for_vector): Likewise.
16721 * explow.c (promote_mode): Use as_a <scalar_mode>.
16722 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16725 Alan Hayward <alan.hayward@arm.com>
16726 David Sherwood <david.sherwood@arm.com>
16728 * target.def (preferred_simd_mode): Take a scalar_mode
16729 instead of a machine_mode.
16730 * targhooks.h (default_preferred_simd_mode): Likewise.
16731 * targhooks.c (default_preferred_simd_mode): Likewise.
16732 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16733 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16734 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16735 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16736 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16737 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16738 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16739 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16741 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16742 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16743 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16744 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16745 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16746 * doc/tm.texi: Regenerate.
16747 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16751 Alan Hayward <alan.hayward@arm.com>
16752 David Sherwood <david.sherwood@arm.com>
16754 * target.def (scalar_mode_supported_p): Take a scalar_mode
16755 instead of a machine_mode.
16756 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16757 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16758 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16759 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16760 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16761 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16762 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16763 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16764 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16765 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16766 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16767 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16768 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16769 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16771 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16772 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16773 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16774 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16775 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16777 * doc/tm.texi: Regenerate.
16779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16780 Alan Hayward <alan.hayward@arm.com>
16781 David Sherwood <david.sherwood@arm.com>
16783 * coretypes.h (opt_scalar_mode): New typedef.
16784 * gdbhooks.py (build_pretty_printers): Handle it.
16785 * machmode.h (mode_iterator::get_2xwider): Add overload for
16787 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16789 * expr.c (convert_mode_scalar): Likewise.
16790 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16791 * optabs.c (expand_float): Likewise.
16792 (expand_fix): Likewise.
16793 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16796 Alan Hayward <alan.hayward@arm.com>
16797 David Sherwood <david.sherwood@arm.com>
16799 * optabs.c (expand_float): Explicitly check for scalars before
16800 using a branching expansion.
16801 (expand_fix): Likewise.
16803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16804 Alan Hayward <alan.hayward@arm.com>
16805 David Sherwood <david.sherwood@arm.com>
16807 * expr.c (convert_mode): Split scalar handling out into...
16808 (convert_mode_scalar): ...this new function. Treat the modes
16811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16812 Alan Hayward <alan.hayward@arm.com>
16813 David Sherwood <david.sherwood@arm.com>
16815 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16817 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16820 Alan Hayward <alan.hayward@arm.com>
16821 David Sherwood <david.sherwood@arm.com>
16823 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16824 rather than a machine_mode.
16825 (fixed_from_string): Likewise.
16826 (fixed_convert): Likewise.
16827 (fixed_convert_from_int): Likewise.
16828 (fixed_convert_from_real): Likewise.
16829 (real_convert_from_fixed): Likewise.
16830 * fixed-value.c (fixed_from_double_int): Likewise.
16831 (fixed_from_string): Likewise.
16832 (fixed_convert): Likewise.
16833 (fixed_convert_from_int): Likewise.
16834 (fixed_convert_from_real): Likewise.
16835 (real_convert_from_fixed): Likewise.
16836 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16839 Alan Hayward <alan.hayward@arm.com>
16840 David Sherwood <david.sherwood@arm.com>
16842 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16843 of separate mode class checks. Do not allow vector modes here.
16844 (immed_wide_int_const): Use as_a <scalar_mode>.
16845 * explow.c (trunc_int_for_mode): Likewise.
16846 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16847 (wi::shwi): Likewise.
16848 (wi::min_value): Likewise.
16849 (wi::max_value): Likewise.
16850 * dwarf2out.c (loc_descriptor): Likewise.
16851 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16852 for CONST_WIDE_INT.
16854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16855 Alan Hayward <alan.hayward@arm.com>
16856 David Sherwood <david.sherwood@arm.com>
16858 * tree.h (SCALAR_TYPE_MODE): New macro.
16859 * expr.c (expand_expr_addr_expr_1): Use it.
16860 (expand_expr_real_2): Likewise.
16861 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16862 (fold_convert_const_fixed_from_int): Likewise.
16863 (fold_convert_const_fixed_from_real): Likewise.
16864 (native_encode_fixed): Likewise
16865 (native_encode_complex): Likewise
16866 (native_encode_vector): Likewise.
16867 (native_interpret_fixed): Likewise.
16868 (native_interpret_real): Likewise.
16869 (native_interpret_complex): Likewise.
16870 (native_interpret_vector): Likewise.
16871 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16872 (simd_clone_adjust_argument_types): Likewise.
16873 (simd_clone_init_simd_arrays): Likewise.
16874 (simd_clone_adjust): Likewise.
16875 * stor-layout.c (layout_type): Likewise.
16876 * tree.c (build_minus_one_cst): Likewise.
16877 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16878 * tree-inline.c (estimate_move_cost): Likewise.
16879 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16880 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16881 (vectorizable_reduction): Likewise.
16882 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16883 (vect_recog_mixed_size_cond_pattern): Likewise.
16884 (check_bool_pattern): Likewise.
16885 (adjust_bool_pattern): Likewise.
16886 (search_type_for_mask_1): Likewise.
16887 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16888 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16889 (vectorizable_load): Likewise.
16890 (vectorizable_store): Likewise.
16891 * ubsan.c (ubsan_encode_value): Likewise.
16892 * varasm.c (output_constant): Likewise.
16894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16895 Alan Hayward <alan.hayward@arm.com>
16896 David Sherwood <david.sherwood@arm.com>
16898 * coretypes.h (scalar_mode): New class.
16899 * machmode.h (scalar_mode): Likewise.
16900 (scalar_mode::includes_p): New function.
16901 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16902 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16903 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16904 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16905 * expmed.c (store_bit_field_1): Likewise.
16906 (extract_bit_field_1): Likewise.
16907 * expr.c (write_complex_part): Likewise.
16908 (read_complex_part): Likewise.
16909 (emit_move_complex_push): Likewise.
16910 (expand_expr_real_2): Likewise.
16911 * function.c (assign_parm_setup_reg): Likewise.
16912 (assign_parms_unsplit_complex): Likewise.
16913 * optabs.c (expand_binop): Likewise.
16914 * rtlanal.c (subreg_get_info): Likewise.
16915 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16916 * varasm.c (output_constant_pool_2): Likewise.
16918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16919 Alan Hayward <alan.hayward@arm.com>
16920 David Sherwood <david.sherwood@arm.com>
16922 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16924 (expmed_mult_highpart_optab): Likewise.
16925 (expmed_mult_highpart): Likewise.
16927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16928 Alan Hayward <alan.hayward@arm.com>
16929 David Sherwood <david.sherwood@arm.com>
16931 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16932 instead of a machine_mode.
16933 (builtin_memset_read_str): Likewise.
16934 * builtins.c (c_readstr): Likewise.
16935 (builtin_memcpy_read_str): Likewise.
16936 (builtin_strncpy_read_str): Likewise.
16937 (builtin_memset_read_str): Likewise.
16938 (builtin_memset_gen_str): Likewise.
16939 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16940 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16941 instead of a machine_mode.
16942 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16944 (make_extraction): Likewise.
16945 (try_widen_shift_mode): Take and return scalar_int_modes instead
16947 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16948 a scalar_int_mode instead of a machine_mode.
16949 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16950 (avr_addr_space_pointer_mode): Likewise.
16951 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16952 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16953 (msp430_unwind_word_mode): Likewise.
16954 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16955 (spu_addr_space_pointer_mode): Likewise.
16956 (spu_addr_space_address_mode): Likewise.
16957 (spu_libgcc_cmp_return_mode): Likewise.
16958 (spu_libgcc_shift_count_mode): Likewise.
16959 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16960 (rl78_addr_space_pointer_mode): Likewise.
16961 (fl78_unwind_word_mode): Likewise.
16962 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16964 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16965 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16966 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16967 (mips_valid_pointer_mode): Likewise.
16968 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16969 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16970 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16972 (ft32_addr_space_address_mode): Likewise.
16973 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16974 scalar_int_mode instead of a machine_mode.
16975 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16977 (m32c_addr_space_address_mode): Likewise.
16978 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16979 (rs6000_eh_return_filter_mode): Likewise.
16980 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16981 (rs6000_eh_return_filter_mode): Likewise.
16982 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16983 (s390_libgcc_shift_count_mode): Likewise.
16984 (s390_unwind_word_mode): Likewise.
16985 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16987 * target.def (mode_rep_extended): Likewise.
16988 (valid_pointer_mode): Likewise.
16989 (addr_space.valid_pointer_mode): Likewise.
16990 (eh_return_filter_mode): Return a scalar_int_mode rather than
16992 (libgcc_cmp_return_mode): Likewise.
16993 (libgcc_shift_count_mode): Likewise.
16994 (unwind_word_mode): Likewise.
16995 (addr_space.pointer_mode): Likewise.
16996 (addr_space.address_mode): Likewise.
16997 * doc/tm.texi: Regenerate.
16998 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17000 (do_jump): Use scalar_int_mode for local variables.
17001 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17002 rather than a machine_mode.
17003 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17004 (scompare_loc_descriptor_wide): Likewise.
17005 (scompare_loc_descriptor_narrow): Likewise.
17006 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17008 * except.c (sjlj_emit_dispatch_table): Likewise.
17009 (expand_builtin_eh_copy_values): Likewise.
17010 * explow.c (convert_memory_address_addr_space_1): Likewise.
17011 Take a scalar_int_mode rather than a machine_mode.
17012 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17013 than a machine_mode.
17014 (memory_address_addr_space): Use scalar_int_mode for local variables.
17015 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17016 rather than a machine_mode.
17017 * expmed.c (mask_rtx): Likewise.
17018 (init_expmed_one_conv): Likewise.
17019 (expand_mult_highpart_adjust): Likewise.
17020 (extract_high_half): Likewise.
17021 (expmed_mult_highpart_optab): Likewise.
17022 (expmed_mult_highpart): Likewise.
17023 (expand_smod_pow2): Likewise.
17024 (expand_sdiv_pow2): Likewise.
17025 (emit_store_flag_int): Likewise.
17026 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17028 (extract_low_bits): Likewise.
17029 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17031 * expr.c (pieces_addr::adjust): Likewise.
17032 (can_store_by_pieces): Likewise.
17033 (store_by_pieces): Likewise.
17034 (clear_by_pieces_1): Likewise.
17035 (expand_expr_addr_expr_1): Likewise.
17036 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17037 (expand_expr_real_1): Likewise.
17038 (try_casesi): Likewise.
17039 * final.c (shorten_branches): Likewise.
17040 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17041 type of "mode" to machine_mode.
17042 * internal-fn.c (expand_arith_overflow_result_store): Take a
17043 scalar_int_mode rather than a machine_mode.
17044 (expand_mul_overflow): Use scalar_int_mode for local variables.
17045 * loop-doloop.c (doloop_modify): Likewise.
17046 (doloop_optimize): Likewise.
17047 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17048 than a machine_mode.
17049 (expand_doubleword_shift_condmove): Likewise.
17050 (expand_doubleword_shift): Likewise.
17051 (expand_doubleword_clz): Likewise.
17052 (expand_doubleword_popcount): Likewise.
17053 (expand_doubleword_parity): Likewise.
17054 (expand_absneg_bit): Use scalar_int_mode for local variables.
17055 (prepare_float_lib_cmp): Likewise.
17056 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17057 rather than a machine_mode.
17058 (convert_memory_address_addr_space): Likewise.
17059 (get_mode_bounds): Likewise.
17060 (get_address_mode): Return a scalar_int_mode rather than a
17062 * rtlanal.c (get_address_mode): Likewise.
17063 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17064 than a machine_mode.
17065 * targhooks.c (default_mode_rep_extended): Likewise.
17066 (default_valid_pointer_mode): Likewise.
17067 (default_addr_space_valid_pointer_mode): Likewise.
17068 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17069 than a machine_mode.
17070 (default_libgcc_cmp_return_mode): Likewise.
17071 (default_libgcc_shift_count_mode): Likewise.
17072 (default_unwind_word_mode): Likewise.
17073 (default_addr_space_pointer_mode): Likewise.
17074 (default_addr_space_address_mode): Likewise.
17075 * targhooks.h (default_eh_return_filter_mode): Likewise.
17076 (default_libgcc_cmp_return_mode): Likewise.
17077 (default_libgcc_shift_count_mode): Likewise.
17078 (default_unwind_word_mode): Likewise.
17079 (default_addr_space_pointer_mode): Likewise.
17080 (default_addr_space_address_mode): Likewise.
17081 (default_mode_rep_extended): Take a scalar_int_mode rather than
17083 (default_valid_pointer_mode): Likewise.
17084 (default_addr_space_valid_pointer_mode): Likewise.
17085 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17087 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17088 rather than a machine_mode.
17089 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17090 for local variables.
17091 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17092 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17093 than a machine_mode.
17095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17096 Alan Hayward <alan.hayward@arm.com>
17097 David Sherwood <david.sherwood@arm.com>
17099 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17100 the mode argument to scalar_int_mode.
17101 (do_jump_by_parts_zero_rtx): Likewise.
17102 (do_jump_by_parts_equality_rtx): Likewise.
17103 (do_jump_by_parts_greater): Take a mode argument.
17104 (do_jump_by_parts_equality): Likewise.
17105 (do_jump_1): Update calls accordingly.
17107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17108 Alan Hayward <alan.hayward@arm.com>
17109 David Sherwood <david.sherwood@arm.com>
17111 * is-a.h (safe_dyn_cast): New function.
17112 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17113 (jump_table_for_label): Likewise.
17114 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17115 instead of an rtx_insn *.
17116 (shorten_branches): Use dyn_cast instead of LABEL_P and
17117 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17118 rtx_jump_table_data::get_data_mode.
17119 (final_scan_insn): Likewise.
17121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17122 Alan Hayward <alan.hayward@arm.com>
17123 David Sherwood <david.sherwood@arm.com>
17125 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17126 trying to combine a full-register integer set with a subreg
17129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17130 Alan Hayward <alan.hayward@arm.com>
17131 David Sherwood <david.sherwood@arm.com>
17133 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17134 that is always either address_mode or pointer_mode.
17136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17137 Alan Hayward <alan.hayward@arm.com>
17138 David Sherwood <david.sherwood@arm.com>
17140 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17141 when the two are known to be equal.
17143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17144 Alan Hayward <alan.hayward@arm.com>
17145 David Sherwood <david.sherwood@arm.com>
17147 * simplify-rtx.c (simplify_const_unary_operation): Use
17148 is_a <scalar_int_mode> instead of checking for a nonzero
17149 precision. Forcibly convert op_mode to a scalar_int_mode
17150 in that case. More clearly differentiate the operand and
17151 result modes and use the former when deciding what the value
17152 of a count-bits operation should be. Use is_int_mode instead
17153 of checking for a MODE_INT. Remove redundant check for whether
17154 this mode has a zero precision.
17156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17157 Alan Hayward <alan.hayward@arm.com>
17158 David Sherwood <david.sherwood@arm.com>
17160 * optabs.c (widen_leading): Change the type of the mode argument
17161 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17162 (widen_bswap): Likewise.
17163 (expand_parity): Likewise.
17164 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17165 (expand_ffs): Likewise.
17166 (epand_unop): Check for scalar integer modes before calling the
17169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17170 Alan Hayward <alan.hayward@arm.com>
17171 David Sherwood <david.sherwood@arm.com>
17173 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17175 (expand_expr_real_1): Update call accordingly.
17177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17178 Alan Hayward <alan.hayward@arm.com>
17179 David Sherwood <david.sherwood@arm.com>
17181 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17182 value_mode arguments. Use scalar_int_mode internally.
17183 (store_bit_field_1): Rename the new integer mode from imode
17184 to op0_mode and use it instead of GET_MODE (op0). Update calls
17185 to store_split_bit_field, store_bit_field_using_insv and
17186 store_fixed_bit_field.
17187 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17188 Use scalar_int_mode internally. Use a bit count rather than a mode
17189 when calculating the largest bit size for get_best_mode.
17190 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17191 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17192 Remove assertion that OP0 has a scalar integer mode.
17193 (store_split_bit_field): Add op0_mode and value_mode arguments.
17194 Update calls to extract_fixed_bit_field.
17195 (extract_bit_field_using_extv): Add an op0_mode argument.
17196 Use scalar_int_mode internally.
17197 (extract_bit_field_1): Rename the new integer mode from imode to
17198 op0_mode and use it instead of GET_MODE (op0). Update calls to
17199 extract_split_bit_field, extract_bit_field_using_extv and
17200 extract_fixed_bit_field.
17201 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17202 to extract_split_bit_field and extract_fixed_bit_field_1.
17203 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17204 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17205 on the target mode.
17206 (extract_split_bit_field): Add an op0_mode argument. Update call
17207 to extract_fixed_bit_field.
17209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17210 Alan Hayward <alan.hayward@arm.com>
17211 David Sherwood <david.sherwood@arm.com>
17213 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17214 * explow.c (hard_function_value): Likewise.
17215 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17216 convert_to_mode call outside the loop.
17217 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17218 for the mode iterator. Require the mode specified by max_pieces
17220 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17222 (copy_blkmode_to_reg): Likewise.
17223 (set_storage_via_setmem): Likewise.
17224 * optabs.c (prepare_cmp_insn): Likewise.
17225 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17226 * stor-layout.c (finish_bitfield_representative): Likewise.
17228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17229 Alan Hayward <alan.hayward@arm.com>
17230 David Sherwood <david.sherwood@arm.com>
17232 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17233 * expr.c (convert_move): Use them.
17234 (convert_modes): Likewise.
17235 (store_expr_with_bounds): Likewise.
17237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17238 Alan Hayward <alan.hayward@arm.com>
17239 David Sherwood <david.sherwood@arm.com>
17241 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17242 parameter for the mode of "x". Remove the "known_x", "known_mode"
17243 and "known_ret" arguments. Change the type of the mode argument
17244 to scalar_int_mode.
17245 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17246 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17247 (reg_num_sign_bit_copies_for_combine): Likewise.
17248 * rtlanal.c (nonzero_bits1): Likewise.
17249 (num_sign_bit_copies1): Likewise.
17250 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17251 (reg_num_sign_bit_copies_general): Likewise.
17252 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17253 (reg_nonzero_bits_general): Likewise.
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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17260 than in subroutines. Return 1 for non-integer modes.
17261 (cached_num_sign_bit_copies): Change the type of the mode parameter
17262 to scalar_int_mode.
17263 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17264 classes. Handle CONST_INT_P first and then check whether X also
17265 has a scalar integer mode. Check the same thing for inner registers
17266 of a SUBREG and for values that are being extended or truncated.
17268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17269 Alan Hayward <alan.hayward@arm.com>
17270 David Sherwood <david.sherwood@arm.com>
17272 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17273 in subroutines. Return the mode mask for non-integer modes.
17274 (cached_nonzero_bits): Change the type of the mode parameter
17275 to scalar_int_mode.
17276 (nonzero_bits1): Likewise. Remove early exit for other mode
17277 classes. Handle CONST_INT_P first and then check whether X
17278 also has a scalar integer mode.
17280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17281 Alan Hayward <alan.hayward@arm.com>
17282 David Sherwood <david.sherwood@arm.com>
17284 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17285 Return a scalar_int_mode and assert that the size is greater than
17287 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17288 (op_by_pieces_d::op_by_pieces_d): Likewise.
17289 (op_by_pieces_d::run): Likewise.
17290 (can_store_by_pieces): Likewise.
17292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17293 Alan Hayward <alan.hayward@arm.com>
17294 David Sherwood <david.sherwood@arm.com>
17296 * combine.c (extract_left_shift): Add a mode argument and update
17298 (make_compound_operation_int): Change the type of the mode parameter
17299 to scalar_int_mode and update the call to extract_left_shift.
17301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17302 Alan Hayward <alan.hayward@arm.com>
17303 David Sherwood <david.sherwood@arm.com>
17305 * combine.c (simplify_and_const_int): Change the type of the mode
17306 parameter to scalar_int_mode.
17307 (simplify_and_const_int_1): Likewise. Update recursive call.
17309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17310 Alan Hayward <alan.hayward@arm.com>
17311 David Sherwood <david.sherwood@arm.com>
17313 * combine.c (simplify_compare_const): Check that the mode is a
17314 scalar_int_mode (rather than VOIDmode) before testing its
17316 (simplify_comparison): Move COMPARISON_P handling out of the
17317 loop and restrict the latter part of the loop to scalar_int_modes.
17318 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17319 and when considering SUBREG_REGs. Use is_int_mode instead of
17320 checking GET_MODE_CLASS against MODE_INT.
17322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17323 Alan Hayward <alan.hayward@arm.com>
17324 David Sherwood <david.sherwood@arm.com>
17326 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17327 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17328 shift_unit_mode and for modes involved in scalar shifts.
17330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17331 Alan Hayward <alan.hayward@arm.com>
17332 David Sherwood <david.sherwood@arm.com>
17334 * combine.c (force_int_to_mode): New function, split out from...
17335 (force_to_mode): ...here. Keep xmode up-to-date and use it
17336 instead of GET_MODE (x).
17338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17339 Alan Hayward <alan.hayward@arm.com>
17340 David Sherwood <david.sherwood@arm.com>
17342 * optabs-query.h (extraction_insn::struct_mode): Change type to
17343 opt_scalar_int_mode and update comment.
17344 (extraction_insn::field_mode): Change type to scalar_int_mode.
17345 (extraction_insn::pos_mode): Likewise.
17346 * combine.c (make_extraction): Update accordingly.
17347 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17348 (get_optab_extraction_insn): Likewise.
17349 * recog.c (simplify_while_replacing): Likewise.
17350 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17351 parameter to opt_scalar_int_mode.
17353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17354 Alan Hayward <alan.hayward@arm.com>
17355 David Sherwood <david.sherwood@arm.com>
17357 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17358 to a scalar_int_mode instead of a machine_mode.
17359 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17360 (get_best_mode): Return a boolean and use a pointer argument to store
17361 the selected mode. Replace the limit mode parameter with a bit limit.
17362 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17363 for the values returned by bit_field_mode_iterator::next_mode.
17364 (store_bit_field): Update call to get_best_mode.
17365 (store_fixed_bit_field): Likewise.
17366 (extract_fixed_bit_field): Likewise.
17367 * expr.c (optimize_bitfield_assignment_op): Likewise.
17368 * fold-const.c (optimize_bit_field_compare): Likewise.
17369 (fold_truth_andor_1): Likewise.
17370 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17371 Update for new type of m_mode.
17372 (get_best_mode): As above.
17374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17375 Alan Hayward <alan.hayward@arm.com>
17376 David Sherwood <david.sherwood@arm.com>
17378 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17379 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17380 (store_bit_field): Check is_a <scalar_int_mode> before calling
17381 strict_volatile_bitfield_p.
17382 (extract_bit_field): Likewise.
17384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17385 Alan Hayward <alan.hayward@arm.com>
17386 David Sherwood <david.sherwood@arm.com>
17388 * target.def (cstore_mode): Return a scalar_int_mode.
17389 * doc/tm.texi: Regenerate.
17390 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17391 * targhooks.h (default_cstore_mode): Likewise.
17392 * targhooks.c (default_cstore_mode): Likewise, using a forced
17394 * expmed.c (emit_cstore): Expect the target of the cstore to be
17397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17398 Alan Hayward <alan.hayward@arm.com>
17399 David Sherwood <david.sherwood@arm.com>
17401 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17403 (niter_desc): Likewise mode.
17404 (iv_analyze): Add a mode parameter.
17406 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17407 and change its type to scalar_int_mode.
17408 * loop-iv.c: Update commentary at head of file.
17409 (iv_constant): Pass the mode paraeter before the rtx it describes
17410 and change its type to scalar_int_mode. Remove VOIDmode handling.
17411 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17412 (iv_extend): Likewise.
17413 (shorten_into_mode): Likewise.
17414 (iv_add): Use scalar_int_mode.
17415 (iv_mult): Likewise.
17416 (iv_shift): Likewise.
17417 (canonicalize_iv_subregs): Likewise.
17418 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17419 it describes and change its mode to scalar_int_mode. Also change
17420 the type of the returned inner_mode to scalar_int_mode.
17421 (get_biv_step): Likewise, turning outer_mode from a pointer
17422 into a direct parameter. Update call to get_biv_step_1.
17423 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17424 iv_constant and get_biv_step.
17425 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17426 and change its type to scalar_int_mode. Don't initialise iv->mode
17427 to VOIDmode and remove later checks for its still being VOIDmode.
17428 Update calls to iv_analyze_op and iv_analyze_expr. Check
17429 is_a <scalar_int_mode> when changing the mode under consideration.
17430 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17431 Update call to iv_analyze_expr.
17432 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17433 inner register is not also a scalar_int_mode. Update call to
17435 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17436 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17437 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17438 separate mode class checks. Update calls to iv_analyze. Remove
17439 fix-up of VOIDmodes after iv_analyze_biv.
17440 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17441 don't have a scalar_int_mode. Update call to biv_p.
17443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17444 Alan Hayward <alan.hayward@arm.com>
17445 David Sherwood <david.sherwood@arm.com>
17447 * cfgexpand.c (convert_debug_memory_address): Use
17448 as_a <scalar_int_mode>.
17449 * combine.c (expand_compound_operation): Likewise.
17450 (make_extraction): Likewise.
17451 (change_zero_ext): Likewise.
17452 (simplify_comparison): Likewise.
17453 * cse.c (cse_insn): Likewise.
17454 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17455 (mem_loc_descriptor): Likewise.
17456 (loc_descriptor): Likewise.
17457 * expmed.c (init_expmed_one_mode): Likewise.
17458 (synth_mult): Likewise.
17459 (emit_store_flag_1): Likewise.
17460 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17461 of a comparison with size.
17462 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17463 (reduce_to_bit_field_precision): Likewise.
17464 * function.c (expand_function_end): Likewise.
17465 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17466 * loop-doloop.c (doloop_modify): Likewise.
17467 * optabs.c (expand_binop): Likewise.
17468 (expand_unop): Likewise.
17469 (expand_copysign_absneg): Likewise.
17470 (prepare_cmp_insn): Likewise.
17471 (maybe_legitimize_operand): Likewise.
17472 * recog.c (const_scalar_int_operand): Likewise.
17473 * rtlanal.c (get_address_mode): Likewise.
17474 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17475 (simplify_cond_clz_ctz): Likewise.
17476 * tree-nested.c (get_nl_goto_field): Likewise.
17477 * tree.c (build_vector_type_for_mode): Likewise.
17478 * var-tracking.c (use_narrower_mode): Likewise.
17480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17481 Alan Hayward <alan.hayward@arm.com>
17482 David Sherwood <david.sherwood@arm.com>
17484 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17485 * builtins.c (expand_builtin_signbit): Use it.
17486 * cfgexpand.c (expand_debug_expr): Likewise.
17487 * dojump.c (do_jump): Likewise.
17488 (do_compare_and_jump): Likewise.
17489 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17490 * expmed.c (make_tree): Likewise.
17491 * expr.c (expand_expr_real_2): Likewise.
17492 (expand_expr_real_1): Likewise.
17493 (try_casesi): Likewise.
17494 * fold-const-call.c (fold_const_call_ss): Likewise.
17495 * fold-const.c (unextend): Likewise.
17496 (extract_muldiv_1): Likewise.
17497 (fold_single_bit_test): Likewise.
17498 (native_encode_int): Likewise.
17499 (native_encode_string): Likewise.
17500 (native_interpret_int): Likewise.
17501 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17502 * internal-fn.c (expand_addsub_overflow): Likewise.
17503 (expand_neg_overflow): Likewise.
17504 (expand_mul_overflow): Likewise.
17505 (expand_arith_overflow): Likewise.
17506 * match.pd: Likewise.
17507 * stor-layout.c (layout_type): Likewise.
17508 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17509 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17510 * tree-ssanames.c (get_range_info): Likewise.
17511 * tree-switch-conversion.c (array_value_type) Likewise.
17512 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17513 (vect_recog_divmod_pattern): Likewise.
17514 (vect_recog_mixed_size_cond_pattern): Likewise.
17515 * tree-vrp.c (extract_range_basic): Likewise.
17516 (simplify_float_conversion_using_ranges): Likewise.
17517 * tree.c (int_fits_type_p): Likewise.
17518 * ubsan.c (instrument_bool_enum_load): Likewise.
17519 * varasm.c (mergeable_string_section): Likewise.
17520 (narrowing_initializer_constant_valid_p): Likewise.
17521 (output_constant): Likewise.
17523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17524 Alan Hayward <alan.hayward@arm.com>
17525 David Sherwood <david.sherwood@arm.com>
17527 * machmode.h (NARROWEST_INT_MODE): New macro.
17528 * expr.c (alignment_for_piecewise_move): Use it instead of
17529 GET_CLASS_NARROWEST_MODE (MODE_INT).
17530 (push_block): Likewise.
17531 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17533 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17536 Alan Hayward <alan.hayward@arm.com>
17537 David Sherwood <david.sherwood@arm.com>
17539 * postreload.c (move2add_valid_value_p): Change the type of the
17540 mode parameter to scalar_int_mode.
17541 (move2add_use_add2_insn): Add a mode parameter and use it instead
17543 (move2add_use_add3_insn): Likewise.
17544 (reload_cse_move2add): Update accordingly.
17546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17547 Alan Hayward <alan.hayward@arm.com>
17548 David Sherwood <david.sherwood@arm.com>
17550 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17552 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17553 * optabs.c (expand_unop): Likewise.
17555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17556 Alan Hayward <alan.hayward@arm.com>
17557 David Sherwood <david.sherwood@arm.com>
17559 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17560 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17561 (popcount_loc_descriptor): Likewise.
17562 (bswap_loc_descriptor): Likewise.
17563 (rotate_loc_descriptor): Likewise.
17564 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17565 calling the functions above.
17567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17568 Alan Hayward <alan.hayward@arm.com>
17569 David Sherwood <david.sherwood@arm.com>
17571 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17573 (try_combine): Likewise.
17574 (simplify_if_then_else): Likewise.
17575 * cse.c (cse_insn): Likewise.
17576 * dwarf2out.c (mem_loc_descriptor): Likewise.
17577 * emit-rtl.c (gen_lowpart_common): Likewise.
17578 * simplify-rtx.c (simplify_truncation): Likewise.
17579 (simplify_binary_operation_1): Likewise.
17580 (simplify_const_relational_operation): Likewise.
17581 (simplify_ternary_operation): Likewise.
17582 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17585 Alan Hayward <alan.hayward@arm.com>
17586 David Sherwood <david.sherwood@arm.com>
17588 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17589 * reload.c (push_reload): Likewise.
17590 (find_reloads): Likewise.
17592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17593 Alan Hayward <alan.hayward@arm.com>
17594 David Sherwood <david.sherwood@arm.com>
17596 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17597 (make_compound_operation_int): Likewise.
17598 (change_zero_ext): Likewise.
17599 * expr.c (convert_move): Likewise.
17600 (convert_modes): Likewise.
17601 * fwprop.c (forward_propagate_subreg): Likewise.
17602 * loop-iv.c (get_biv_step_1): Likewise.
17603 * optabs.c (widen_operand): Likewise.
17604 * postreload.c (move2add_valid_value_p): Likewise.
17605 * recog.c (simplify_while_replacing): Likewise.
17606 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17607 (simplify_binary_operation_1): Likewise. Remove redundant
17608 mode equality check.
17610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * combine.c (combine_simplify_rtx): Add checks for
17615 is_a <scalar_int_mode>.
17616 (simplify_if_then_else): Likewise.
17617 (make_field_assignment): Likewise.
17618 (simplify_comparison): Likewise.
17619 * ifcvt.c (noce_try_bitop): Likewise.
17620 * loop-invariant.c (canonicalize_address_mult): Likewise.
17621 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17624 Alan Hayward <alan.hayward@arm.com>
17625 David Sherwood <david.sherwood@arm.com>
17627 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17628 is_a <scalar_int_mode> instead of != BLKmode.
17630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17631 Alan Hayward <alan.hayward@arm.com>
17632 David Sherwood <david.sherwood@arm.com>
17634 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17635 instead of != VOIDmode.
17636 * combine.c (if_then_else_cond): Likewise.
17637 (change_zero_ext): Likewise.
17638 * dwarf2out.c (mem_loc_descriptor): Likewise.
17639 (loc_descriptor): Likewise.
17640 * rtlanal.c (canonicalize_condition): Likewise.
17641 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17644 Alan Hayward <alan.hayward@arm.com>
17645 David Sherwood <david.sherwood@arm.com>
17647 * simplify-rtx.c (simplify_binary_operation_1): Use
17648 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17651 Alan Hayward <alan.hayward@arm.com>
17652 David Sherwood <david.sherwood@arm.com>
17654 * wide-int.h (int_traits<unsigned char>) New class.
17655 (int_traits<unsigned short>) Likewise.
17656 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17657 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17659 * combine.c (set_nonzero_bits_and_sign_copies): Use
17660 is_a <scalar_int_mode>.
17661 (find_split_point): Likewise.
17662 (combine_simplify_rtx): Likewise.
17663 (simplify_logical): Likewise.
17664 (expand_compound_operation): Likewise.
17665 (expand_field_assignment): Likewise.
17666 (make_compound_operation): Likewise.
17667 (extended_count): Likewise.
17668 (change_zero_ext): Likewise.
17669 (simplify_comparison): Likewise.
17670 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17671 (ucompare_loc_descriptor): Likewise.
17672 (minmax_loc_descriptor): Likewise.
17673 (mem_loc_descriptor): Likewise.
17674 (loc_descriptor): Likewise.
17675 * expmed.c (init_expmed_one_mode): Likewise.
17676 * lra-constraints.c (lra_constraint_offset): Likewise.
17677 * optabs.c (prepare_libcall_arg): Likewise.
17678 * postreload.c (move2add_note_store): Likewise.
17679 * reload.c (operands_match_p): Likewise.
17680 * rtl.h (load_extend_op): Likewise.
17681 * rtlhooks.c (gen_lowpart_general): Likewise.
17682 * simplify-rtx.c (simplify_truncation): Likewise.
17683 (simplify_unary_operation_1): Likewise.
17684 (simplify_binary_operation_1): Likewise.
17685 (simplify_const_binary_operation): Likewise.
17686 (simplify_const_relational_operation): Likewise.
17687 (simplify_subreg): Likewise.
17688 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17689 * var-tracking.c (adjust_mems): Likewise.
17690 (prepare_call_arguments): Likewise.
17692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17693 Alan Hayward <alan.hayward@arm.com>
17694 David Sherwood <david.sherwood@arm.com>
17696 * machmode.h (is_int_mode): New fuction.
17697 * combine.c (find_split_point): Use it.
17698 (combine_simplify_rtx): Likewise.
17699 (simplify_if_then_else): Likewise.
17700 (simplify_set): Likewise.
17701 (simplify_shift_const_1): Likewise.
17702 (simplify_comparison): Likewise.
17703 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17704 * cse.c (notreg_cost): Likewise.
17705 (cse_insn): Likewise.
17706 * cselib.c (cselib_lookup_1): Likewise.
17707 * dojump.c (do_jump_1): Likewise.
17708 (do_compare_rtx_and_jump): Likewise.
17709 * dse.c (get_call_args): Likewise.
17710 * dwarf2out.c (rtl_for_decl_init): Likewise.
17711 (native_encode_initializer): Likewise.
17712 * expmed.c (emit_store_flag_1): Likewise.
17713 (emit_store_flag): Likewise.
17714 * expr.c (convert_modes): Likewise.
17715 (store_field): Likewise.
17716 (expand_expr_real_1): Likewise.
17717 * fold-const.c (fold_read_from_constant_string): Likewise.
17718 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17719 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17720 * optabs.c (expand_binop): Likewise.
17721 (expand_unop): Likewise.
17722 (expand_abs_nojump): Likewise.
17723 (expand_one_cmpl_abs_nojump): Likewise.
17724 * simplify-rtx.c (mode_signbit_p): Likewise.
17725 (val_signbit_p): Likewise.
17726 (val_signbit_known_set_p): Likewise.
17727 (val_signbit_known_clear_p): Likewise.
17728 (simplify_relational_operation_1): Likewise.
17729 * tree.c (vector_type_mode): Likewise.
17731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17732 Alan Hayward <alan.hayward@arm.com>
17733 David Sherwood <david.sherwood@arm.com>
17735 * machmode.h (smallest_mode_for_size): Fix formatting.
17736 (smallest_int_mode_for_size): New function.
17737 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17738 instead of smallest_mode_for_size.
17739 * combine.c (make_extraction): Likewise.
17740 * config/arc/arc.c (arc_expand_movmem): Likewise.
17741 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17742 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17743 * config/s390/s390.c (s390_expand_insv): Likewise.
17744 * config/sparc/sparc.c (assign_int_registers): Likewise.
17745 * config/spu/spu.c (spu_function_value): Likewise.
17746 (spu_function_arg): Likewise.
17747 * coverage.c (get_gcov_type): Likewise.
17748 (get_gcov_unsigned_t): Likewise.
17749 * dse.c (find_shift_sequence): Likewise.
17750 * expmed.c (store_bit_field_1): Likewise.
17751 * expr.c (convert_move): Likewise.
17752 (store_field): Likewise.
17753 * internal-fn.c (expand_arith_overflow): Likewise.
17754 * optabs-query.c (get_best_extraction_insn): Likewise.
17755 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17756 * stor-layout.c (layout_type): Likewise.
17757 (initialize_sizetypes): Likewise.
17758 * targhooks.c (default_get_mask_mode): Likewise.
17759 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17762 Alan Hayward <alan.hayward@arm.com>
17763 David Sherwood <david.sherwood@arm.com>
17765 * machmode.h (opt_mode::else_blk): New function.
17766 (int_mode_for_mode): Declare.
17767 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17768 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17770 * cfgexpand.c (expand_debug_expr): Likewise.
17771 * combine.c (gen_lowpart_or_truncate): Likewise.
17772 (gen_lowpart_for_combine): Likewise.
17773 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17774 * config/avr/avr.c (avr_to_int_mode): Likewise.
17775 (avr_out_plus_1): Likewise.
17776 (avr_out_plus): Likewise.
17777 (avr_out_round): Likewise.
17778 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17779 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17780 (s390_expand_vcond): Likewise.
17781 * config/spu/spu.c (spu_split_immediate): Likewise.
17782 (spu_expand_mov): Likewise.
17783 * dse.c (get_stored_val): Likewise.
17784 * expmed.c (store_bit_field_1): Likewise.
17785 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17787 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17788 (extract_low_bits): Likewise.
17789 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17790 handling rather than repeating the check.
17791 (emit_group_store): Likewise.
17792 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17793 * optabs.c (expand_absneg_bit): Likewise.
17794 (expand_copysign_absneg): Likewise.
17795 (expand_copysign_bit): Likewise.
17796 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17797 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17798 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17799 * var-tracking.c (prepare_call_arguments): Likewise.
17800 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17801 int_mode_for_mode instead of mode_for_size.
17802 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17805 Alan Hayward <alan.hayward@arm.com>
17806 David Sherwood <david.sherwood@arm.com>
17808 * machmode.h (int_mode_for_size): New function.
17809 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17810 instead of mode_for_size.
17811 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17813 * combine.c (expand_field_assignment): Use int_mode_for_size
17814 instead of mode_for_size.
17815 (make_extraction): Likewise.
17816 (simplify_shift_const_1): Likewise.
17817 (simplify_comparison): Likewise.
17818 * dojump.c (do_jump): Likewise.
17819 * dwarf2out.c (mem_loc_descriptor): Likewise.
17820 * emit-rtl.c (init_derived_machine_modes): Likewise.
17821 * expmed.c (flip_storage_order): Likewise.
17822 (convert_extracted_bit_field): Likewise.
17823 * expr.c (copy_blkmode_from_reg): Likewise.
17824 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17825 * internal-fn.c (expand_mul_overflow): Likewise.
17826 * lower-subreg.c (simple_move): Likewise.
17827 * optabs-libfuncs.c (init_optabs): Likewise.
17828 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17829 * tree.c (vector_type_mode): Likewise.
17830 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17831 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17832 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17833 * tree-vect-stmts.c (vectorizable_load): Likewise.
17834 (vectorizable_store): Likewise.
17836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17837 Alan Hayward <alan.hayward@arm.com>
17838 David Sherwood <david.sherwood@arm.com>
17840 * coretypes.h (pod_mode): New type.
17841 (scalar_int_mode_pod): New typedef.
17842 * machmode.h (pod_mode): New class.
17843 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17844 * genmodes.c (emit_mode_int_n): Update accordingly.
17845 * lower-subreg.h (target_lower_subreg): Change type to
17846 scalar_int_mode_pod.
17847 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17848 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17855 machine_mode to scalar_int_mode.
17856 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17857 (rs6000_option_override_internal): Remove cast to int.
17858 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17859 machine_mode to scalar_int_mode.
17860 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17861 (rs6000_option_override_internal): Remove cast to int.
17862 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17863 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17865 * config/s390/s390.c (s390_expand_builtin): Likewise.
17866 * coretypes.h (scalar_int_mode): New type.
17867 (opt_scalar_int_mode): New typedef.
17868 * machmode.h (scalar_int_mode): New class.
17869 (scalar_int_mode::includes_p): New function.
17870 (byte_mode): Change type to scalar_int_mode.
17871 (word_mode): Likewise.
17872 (ptr_mode): Likewise.
17873 * emit-rtl.c (byte_mode): Likewise.
17874 (word_mode): Likewise.
17875 (ptr_mode): Likewise.
17876 (init_derived_machine_modes): Update accordingly.
17877 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17878 and MODE_PARTIAL_INT.
17879 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17880 opt_scalar_int_mode.
17882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17883 Alan Hayward <alan.hayward@arm.com>
17884 David Sherwood <david.sherwood@arm.com>
17886 * target.def (libgcc_floating_mode_supported_p): Take a
17888 * doc/tm.texi: Regenerate.
17889 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17891 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17892 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17896 Alan Hayward <alan.hayward@arm.com>
17897 David Sherwood <david.sherwood@arm.com>
17899 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17900 * doc/tm.texi: Regenerate.
17901 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17902 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17903 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17904 * targhooks.h (default_floatn_mode): Likewise.
17905 * targhooks.c (default_floatn_mode): Likewise.
17906 * tree.c (build_common_tree_nodes): Update accordingly.
17908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17909 Alan Hayward <alan.hayward@arm.com>
17910 David Sherwood <david.sherwood@arm.com>
17912 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17913 (mode_iterator::iterate_p): Likewise.
17914 (mode_iterator::get_wider): Likewise.
17915 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17918 Alan Hayward <alan.hayward@arm.com>
17919 David Sherwood <david.sherwood@arm.com>
17921 * coretypes.h (opt_scalar_float_mode): New typedef.
17922 * machmode.h (float_mode_for_size): New function.
17923 * emit-rtl.c (double_mode): Delete.
17924 (init_emit_once): Use float_mode_for_size.
17925 * stor-layout.c (layout_type): Likewise.
17926 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17929 Alan Hayward <alan.hayward@arm.com>
17930 David Sherwood <david.sherwood@arm.com>
17932 * output.h (assemble_real): Take a scalar_float_mode.
17933 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17934 * config/arm/arm.md (consttable_4): Likewise.
17935 (consttable_8): Likewise.
17936 (consttable_16): Likewise.
17937 * config/mips/mips.md (consttable_float): Likewise.
17938 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17939 * varasm.c (assemble_real): Take a scalar_float_mode.
17940 (output_constant_pool_2): Update accordingly.
17941 (output_constant): Likewise.
17943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17944 Alan Hayward <alan.hayward@arm.com>
17945 David Sherwood <david.sherwood@arm.com>
17947 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17948 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17949 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17950 (native_encode_real): Likewise.
17951 (native_interpret_real): Likewise.
17952 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17953 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17956 Alan Hayward <alan.hayward@arm.com>
17957 David Sherwood <david.sherwood@arm.com>
17959 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17960 <scalar_float_mode>. Simplify.
17961 (gen_extend_conv_libfunc): Likewise.
17963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17964 Alan Hayward <alan.hayward@arm.com>
17965 David Sherwood <david.sherwood@arm.com>
17967 * coretypes.h (scalar_float_mode): New type.
17968 * machmode.h (mode_traits::from_int): Use machine_mode if
17969 USE_ENUM_MODES is defined.
17970 (is_a): New function.
17972 (dyn_cast): Likewise.
17973 (scalar_float_mode): New class.
17974 (scalar_float_mode::includes_p): New function.
17975 (is_float_mode): Likewise.
17976 * gdbhooks.py (MachineModePrinter): New class.
17977 (build_pretty_printer): Use it for scalar_float_mode.
17978 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17979 (format_helper::format_helper): Turn into a template.
17980 * genmodes.c (get_mode_class): New function.
17981 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17982 or machine_mode if none.
17983 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17984 as_a <scalar_float_mode>.
17985 * dwarf2out.c (mem_loc_descriptor): Likewise.
17986 (insert_float): Likewise.
17987 (add_const_value_attribute): Likewise.
17988 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17989 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17990 (expand_unop): Update accordingly.
17991 (expand_abs_nojump): Likewise.
17992 (expand_copysign_absneg): Take a scalar_float_mode.
17993 (expand_copysign_bit): Likewise.
17994 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
18001 * machmode.h (opt_mode): Likewise.
18002 (opt_mode::else_void): New function.
18003 (opt_mode::require): Likewise.
18004 (opt_mode::exists): Likewise.
18005 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18006 (GET_MODE_2XWIDER_MODE): Likewise.
18007 (mode_iterator::get_wider): Update accordingly.
18008 (mode_iterator::get_2xwider): Likewise.
18009 (mode_iterator::get_known_wider): Likewise, turning into a template.
18010 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18011 forcing a wider mode to exist.
18012 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18013 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18014 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18015 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18016 * lower-subreg.c (init_lower_subreg): Likewise.
18017 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18018 on the final iteration.
18019 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18020 a wider mode exists before asking for a move pattern.
18021 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18022 forcing a wider mode to exist.
18023 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18024 returning false if no such mode exists.
18025 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18026 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18027 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18028 Avoid checking for a MODE_INT if we already know the mode is not a
18030 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18031 forcing a wider mode to exist.
18032 (expmed_mult_highpart_optab): Likewise.
18033 (expmed_mult_highpart): Likewise.
18034 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18036 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18037 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18038 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18039 * internal-fn.c (expand_mul_overflow): Update use of
18040 GET_MODE_2XWIDER_MODE.
18041 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18042 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18043 GET_MODE_WIDER_MODE.
18044 (convert_plusminus_to_widen): Likewise.
18045 * tree-switch-conversion.c (array_value_type): Likewise.
18046 * var-tracking.c (emit_note_insn_var_location): Likewise.
18047 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18048 Return false inside rather than outside the loop if no wider mode
18050 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18051 and GET_MODE_2XWIDER_MODE
18052 (can_compare_p): Use else_void.
18053 * gdbhooks.py (OptMachineModePrinter): New class.
18054 (build_pretty_printer): Use it for opt_mode.
18056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18057 Alan Hayward <alan.hayward@arm.com>
18058 David Sherwood <david.sherwood@arm.com>
18060 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18061 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18064 Alan Hayward <alan.hayward@arm.com>
18065 David Sherwood <david.sherwood@arm.com>
18067 * machmode.h (mode_traits): New structure.
18068 (get_narrowest_mode): New function.
18069 (mode_iterator::start): Likewise.
18070 (mode_iterator::iterate_p): Likewise.
18071 (mode_iterator::get_wider): Likewise.
18072 (mode_iterator::get_known_wider): Likewise.
18073 (mode_iterator::get_2xwider): Likewise.
18074 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18075 (FOR_EACH_MODE): Likewise.
18076 (FOR_EACH_MODE_FROM): Likewise.
18077 (FOR_EACH_MODE_UNTIL): Likewise.
18078 (FOR_EACH_WIDER_MODE): Likewise.
18079 (FOR_EACH_2XWIDER_MODE): Likewise.
18080 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18081 * combine.c (simplify_comparison): Likewise
18082 * config/i386/i386.c (type_natural_mode): Likewise.
18083 * cse.c (cse_insn): Likewise.
18084 * dse.c (find_shift_sequence): Likewise.
18085 * emit-rtl.c (init_derived_machine_modes): Likewise.
18086 (init_emit_once): Likewise.
18087 * explow.c (hard_function_value): Likewise.
18088 * expmed.c (extract_fixed_bit_field_1): Likewise.
18089 (extract_bit_field_1): Likewise.
18090 (expand_divmod): Likewise.
18091 (emit_store_flag_1): Likewise.
18092 * expr.c (init_expr_target): Likewise.
18093 (convert_move): Likewise.
18094 (alignment_for_piecewise_move): Likewise.
18095 (widest_int_mode_for_size): Likewise.
18096 (emit_block_move_via_movmem): Likewise.
18097 (copy_blkmode_to_reg): Likewise.
18098 (set_storage_via_setmem): Likewise.
18099 (compress_float_constant): Likewise.
18100 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18101 * optabs-query.c (get_best_extraction_insn): Likewise.
18102 * optabs.c (expand_binop): Likewise.
18103 (expand_twoval_unop): Likewise.
18104 (expand_twoval_binop): Likewise.
18105 (widen_leading): Likewise.
18106 (widen_bswap): Likewise.
18107 (expand_parity): Likewise.
18108 (expand_unop): Likewise.
18109 (prepare_cmp_insn): Likewise.
18110 (prepare_float_lib_cmp): Likewise.
18111 (expand_float): Likewise.
18112 (expand_fix): Likewise.
18113 (expand_sfix_optab): Likewise.
18114 * postreload.c (move2add_use_add2_insn): Likewise.
18115 * reg-stack.c (reg_to_stack): Likewise.
18116 * reginfo.c (choose_hard_reg_mode): Likewise.
18117 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18118 * stor-layout.c (mode_for_size): Likewise.
18119 (smallest_mode_for_size): Likewise.
18120 (mode_for_vector): Likewise.
18121 (finish_bitfield_representative): Likewise.
18122 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18123 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18124 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18125 * var-tracking.c (prepare_call_arguments): Likewise.
18127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18128 Alan Hayward <alan.hayward@arm.com>
18129 David Sherwood <david.sherwood@arm.com>
18131 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18132 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18133 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18134 * machmode.h (mode_size): Move earlier in file.
18135 (mode_precision): Likewise.
18136 (mode_inner): Likewise.
18137 (mode_nunits): Likewise.
18138 (mode_unit_size): Likewise.
18139 (unit_unit_precision): Likewise.
18140 (mode_wider): Likewise.
18141 (mode_2xwider): Likewise.
18142 (machine_mode): New class.
18143 (mode_to_bytes): New function.
18144 (mode_to_bits): Likewise.
18145 (mode_to_precision): Likewise.
18146 (mode_to_inner): Likewise.
18147 (mode_to_unit_size): Likewise.
18148 (mode_to_unit_precision): Likewise.
18149 (mode_to_nunits): Likewise.
18150 (GET_MODE_SIZE): Use mode_to_bytes.
18151 (GET_MODE_BITSIZE): Use mode_to_bits.
18152 (GET_MODE_PRECISION): Use mode_to_precision.
18153 (GET_MODE_INNER): Use mode_to_inner.
18154 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18155 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18156 (GET_MODE_NUNITS): Use mode_to_nunits.
18157 * system.h (ALWAYS_INLINE): New macro.
18158 * config/powerpcspe/powerpcspe-c.c
18159 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18160 int for arg1_mode and arg2_mode.
18162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18163 Alan Hayward <alan.hayward@arm.com>
18164 David Sherwood <david.sherwood@arm.com>
18166 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18167 Prefix mode names with E_ in case statements.
18168 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18169 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18170 (aarch64_split_simd_move): Likewise.
18171 (aarch64_gen_storewb_pair): Likewise.
18172 (aarch64_gen_loadwb_pair): Likewise.
18173 (aarch64_gen_store_pair): Likewise.
18174 (aarch64_gen_load_pair): Likewise.
18175 (aarch64_get_condition_code_1): Likewise.
18176 (aarch64_constant_pool_reload_icode): Likewise.
18177 (get_rsqrte_type): Likewise.
18178 (get_rsqrts_type): Likewise.
18179 (get_recpe_type): Likewise.
18180 (get_recps_type): Likewise.
18181 (aarch64_gimplify_va_arg_expr): Likewise.
18182 (aarch64_simd_container_mode): Likewise.
18183 (aarch64_emit_load_exclusive): Likewise.
18184 (aarch64_emit_store_exclusive): Likewise.
18185 (aarch64_expand_compare_and_swap): Likewise.
18186 (aarch64_gen_atomic_cas): Likewise.
18187 (aarch64_emit_bic): Likewise.
18188 (aarch64_emit_atomic_swap): Likewise.
18189 (aarch64_emit_atomic_load_op): Likewise.
18190 (aarch64_evpc_trn): Likewise.
18191 (aarch64_evpc_uzp): Likewise.
18192 (aarch64_evpc_zip): Likewise.
18193 (aarch64_evpc_ext): Likewise.
18194 (aarch64_evpc_rev): Likewise.
18195 (aarch64_evpc_dup): Likewise.
18196 (aarch64_gen_ccmp_first): Likewise.
18197 (aarch64_gen_ccmp_next): Likewise.
18198 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18199 (alpha_emit_xfloating_libcall): Likewise.
18200 (emit_insxl): Likewise.
18201 (alpha_arg_type): Likewise.
18202 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18203 (arc_preferred_simd_mode): Likewise.
18204 (arc_secondary_reload): Likewise.
18205 (get_arc_condition_code): Likewise.
18206 (arc_print_operand): Likewise.
18207 (arc_legitimate_constant_p): Likewise.
18208 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18209 * config/arc/arc.md (casesi_load): Likewise.
18210 (casesi_compact_jump): Likewise.
18211 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18212 (cc_use_register): Likewise.
18213 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18214 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18215 (arm_init_iwmmxt_builtins): Likewise.
18216 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18217 (neon_expand_vector_init): Likewise.
18218 (arm_attr_length_move_neon): Likewise.
18219 (maybe_get_arm_condition_code): Likewise.
18220 (arm_emit_vector_const): Likewise.
18221 (arm_preferred_simd_mode): Likewise.
18222 (arm_output_iwmmxt_tinsr): Likewise.
18223 (thumb1_output_casesi): Likewise.
18224 (thumb2_output_casesi): Likewise.
18225 (arm_emit_load_exclusive): Likewise.
18226 (arm_emit_store_exclusive): Likewise.
18227 (arm_expand_compare_and_swap): Likewise.
18228 (arm_evpc_neon_vuzp): Likewise.
18229 (arm_evpc_neon_vzip): Likewise.
18230 (arm_evpc_neon_vrev): Likewise.
18231 (arm_evpc_neon_vtrn): Likewise.
18232 (arm_evpc_neon_vext): Likewise.
18233 (arm_validize_comparison): Likewise.
18234 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18235 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18236 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18237 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18238 (c6x_preferred_simd_mode): Likewise.
18239 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18240 (epiphany_rtx_costs): Likewise.
18241 * config/epiphany/predicates.md (proper_comparison_operator):
18243 * config/frv/frv.c (condexec_memory_operand): Likewise.
18244 (frv_emit_move): Likewise.
18245 (output_move_single): Likewise.
18246 (output_condmove_single): Likewise.
18247 (frv_hard_regno_mode_ok): Likewise.
18248 (frv_matching_accg_mode): Likewise.
18249 * config/h8300/h8300.c (split_adds_subs): Likewise.
18250 (h8300_rtx_costs): Likewise.
18251 (h8300_print_operand): Likewise.
18252 (compute_mov_length): Likewise.
18253 (output_logical_op): Likewise.
18254 (compute_logical_op_length): Likewise.
18255 (compute_logical_op_cc): Likewise.
18256 (h8300_shift_needs_scratch_p): Likewise.
18257 (output_a_shift): Likewise.
18258 (compute_a_shift_length): Likewise.
18259 (compute_a_shift_cc): Likewise.
18260 (expand_a_rotate): Likewise.
18261 (output_a_rotate): Likewise.
18262 * config/i386/i386.c (classify_argument): Likewise.
18263 (function_arg_advance_32): Likewise.
18264 (function_arg_32): Likewise.
18265 (function_arg_64): Likewise.
18266 (function_value_64): Likewise.
18267 (ix86_gimplify_va_arg): Likewise.
18268 (ix86_legitimate_constant_p): Likewise.
18269 (put_condition_code): Likewise.
18270 (split_double_mode): Likewise.
18271 (ix86_avx256_split_vector_move_misalign): Likewise.
18272 (ix86_expand_vector_logical_operator): Likewise.
18273 (ix86_split_idivmod): Likewise.
18274 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18275 (ix86_build_const_vector): Likewise.
18276 (ix86_build_signbit_mask): Likewise.
18277 (ix86_match_ccmode): Likewise.
18278 (ix86_cc_modes_compatible): Likewise.
18279 (ix86_expand_branch): Likewise.
18280 (ix86_expand_sse_cmp): Likewise.
18281 (ix86_expand_sse_movcc): Likewise.
18282 (ix86_expand_int_sse_cmp): Likewise.
18283 (ix86_expand_vec_perm_vpermi2): Likewise.
18284 (ix86_expand_vec_perm): Likewise.
18285 (ix86_expand_sse_unpack): Likewise.
18286 (ix86_expand_int_addcc): Likewise.
18287 (ix86_split_to_parts): Likewise.
18288 (ix86_vectorize_builtin_gather): Likewise.
18289 (ix86_vectorize_builtin_scatter): Likewise.
18290 (avx_vpermilp_parallel): Likewise.
18291 (inline_memory_move_cost): Likewise.
18292 (ix86_tieable_integer_mode_p): Likewise.
18293 (x86_maybe_negate_const_int): Likewise.
18294 (ix86_expand_vector_init_duplicate): Likewise.
18295 (ix86_expand_vector_init_one_nonzero): Likewise.
18296 (ix86_expand_vector_init_one_var): Likewise.
18297 (ix86_expand_vector_init_concat): Likewise.
18298 (ix86_expand_vector_init_interleave): Likewise.
18299 (ix86_expand_vector_init_general): Likewise.
18300 (ix86_expand_vector_set): Likewise.
18301 (ix86_expand_vector_extract): Likewise.
18302 (emit_reduc_half): Likewise.
18303 (ix86_emit_i387_round): Likewise.
18304 (ix86_mangle_type): Likewise.
18305 (ix86_expand_round_sse4): Likewise.
18306 (expand_vec_perm_blend): Likewise.
18307 (canonicalize_vector_int_perm): Likewise.
18308 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18309 (expand_vec_perm_1): Likewise.
18310 (expand_vec_perm_interleave3): Likewise.
18311 (expand_vec_perm_even_odd_pack): Likewise.
18312 (expand_vec_perm_even_odd_1): Likewise.
18313 (expand_vec_perm_broadcast_1): Likewise.
18314 (ix86_vectorize_vec_perm_const_ok): Likewise.
18315 (ix86_expand_vecop_qihi): Likewise.
18316 (ix86_expand_mul_widen_hilo): Likewise.
18317 (ix86_expand_sse2_abs): Likewise.
18318 (ix86_expand_pextr): Likewise.
18319 (ix86_expand_pinsr): Likewise.
18320 (ix86_preferred_simd_mode): Likewise.
18321 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18322 * config/i386/sse.md (*andnot<mode>3): Likewise.
18323 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18324 (*<code><mode>3): Likewise.
18325 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18326 (ia64_expand_atomic_op): Likewise.
18327 (ia64_arg_type): Likewise.
18328 (ia64_mode_to_int): Likewise.
18329 (ia64_scalar_mode_supported_p): Likewise.
18330 (ia64_vector_mode_supported_p): Likewise.
18331 (expand_vec_perm_broadcast): Likewise.
18332 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18333 (iq2000_function_arg_advance): Likewise.
18334 (iq2000_function_arg): Likewise.
18335 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18336 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18337 (m68k_libcall_value): Likewise.
18338 (m68k_function_value): Likewise.
18339 (sched_attr_op_type): Likewise.
18340 * config/mcore/mcore.c (mcore_output_move): Likewise.
18341 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18343 (microblaze_function_arg): Likewise.
18344 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18345 (mips_print_operand): Likewise.
18346 (mips_mode_ok_for_mov_fmt_p): Likewise.
18347 (mips_vector_mode_supported_p): Likewise.
18348 (mips_preferred_simd_mode): Likewise.
18349 (mips_expand_vpc_loongson_even_odd): Likewise.
18350 (mips_expand_vec_unpack): Likewise.
18351 (mips_expand_vi_broadcast): Likewise.
18352 (mips_expand_vector_init): Likewise.
18353 (mips_expand_vec_reduc): Likewise.
18354 (mips_expand_msa_cmp): Likewise.
18355 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18356 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18357 (cc_flags_for_mode): Likewise.
18358 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18359 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18360 (nds32_output_casesi_pc_relative): Likewise.
18361 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18362 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18363 (nvptx_gen_unpack): Likewise.
18364 (nvptx_gen_pack): Likewise.
18365 (nvptx_gen_shuffle): Likewise.
18366 (nvptx_gen_wcast): Likewise.
18367 (nvptx_preferred_simd_mode): Likewise.
18368 * config/pa/pa.c (pa_secondary_reload): Likewise.
18369 * config/pa/predicates.md (base14_operand): Likewise.
18370 * config/powerpcspe/powerpcspe-c.c
18371 (altivec_resolve_overloaded_builtin): Likewise.
18372 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18374 (rs6000_preferred_simd_mode): Likewise.
18375 (output_vec_const_move): Likewise.
18376 (rs6000_expand_vector_extract): Likewise.
18377 (rs6000_split_vec_extract_var): Likewise.
18378 (reg_offset_addressing_ok_p): Likewise.
18379 (rs6000_legitimate_offset_address_p): Likewise.
18380 (rs6000_legitimize_address): Likewise.
18381 (rs6000_emit_set_const): Likewise.
18382 (rs6000_const_vec): Likewise.
18383 (rs6000_emit_move): Likewise.
18384 (spe_build_register_parallel): Likewise.
18385 (rs6000_darwin64_record_arg_recurse): Likewise.
18386 (swap_selector_for_mode): Likewise.
18387 (spe_init_builtins): Likewise.
18388 (paired_init_builtins): Likewise.
18389 (altivec_init_builtins): Likewise.
18390 (do_load_for_compare): Likewise.
18391 (rs6000_generate_compare): Likewise.
18392 (rs6000_expand_float128_convert): Likewise.
18393 (emit_load_locked): Likewise.
18394 (emit_store_conditional): Likewise.
18395 (rs6000_output_function_epilogue): Likewise.
18396 (rs6000_handle_altivec_attribute): Likewise.
18397 (rs6000_function_value): Likewise.
18398 (emit_fusion_gpr_load): Likewise.
18399 (emit_fusion_p9_load): Likewise.
18400 (emit_fusion_p9_store): Likewise.
18401 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18402 (fusion_gpr_mem_load): Likewise.
18403 (fusion_addis_mem_combo_load): Likewise.
18404 (fusion_addis_mem_combo_store): Likewise.
18405 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18406 (fusion_gpr_mem_load): Likewise.
18407 (fusion_addis_mem_combo_load): Likewise.
18408 (fusion_addis_mem_combo_store): Likewise.
18409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18411 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18412 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18413 (rs6000_preferred_simd_mode): Likewise.
18414 (output_vec_const_move): Likewise.
18415 (rs6000_expand_vector_extract): Likewise.
18416 (rs6000_split_vec_extract_var): Likewise.
18417 (reg_offset_addressing_ok_p): Likewise.
18418 (rs6000_legitimate_offset_address_p): Likewise.
18419 (rs6000_legitimize_address): Likewise.
18420 (rs6000_emit_set_const): Likewise.
18421 (rs6000_const_vec): Likewise.
18422 (rs6000_emit_move): Likewise.
18423 (rs6000_darwin64_record_arg_recurse): Likewise.
18424 (swap_selector_for_mode): Likewise.
18425 (paired_init_builtins): Likewise.
18426 (altivec_init_builtins): Likewise.
18427 (rs6000_expand_float128_convert): Likewise.
18428 (emit_load_locked): Likewise.
18429 (emit_store_conditional): Likewise.
18430 (rs6000_output_function_epilogue): Likewise.
18431 (rs6000_handle_altivec_attribute): Likewise.
18432 (rs6000_function_value): Likewise.
18433 (emit_fusion_gpr_load): Likewise.
18434 (emit_fusion_p9_load): Likewise.
18435 (emit_fusion_p9_store): Likewise.
18436 * config/rx/rx.c (rx_gen_move_template): Likewise.
18437 (flags_from_mode): Likewise.
18438 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18439 (s390_slb_comparison): Likewise.
18440 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18441 (s390_vector_mode_supported_p): Likewise.
18442 (s390_cc_modes_compatible): Likewise.
18443 (s390_match_ccmode_set): Likewise.
18444 (s390_canonicalize_comparison): Likewise.
18445 (s390_emit_compare_and_swap): Likewise.
18446 (s390_branch_condition_mask): Likewise.
18447 (s390_rtx_costs): Likewise.
18448 (s390_secondary_reload): Likewise.
18449 (__SECONDARY_RELOAD_CASE): Likewise.
18450 (s390_expand_cs): Likewise.
18451 (s390_preferred_simd_mode): Likewise.
18452 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18453 * config/sh/sh.c (sh_print_operand): Likewise.
18454 (dump_table): Likewise.
18455 (sh_secondary_reload): Likewise.
18456 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18457 * config/sh/sh.md (casesi_worker_1): Likewise.
18458 (casesi_worker_2): Likewise.
18459 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18460 (fcc_comparison_operator): Likewise.
18461 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18462 (emit_soft_tfmode_cvt): Likewise.
18463 (sparc_preferred_simd_mode): Likewise.
18464 (output_cbranch): Likewise.
18465 (sparc_print_operand): Likewise.
18466 (sparc_expand_vec_perm_bmask): Likewise.
18467 (vector_init_bshuffle): Likewise.
18468 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18469 (spu_vector_mode_supported_p): Likewise.
18470 (spu_expand_insv): Likewise.
18471 (spu_emit_branch_or_set): Likewise.
18472 (spu_handle_vector_attribute): Likewise.
18473 (spu_builtin_splats): Likewise.
18474 (spu_builtin_extract): Likewise.
18475 (spu_builtin_promote): Likewise.
18476 (spu_expand_sign_extend): Likewise.
18477 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18478 (tilegx_simd_int): Likewise.
18479 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18480 (tilepro_simd_int): Likewise.
18481 * config/v850/v850.c (const_double_split): Likewise.
18482 (v850_print_operand): Likewise.
18483 (ep_memory_offset): Likewise.
18484 * config/vax/vax.c (vax_rtx_costs): Likewise.
18485 (vax_output_int_move): Likewise.
18486 (vax_output_int_add): Likewise.
18487 (vax_output_int_subtract): Likewise.
18488 * config/visium/predicates.md (visium_branch_operator): Likewise.
18489 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18490 (visium_print_operand_address): Likewise.
18491 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18492 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18493 (xtensa_expand_conditional_branch): Likewise.
18494 (xtensa_copy_incoming_a7): Likewise.
18495 (xtensa_output_literal): Likewise.
18496 * dfp.c (decimal_real_maxval): Likewise.
18497 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18500 Alan Hayward <alan.hayward@arm.com>
18501 David Sherwood <david.sherwood@arm.com>
18503 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18504 (mode_nunits_inline): Likewise.
18505 (mode_inner_inline): Likewise.
18506 (mode_unit_size_inline): Likewise.
18507 (mode_unit_precision_inline): Likewise.
18508 (emit_insn_modes_h): Likewise. Also emit a #define of the
18510 (emit_mode_wider): Add an E_ prefix to mode names.
18511 (emit_mode_complex): Likewise.
18512 (emit_mode_inner): Likewise.
18513 (emit_mode_adjustments): Likewise.
18514 (emit_mode_int_n): Likewise.
18515 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18516 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18517 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18518 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18519 (CRC32_BUILTIN, ENTRY): Likewise.
18520 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18521 (aarch64_pop_regs): Likewise.
18522 (aarch64_process_components): Likewise.
18523 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18524 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18525 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18526 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18527 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18528 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18530 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18531 (pcmpistr): Likewise.
18532 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18533 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18534 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18536 * config/rl78/rl78.c (mduc_regs): Likewise.
18537 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18538 (htm_expand_builtin): Likewise.
18539 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18540 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18541 * config/xtensa/xtensa.c (print_operand): Likewise.
18542 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18543 (NUM_MODE_VECTOR_INT): Likewise.
18544 * genoutput.c (null_operand): Likewise.
18545 (output_operand_data): Likewise.
18546 * genrecog.c (print_parameter_value): Likewise.
18547 * lra.c (debug_operand_data): Likewise.
18549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18551 * dwarf2out.c (scompare_loc_descriptor_wide)
18552 (scompare_loc_descriptor_narrow): New functions, split out from...
18553 (scompare_loc_descriptor): ...here.
18554 * expmed.c (emit_store_flag_int): New function, split out from...
18555 (emit_store_flag): ...here.
18557 2017-08-30 Richard Biener <rguenther@suse.de>
18559 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18560 (dwarf2out_early_finish): Move setting of AT_pubnames from
18561 early debug output to early finish.
18563 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18565 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18566 and -mdata-region to the assembler.
18568 2017-08-30 Richard Biener <rguenther@suse.de>
18570 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18572 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18573 (dwarf2out_early_global_decl): Only generate a DIE for the
18574 abstract origin if it doesn't already exist or is a declaration DIE.
18575 (resolve_addr): Do not add the linkage name twice when
18576 generating a stub DIE for the DW_TAG_GNU_call_site target.
18578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18580 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18581 Use machine_mode rather than int for arg1_mode.
18583 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18586 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18587 that the second argument of the built-in functions to unpack
18588 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18589 switch statement instead a lot of if statements.
18590 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18591 Allow 64-bit values to be in Altivec registers as well as
18592 traditional floating point registers.
18593 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18595 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18597 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18598 MAX_REGS_PER_ADDRESS == 1.
18600 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18602 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18603 * config/i386/i386.c (ix86_option_override_internal): Simplify
18604 setting of opts->x_flag_entry.
18606 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18607 Jakub Jelinek <jakub@redhat.com>
18608 Richard Biener <rguenther@suse.de>
18610 PR tree-optimization/81503
18611 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18612 folded constant fits in the target type; reorder tests for clarity.
18614 2017-08-29 Martin Liska <mliska@suse.cz>
18616 * passes.def: Include pass_lower_switch.
18617 * stmt.c (dump_case_nodes): Remove and move to
18618 tree-switch-conversion.
18619 (case_values_threshold): Likewise.
18620 (expand_switch_as_decision_tree_p): Likewise.
18621 (emit_case_decision_tree): Likewise.
18622 (expand_case): Likewise.
18623 (balance_case_nodes): Likewise.
18624 (node_has_low_bound): Likewise.
18625 (node_has_high_bound): Likewise.
18626 (node_is_bounded): Likewise.
18627 (emit_case_nodes): Likewise.
18628 (struct simple_case_node): New struct.
18629 (add_case_node): Remove.
18630 (emit_case_dispatch_table): Use vector instead of case_list.
18631 (reset_out_edges_aux): Remove.
18632 (compute_cases_per_edge): Likewise.
18633 (expand_case): Build list of simple_case_node.
18634 (expand_sjlj_dispatch_table): Use it.
18635 * tree-switch-conversion.c (struct case_node): Moved from
18636 stmt.c and adjusted.
18637 (emit_case_nodes): Likewise.
18638 (node_has_low_bound): Likewise.
18639 (node_has_high_bound): Likewise.
18640 (node_is_bounded): Likewise.
18641 (case_values_threshold): Likewise.
18642 (reset_out_edges_aux): Likewise.
18643 (compute_cases_per_edge): Likewise.
18644 (add_case_node): Likewise.
18645 (dump_case_nodes): Likewise.
18646 (balance_case_nodes): Likewise.
18647 (expand_switch_as_decision_tree_p): Likewise.
18648 (emit_jump): Likewise.
18649 (emit_case_decision_tree): Likewise.
18650 (try_switch_expansion): Likewise.
18651 (do_jump_if_equal): Likewise.
18652 (emit_cmp_and_jump_insns): Likewise.
18653 (fix_phi_operands_for_edge): New function.
18654 (record_phi_operand_mapping): Likewise.
18655 (class pass_lower_switch): New pass.
18656 (pass_lower_switch::execute): New function.
18657 (make_pass_lower_switch): Likewise.
18658 (conditional_probability):
18659 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18660 * tree-pass.h: Add make_pass_lower_switch.
18662 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18665 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18668 2017-08-29 Richard Biener <rguenther@suse.de>
18670 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18671 we do not add a DW_AT_inline attribute twice.
18672 (gen_subprogram_die): Remove code setting DW_AT_inline on
18673 DECL_ABSTRACT_P nodes.
18675 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18677 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18678 calls to internal functions.
18679 (gimplify_modify_expr): Likewise.
18680 * tree-call-cdce.c (use_internal_fn): Likewise.
18681 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18682 (convert_to_divmod): Set the nothrow flag.
18683 * tree-if-conv.c (predicate_mem_writes): Likewise.
18684 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18685 (vectorizable_call): Likewise.
18686 (vectorizable_store): Likewise.
18687 (vectorizable_load): Likewise.
18688 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18689 (vect_recog_mask_conversion_pattern): Likewise.
18691 2017-08-29 Martin Liska <mliska@suse.cz>
18694 * gcc.c (driver_handle_option): Add new argument.
18695 * opts-common.c (handle_option): Pass
18696 target_option_override_hook.
18697 * opts-global.c (lang_handle_option): Add new option.
18698 (set_default_handlers): Add new argument.
18699 (decode_options): Likewise.
18700 * opts.c (target_handle_option): Likewise.
18701 (common_handle_option): Call target_option_override_hook.
18702 * opts.h (struct cl_option_handler_func): Add hook for
18703 target option override.
18704 (struct cl_option_handlers): Likewise.
18705 (set_default_handlers): Add new argument.
18706 (decode_options): Likewise.
18707 (common_handle_option): Likewise.
18708 (target_handle_option): Likewise.
18709 * toplev.c (toplev::main): Pass targetm.target_option.override
18712 2017-08-29 Richard Biener <rguenther@suse.de>
18713 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18715 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18716 life to the active subtree.
18718 2017-08-28 Jeff Law <law@redhat.com>
18720 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18721 derive_equivalences.
18722 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18723 Code moved into....
18724 (edge_info::derive_equivalences): New private member function
18726 * tree-ssa-dom.c (class edge_info): Changed from a struct
18727 to a class. Add ctor/dtor, methods and data members.
18728 (edge_info::edge_info): Renamed from allocate_edge_info.
18729 Initialize additional members.
18730 (edge_info::~edge_info): New.
18731 (free_dom_edge_info): Delete the edge info.
18732 (record_edge_info): Use new class & associated member functions.
18733 Tighten forms for testing for edge equivalences.
18734 (record_temporary_equivalences): Iterate over the simple
18735 equivalences rather than assuming there's only one per edge.
18736 (cprop_into_successor_phis): Iterate over the simple
18737 equivalences rather than assuming there's only one per edge.
18738 (optimize_stmt): Use operand_equal_p rather than pointer
18739 equality for mini-DSE code.
18741 2017-08-28 Nathan Sidwell <nathan@acm.org>
18743 * gcc.c (execute): Fold SIGPIPE handling into switch
18744 statement. Adjust internal error message.
18746 2017-08-28 Richard Biener <rguenther@suse.de>
18749 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18750 Do nothing for removed DIEs.
18752 2017-08-28 Richard Biener <rguenther@suse.de>
18754 PR tree-optimization/81977
18755 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18758 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18761 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18762 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18763 using targetm.gen_mem_thread_fence.
18765 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18768 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18769 predicate to register_operand. Reorder operands.
18770 (*btr<mode>): Ditto.
18771 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18772 (*btr<mode>_mask): Ditto.
18774 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18776 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18777 * config/rs6000/xmmintrin.h: New file.
18778 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18780 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18783 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18784 parameter and_insn and return it.
18785 (recombine_lvx_pattern): Insert a copy to ensure availability of
18786 the base register of the copied masking operation at the point of
18787 the instruction replacement.
18788 (recombine_stvx_pattern): Likewise.
18790 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18792 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18793 undocumented switches.
18794 (-mpower9-dform-vector): Likewise.
18795 (-mpower9-dform): Likewise.
18796 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18797 comments to delete references to -mpower9-dform* switches.
18798 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18799 Delete reference to -mpower9-dform* switches, test for
18800 -mpower9-vector instead.
18801 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18802 (OTHER_P9_VECTOR_MASKS): Likewise.
18803 (POWERPC_MASKS): Likewise.
18804 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18805 tests against -mpower9-dform* to -mpower9-vector. Delete code
18806 that checked for -mpower9-dform* consistancy with other options.
18807 Add test for -mpower9-misc to enable other power9 switches.
18808 (rs6000_init_hard_regno_mode_ok): Likewise.
18809 (rs6000_option_override_internal): Likewise.
18810 (rs6000_emit_prologue): Likewise.
18811 (rs6000_emit_epilogue): Likewise.
18812 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18813 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18814 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18816 (emit_fusion_p9_store): Likewise.
18817 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18818 resetting these macros if the assembler does not support ISA 3.0
18820 (TARGET_P9_DFORM_VECTOR): Likewise.
18821 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18822 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18824 2017-08-25 Alan Modra <amodra@gmail.com>
18826 PR rtl-optimization/81747
18827 * cse.c (cse_extended_basic_block): Don't attempt to record
18828 equivalences for degenerate conditional branches that branch
18829 to their fall-through.
18831 2017-08-24 Martin Sebor <msebor@redhat.com>
18833 PR middle-end/81908
18834 * gimple-fold.c (size_must_be_zero_p): New function.
18835 (gimple_fold_builtin_memory_op): Call it.
18837 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18839 * config/rs6000/mm_malloc.h: New file.
18841 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18843 PR tree-optimization/81913
18844 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18845 analysis when either IVs in condition can wrap.
18847 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18849 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18850 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18852 2017-08-24 Richard Biener <rguenther@suse.de>
18855 * targhooks.c (default_target_can_inline_p): Properly
18856 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18857 is present and always compare.
18858 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18859 infer -mfpmath=sse from TARGET_SSE_P.
18860 (ix86_can_inline_p): Properly use target_option_default_node when
18861 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18863 2017-08-24 Richard Biener <rguenther@suse.de>
18866 * dwarf2out.c (output_die): Handle flag_generate_offload like
18868 (output_comp_unit): Likewise.
18869 (gen_array_type_die): Likewise.
18870 (dwarf2out_early_finish): Likewise.
18871 (note_variable_value_in_expr): Likewise.
18872 (dwarf2out_finish): Likewise. Adjust assert.
18873 * cgraphunit.c (symbol_table::compile): Move setting of
18874 flag_generate_offload earlier ...
18875 (symbol_table::finalize_compilation_unit): ... here, before
18876 early debug finalization.
18878 2017-08-24 Richard Biener <rguenther@suse.de>
18880 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18881 and ipa-fnsummary.h.
18882 (ix86_can_inline_p): When ix86_fpmath flags do not match
18883 check whether the callee uses FP math at all.
18885 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18887 PR middle-end/81931
18888 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18889 instead of TYPE_PRECISION.
18891 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18892 Alan Hayward <alan.hayward@arm.com>
18893 David Sherwood <david.sherwood@arm.com>
18895 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18897 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18899 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18901 (ix86_frame::stack_realign_allocate): New field.
18902 (struct machine_frame_state): Modify comments.
18903 (machine_frame_state::sp_realigned_fp_end): New field.
18904 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18905 layout calculation.
18906 (sp_valid_at): Add assertion to assure no attempt to access invalid
18907 offset of a realigned stack.
18908 (fp_valid_at): Likewise.
18909 (choose_baseaddr): Modify comments.
18910 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18911 ix86_expand_prologue.
18912 (ix86_expand_prologue): Modify stack realignment and allocation.
18913 (ix86_expand_epilogue): Modify comments.
18914 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18915 avx2_runtime, avx512f, and avx512f_runtime.
18917 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18919 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18920 (mstackrealign): Do not init to -1.
18921 * config/i386/i386.c (ix86_option_override_internal):
18922 Check opts_set, not opts when setting default value of
18923 opts->x_ix86_force_align_arg_pointer.
18925 2017-08-23 Richard Biener <rguenther@suse.de>
18927 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18928 lang_hooks.decl_printable_name.
18929 * print-rtl-function.c (print_rtx_function): Likewise.
18930 * tree-pretty-print.c (dump_function_header): Likewise.
18932 2017-08-23 Richard Biener <rguenther@suse.de>
18935 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18936 -g0 at compile-time.
18938 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18940 PR middle-end/19706
18941 * doc/sourcebuild.texi (Other hardware attributes):
18944 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18946 PR middle-end/19706
18947 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18948 Fix single-use check.
18950 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18952 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18954 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18956 * doc/install.texi: Modify to add more details on running selected
18959 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18961 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18962 is combined with -mabi=ms.
18963 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18966 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18968 PR tree-optimization/81488
18969 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18970 and cached_basis fields.
18971 (MAX_SPREAD): New constant.
18972 (alloc_cand_and_find_basis): Initialize new fields.
18973 (clear_visited): New function.
18974 (create_phi_basis_1): Rename from create_phi_basis, set visited
18975 and cached_basis fields.
18976 (create_phi_basis): New wrapper function.
18977 (phi_add_costs_1): Rename from phi_add_costs, add spread
18978 parameter, set visited field, short-circuit when limits reached.
18979 (phi_add_costs): New wrapper function.
18980 (record_phi_increments_1): Rename from record_phi_increments, set
18982 (record_phi_increments): New wrapper function.
18983 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18984 (phi_incr_cost): New wrapper function.
18985 (all_phi_incrs_profitable_1): Rename from
18986 all_phi_incrs_profitable, set visited field.
18987 (all_phi_incrs_profitable): New wrapper function.
18989 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18990 Alan Hayward <alan.hayward@arm.com>
18991 David Sherwood <david.sherwood@arm.com>
18993 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18994 that takes the outer and inner modes.
18995 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18996 comparison as the canonical test for a paradoxical subreg.
18997 * combine.c (simplify_set): Use paradoxical_subreg_p.
18998 (make_extraction): Likewise.
18999 (force_to_mode): Likewise.
19000 (rtx_equal_for_field_assignment_p): Likewise.
19001 (gen_lowpart_for_combine): Likewise.
19002 (simplify_comparison): Likewise.
19003 * cse.c (equiv_constant): Likewise.
19004 * expmed.c (store_bit_field_1): Likewise.
19005 * final.c (alter_subreg): Likewise.
19006 * fwprop.c (propagate_rtx): Likewise.
19007 (forward_propagate_subreg): Likewise.
19008 * ira-conflicts.c (ira_build_conflicts): Likewise.
19009 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19010 * lra-constraints.c (curr_insn_transform): Likewise.
19011 (split_reg): Likewise.
19012 * lra-eliminations.c (move_plus_up): Likewise.
19013 (lra_eliminate_regs_1): Likewise.
19014 * recog.c (general_operand): Likewise.
19015 * ree.c (combine_reaching_defs): Likewise.
19016 * reload.c (push_reload): Likewise.
19017 (find_reloads): Likewise.
19018 * reload1.c (elimination_effects): Likewise.
19019 (compute_reload_subreg_offset): Likewise.
19020 (choose_reload_regs): Likewise.
19021 * rtlanal.c (subreg_lsb_1): Likewise.
19022 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19023 (simplify_subreg): Likewise.
19024 * var-tracking.c (track_loc_p): Likewise.
19025 * emit-rtl.c (byte_lowpart_offset): Likewise.
19026 (paradoxical_subreg_p): Delete out-of-line definition.
19028 2017-08-22 Jeff Law <law@redhat.com>
19030 PR tree-optimization/81741
19031 PR tree-optimization/71947
19032 * tree-ssa-dom.c: Include tree-inline.h.
19033 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19034 equivalences if one is more expensive to compute than the other.
19035 * tree-ssa-scopedtables.h (class const_or_copies): Make
19036 record_const_or_copy_raw method private.
19037 (class avail_exprs_stack): New method simplify_binary_operation.
19038 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19039 avail_exprs_stack::simplify_binary_operation as needed.
19040 (avail_exprs_stack::simplify_binary_operation): New function.
19042 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19044 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19045 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19046 (DOT_SYMBOLS): Likewise.
19047 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19048 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19049 (RS6000_ABI_NAME): Likewise.
19050 (TARGET_CMODEL): Likewise.
19051 (TOC_SECTION_ASM_OP): Likewise.
19052 (SET_CMODEL): New macro.
19053 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19055 2017-08-22 Richard Biener <rguenther@suse.de>
19057 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19058 to allow for free-lang-data replacements similar to verify_type_variant.
19060 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19061 Alan Hayward <alan.hayward@arm.com>
19062 David Sherwood <david.sherwood@arm.com>
19064 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19065 VOIDmode for the LABEL_REF.
19067 2017-08-22 Richard Biener <rguenther@suse.de>
19069 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19071 2017-08-22 Richard Biener <rguenther@suse.de>
19073 * common.opt (feliminate-dwarf2-dups): Ignore.
19074 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19075 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19076 same_die_p_wrap, compute_section_prefix,
19077 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19078 (comdat_symbol_id, comdat_symbol_number): Likewise.
19079 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19081 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19082 (output_die): Mark unreachable path unreachable.
19083 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19084 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19085 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19086 (dwarf2out_early_finish): Likewise.
19088 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19090 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19092 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19095 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19096 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19097 (avr_attribute_table) <io, io_low, address>: Initialize
19098 .decl_required with true.
19100 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19102 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19103 undocumented debugging options.
19104 (-mvsx-scalar-double): Likewise.
19105 (-mallow-df-permute): Likewise.
19106 (-mvectorize-builtins): Likewise.
19107 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19108 (rs6000_builtin_vectorized_function): Likewise.
19109 (rs6000_builtin_md_vectorized_function): Likewise.
19110 (rs6000_opt_vars): Likewise.
19112 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19115 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19116 (*btrq_imm): Rename from *btrq.
19117 (*btcq_imm): Rename from *btcq.
19118 (btsc): New code attribute.
19119 (*<btsc><mode>): New insn pattern.
19120 (*btr<mode>): Ditto.
19121 (*<btsc><mode>_mask): New insn_and_split pattern.
19122 (*btr<mode>_mask): Ditto.
19124 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19125 Alan Hayward <alan.hayward@arm.com>
19126 David Sherwood <david.sherwood@arm.com>
19128 * function.c (pad_below): Simplify padding calculation.
19130 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19131 Alan Hayward <alan.hayward@arm.com>
19132 David Sherwood <david.sherwood@arm.com>
19134 * target.def (function_prologue): Remove frame size argument.
19135 (function_epilogue): Likewise.
19136 * doc/tm.texi: Regenerate.
19137 * final.c (final_start_function): Update call to function_prologue.
19138 (final_end_function): Update call to function_epilogue.
19139 (default_function_pro_epilogue): Remove frame size argument.
19140 * output.h (default_function_pro_epilogue): Likewise.
19141 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19142 (arm_output_function_prologue): Likewise.
19143 * config/frv/frv.c (frv_function_prologue): Likewise.
19144 (frv_function_epilogue): Likewise.
19145 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19146 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19147 (ia64_output_function_epilogue): Likewise.
19148 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19149 (m32r_output_function_epilogue): Likewise.
19150 * config/microblaze/microblaze.c (microblaze_function_prologue)
19151 (microblaze_function_epilogue): Likewise.
19152 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19153 (mips_output_function_epilogue): Likewise.
19154 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19155 (mmix_target_asm_function_epilogue): Likewise.
19156 * config/msp430/msp430.c (msp430_start_function): Likewise.
19157 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19158 (nds32_asm_function_epilogue): Likewise.
19159 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19160 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19161 (pa_output_function_epilogue): Likewise.
19162 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19163 (rs6000_output_function_epilogue): Likewise.
19164 * config/rl78/rl78.c (rl78_start_function): Likewise.
19165 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19166 (rs6000_output_function_epilogue): Likewise.
19167 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19168 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19169 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19170 (sparc_asm_function_epilogue): Likewise.
19172 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19174 * tree.h (type_has_mode_precision_p): New function.
19175 * convert.c (convert_to_integer_1): Use it.
19176 * expr.c (expand_expr_real_2): Likewise.
19177 (expand_expr_real_1): Likewise.
19178 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19179 * match.pd: Likewise.
19180 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19181 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19182 * tree-tailcall.c (process_assignment): Likewise.
19183 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19184 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19185 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19186 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19187 (vectorizable_assignment): Likewise.
19188 (vectorizable_shift): Likewise.
19189 (vectorizable_operation): Likewise.
19190 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19192 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19194 * match.pd: Add pow (C, x) simplification.
19196 2017-08-21 Richard Biener <rguenther@suse.de>
19198 PR tree-optimization/81900
19199 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19200 for blocks with abnormal predecessors.
19201 (compute_antic): Do not set visited flag prematurely.
19203 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19206 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19208 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19210 * stor-layout.h (vector_type_mode): Move to...
19211 * tree.h (vector_type_mode): ...here.
19212 * stor-layout.c (vector_type_mode): Move to...
19213 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19215 2017-08-21 Richard Biener <rguenther@suse.de>
19217 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19218 register_external_die hooks.
19219 (debug_false_tree_charstarstar_uhwistar): Declare.
19220 (debug_nothing_tree_charstar_uhwi): Likewise.
19221 * debug.c (do_nothing_debug_hooks): Adjust.
19222 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19223 (debug_nothing_tree_charstar_uhwi): Likewise.
19224 * dbxout.c (dbx_debug_hooks): Adjust.
19225 (xcoff_debug_hooks): Likewise.
19226 * sdbout.c (sdb_debug_hooks): Likewise.
19227 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19228 * dwarf2out.c (macinfo_label_base): New global.
19229 (dwarf2out_register_external_die): New function for the
19230 register_external_die hook.
19231 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19232 (dwarf2_debug_hooks): Use them.
19233 (dwarf2_lineno_debug_hooks): Adjust.
19234 (struct die_struct): Add with_offset flag.
19235 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19236 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19237 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19238 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19239 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19240 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19241 defining section names for the early LTO debug variants.
19242 (reset_indirect_string): New helper.
19243 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19244 (print_dw_val): Add support for offsetted symbol references.
19245 (get_ultimate_context): Split out from is_cxx.
19246 (is_cxx): Use get_ultimate_context.
19247 (is_fortran): Add decl overload.
19248 (compute_comp_unit_symbol): Split out worker from
19249 compute_section_prefix.
19250 (compute_section_prefix): Call compute_comp_unit_symbol and
19251 set comdat_type_p here.
19252 (output_die): Skip DIE symbol output for the LTO added one.
19253 Handle DIE symbol references with offset.
19254 (output_comp_unit): Guard section name mangling properly.
19255 For LTO debug sections emit a symbol at the section beginning
19256 which we use to refer to its DIEs.
19257 (add_abstract_origin_attribute): For DIEs registered via
19258 dwarf2out_register_external_die directly refer to the early
19259 DIE rather than indirectly through the shadow one we created.
19260 Remove obsolete call to dwarf2out_abstract_function for
19261 non-function/block origins.
19262 (gen_array_type_die): When generating early LTO debug do
19263 not emit DW_AT_string_length.
19264 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19265 late when in LTO. As suggested place a gcc_unreachable for
19266 the DECL_ABSTRACT_P case.
19267 (gen_subprogram_die): Avoid another specification DIE
19268 for early built declarations/definitions for the late LTO case.
19269 (gen_variable_die): Add type references for late duplicated VLA dies
19271 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19272 we have the abstract instance already.
19273 (process_scope_var): Adjust decl DIE contexts in LTO which
19274 first puts them in limbo.
19275 (gen_decl_die): Do not generate type DIEs late apart from
19276 types for VLAs or for decls we do not yet have a DIE. Do not
19277 call dwarf2out_abstract_function late.
19278 (dwarf2out_early_global_decl): Make sure to create DIEs
19279 for abstract instances of a decl first.
19280 (dwarf2out_late_global_decl): Adjust comment.
19281 (output_macinfo_op): With multiple macro sections use
19282 macinfo_label_base to distinguish labels.
19283 (output_macinfo): Likewise. Update macinfo_label_base.
19284 Pass in the line info label.
19285 (note_variable_value_in_expr): When generating LTO resolve
19286 all variable values here by generating DIEs as needed.
19287 (init_sections_and_labels): Add early LTO debug flag parameter
19288 and generate different sections and names if set. Add generation
19289 counter for the labels so we can have multiple of them.
19290 (reset_dies): Helper to allow DIEs to be output multiple times.
19291 (dwarf2out_finish): When outputting DIEs to the fat part of an
19292 LTO object first reset DIEs.
19293 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19294 (modified_type_die): Check for decl_ultimate_origin being self
19296 (gen_type_die_with_usage): Likewise.
19297 (gen_typedef_die): Allow decl_ultimate_origin being self.
19298 (set_decl_abstract_flags): Remove.
19299 (set_block_abstract_flags): Likewise.
19300 (dwarf2out_abstract_function): Treat the early generated DIEs
19301 as the abstract copy and only add DW_AT_inline and
19302 DW_AT_artificial here and call set_decl_origin_self.
19303 If the DIE has an abstract origin don't do anything.
19304 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19305 if we have none yet (Go fails to build one, PR78628).
19306 (variably_modified_type_p): Prevent endless recursion for Ada
19307 cyclic pointer types.
19308 * lto-streamer-in.c: Include debug.h.
19309 (dref_queue): New global.
19310 (lto_read_tree_1): Stream in DIE references.
19311 (lto_input_tree): Register DIE references.
19312 (input_function): Stream DECL_DEBUG_ARGS.
19313 * lto-streamer-out.c: Include debug.h.
19314 (lto_write_tree_1): Output DIE references.
19315 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19316 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19317 (output_function): Stream DECL_DEBUG_ARGS.
19318 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19319 Stream DECL_ABSTRACT_ORIGIN.
19320 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19321 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19322 DECL_CONTEXT for file-scope decls.
19323 * lto-streamer.h (struct dref_entry): Declare.
19324 (dref_queue): Likewise.
19325 * cfgexpand.c (pass_expand::execute): Do not call the
19326 outlining_inline_function hook here.
19327 * lto-wrapper.c (debug_obj): New global.
19328 (tool_cleanup): Unlink it if required.
19329 (debug_objcopy): New function.
19330 (run_gcc): Handle early debug sections in the IL files by
19331 extracting them to separate files, partially linkin them and
19332 feeding the result back as result to the linker.
19333 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19334 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19335 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19336 sections into a separate segment.
19337 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19339 (darwin_asm_dwarf_section): Likewise.
19340 (darwin_asm_output_dwarf_offset): Likewise.
19341 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19343 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19344 Alan Hayward <alan.hayward@arm.com>
19345 David Sherwood <david.sherwood@arm.com>
19347 * read-md.h (md_reader::record_potential_iterator_use): Replace
19348 pointer argument with an rtx and an index.
19349 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19350 (apply_mode_iterator): Likewise.
19351 (apply_code_iterator): Likewise.
19352 (apply_int_iterator): Likewise.
19353 (apply_subst_iterator): Likewise.
19354 (record_iterator_use): Likewise.
19355 (record_attribute_use): Likewise.
19356 (md_reader::record_potential_iterator_use): Likewise. Update calls
19357 to record_iterator_use and apply_iterator.
19358 (iterator_use): Replace ptr with x and index.
19359 (attribute_use): Likewise.
19360 (apply_attribute_uses): Update calls to apply_iterator.
19361 (apply_iterators): Likewise. Update initialization of iterator_use.
19362 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19363 and record_potential_iterator_use.
19364 (rtx_reader::read_rtx_operand): Likewise.
19366 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19367 Alan Hayward <alan.hayward@arm.com>
19368 David Sherwood <david.sherwood@arm.com>
19370 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19373 2017-08-21 Richard Biener <rguenther@suse.de>
19375 PR middle-end/81884
19376 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19377 at struct end conservatively when comparing common bases.
19379 2017-08-21 Richard Biener <rguenther@suse.de>
19381 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19382 (mem_ref_in_stmt): Remove.
19383 (determine_max_movement): Use ref index to get at the reference.
19384 (invariantness_dom_walker::before_dom_children): Deal with
19385 lim data already initialized.
19386 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19388 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19390 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19391 (TARGET_ISA_ROUND): Ditto.
19392 (TARGET_ROUND): Ditto.
19393 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19394 * config/i386/i386.md: Ditto.
19395 * config/i386/sse.md: Ditto.
19396 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19397 with OPTION_MASK_ISA_SSE4_1.
19399 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19402 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19403 __builtin_ia32_lzcnt_u16.
19405 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19408 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19409 (rs6000_set_current_function): Rewrite function to use it.
19411 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19414 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19415 and TYPE_WARN_IF_NOT_ALIGN.
19416 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19417 (handle_warn_if_not_align): New.
19418 (place_union_field): Call handle_warn_if_not_align.
19419 (place_field): Call handle_warn_if_not_align.
19420 Copy TYPE_WARN_IF_NOT_ALIGN.
19421 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19422 (layout_type): Likewise.
19423 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19425 (tree_decl_common): Add warn_if_not_align.
19426 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19427 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19428 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19429 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19430 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19431 * doc/extend.texi: Document warn_if_not_aligned attribute.
19432 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19434 2017-08-17 Martin Liska <mliska@suse.cz>
19437 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19438 (get_data_dependence): Use it as pointer type.
19439 (distribute_loop): Likewise.
19441 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19443 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19444 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19445 (p8_vmrgow_<mode>_direct): New define_insn.
19446 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19447 handle endianness for vmrgew and vmrgow permute patterns.
19449 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19451 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19452 * config/rs6000/rs6000-cpus.def: Remove comment.
19453 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19454 (POWERPC_MASKS): Likewise.
19455 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19456 use of TARGET_VSX_TIMODE.
19457 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19458 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19459 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19460 (rs6000_option_override_internal): Remove dead code.
19461 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19462 (rs6000_legitimize_reload_address): Likewise.
19463 (rs6000_legitimate_address_p): Likewise.
19464 (rs6000_opt_masks): Delete "vsx-timode".
19465 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19466 from function comment.
19467 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19468 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19469 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19471 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19472 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19473 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19474 (VSX_TI): Likewise.
19476 (define_peephole2): Likewise.
19478 2017-08-17 Martin Sebor <msebor@redhat.com>
19481 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19482 past the end of an array.
19483 (test_pp_format): Add test cases.
19485 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19487 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19488 missing ECF_NOTHROW flags.
19490 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19493 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19494 operands residing in integer registers.
19495 (*vsx_le_perm_load_<mode>): Likewise.
19496 (*vsx_le_perm_store_<mode>): Likewise.
19497 (define_peephole2): Add peepholes to optimize the above.
19499 2017-08-17 Marek Polacek <polacek@redhat.com>
19501 PR middle-end/81814
19502 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19503 to mimic what shorten_compare did. Change the return type to bool.
19504 (fold_cond_expr_with_comparison): Update call to
19505 operand_equal_for_comparison_p.
19506 (fold_ternary_loc): Likewise.
19508 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19510 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19512 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19514 2017-08-17 Richard Biener <rguenther@suse.de>
19516 * tree-ssa-structalias.c (solve_graph): When propagating
19517 to successors update the graphs succ edges and avoid duplicate work.
19519 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19522 * config/i386/i386.c (ix86_option_override_internal): Save target
19523 specific options after ix86_stack_protector_guard_reg was changed.
19525 2017-08-17 Richard Biener <rguenther@suse.de>
19527 PR tree-optimization/81827
19528 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19529 (new_var_info): Initialize it conservatively.
19530 (get_call_vi): Mark register vars.
19531 (new_scalar_tmp_constraint_exp): Likewise.
19532 (handle_rhs_call): Likewise.
19533 (handle_const_call): Likewise.
19534 (create_function_info_for): Likewise.
19535 (solve_constraints): Sort varinfos to separate register from
19536 non-register vars to pack points-to solution bitmaps during
19539 2017-08-17 Marek Polacek <polacek@redhat.com>
19541 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19543 2017-08-17 Richard Biener <rguenther@suse.de>
19545 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19546 to true when overflow is undefined and we saturated the result.
19548 2017-08-17 Alan Modra <amodra@gmail.com>
19551 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19552 Don't use store multiple if only one reg needs saving.
19553 (interesting_frame_related_regno): New function.
19554 (rs6000_frame_related): Don't emit frame info for regs that
19556 (rs6000_emit_epilogue): Likewise.
19558 2017-08-16 Nathan Sidwell <nathan@acm.org>
19560 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19561 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19562 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19563 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19564 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19565 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19566 (verify_type): Adjust for TYPE_BINFO move.
19567 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19568 process TYPE_BINFO directly.
19569 (hash_tree): Likewise.
19570 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19572 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19575 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19577 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19579 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19582 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19583 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19584 Add ix86_binary_operator_ok to insn constraint.
19585 (*iordi_1_bts): Ditto.
19586 (*xordi_1_btc): Ditto.
19587 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19588 Update corresponding peephole2 pattern.
19592 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19594 PR tree-optimization/81832
19595 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19596 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19598 2017-08-16 Marek Polacek <polacek@redhat.com>
19601 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19602 regarding size_zero_node.
19604 2017-08-16 Martin Liska <mliska@suse.cz>
19607 * config.gcc: Respect previously set extra_objs in case
19610 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19612 PR tree-optimization/81835
19613 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19614 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19615 not depend on the phi.
19617 2017-08-16 Alan Modra <amodra@gmail.com>
19619 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19622 2017-08-16 Alan Modra <amodra@gmail.com>
19624 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19625 (save_reg_p): ..into this. Update all callers.
19626 (first_reg_to_save): Simplify.
19628 2017-08-16 Alan Modra <amodra@gmail.com>
19630 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19633 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19637 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19638 constant count if that count is less than 32.
19640 2017-08-15 Nathan Sidwell <nathan@acm.org>
19642 * gcc.c (execute): Emit friendlier message if inferior is killed
19643 by an external cause.
19645 2017-08-15 Richard Biener <rguenther@suse.de>
19647 PR tree-optimization/81790
19648 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19649 CONSTRUCTORs from simplifying and VN.
19651 2017-08-14 Martin Sebor <msebor@redhat.com>
19653 * builtin-attrs.def: Add comments.
19655 2017-08-14 Martin Sebor <msebor@redhat.com>
19658 * doc/extend.texi (attribute nonstring): Document new attribute.
19660 2017-08-14 Martin Sebor <msebor@redhat.com>
19663 * tree-diagnostic.c (default_tree_printer): Handle %G.
19664 * gimple-pretty-print.h (percent_G_format): Declare new function.
19665 * gimple-pretty-print.c (percent_G_format): Define.
19666 * tree-pretty-print.c (percent_K_format): Add argument.
19668 2017-08-14 Martin Sebor <msebor@redhat.com>
19670 PR translation/79998
19671 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19672 Remove a stray space.
19674 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19677 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19678 (*iordi_1_bts): Ditto.
19679 (*xordi_1_btc): Ditto.
19681 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19684 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19685 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19687 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19688 quoted strings, and make more translator-friendly.
19689 (darwin_rs6000_override_options): Likewise.
19690 (rs6000_option_override_internal): Likewise.
19691 (rs6000_return_in_memory): Fix overlong line.
19692 (init_cmulative_args): Use quoted strings, and make more
19693 translator-friendly.
19694 (rs6000_pass_by_reference): Fix overlong line.
19695 (def_builtin): Use quoted strings.
19696 (altivec_expand_predicate_builtin): Use quoted strings, and make
19697 more translator-friendly.
19698 (htm_expand_builtin): Use quoted strings.
19699 (cpu_expand_builtin): Use quoted strings, and make more
19700 translator-friendly.
19701 (altivec_expand_builtin): Likewise.
19702 (paired_expand_predicate_builtin): Likewise.
19703 (rs6000_invalid_builtin): Likewise.
19704 (builtin_function_type): Use quoted strings.
19705 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19706 more translator-friendly.
19707 (rs6000_trampoline_init): Likewise.
19708 (rs6000_handle_altivec_attribute): Likewise.
19709 (rs6000_inner_target_options): Use quoted strings.
19710 (rs6000_disable_incompatible_switches): Likewise.
19711 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19712 strings, and make more translator-friendly.
19713 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19715 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19717 PR tree-optimization/81799
19718 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19719 cond_expr to simple gimple operand.
19721 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19723 PR middle-end/46932
19724 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19726 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19730 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19731 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19732 TARGET_GASISR_PROLOGUES.
19733 * config/avr/avr.c (avr_option_override): Same.
19734 (avr_pass_pre_proep::execute): Same.
19736 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19739 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19740 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19742 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19744 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19745 define_insn_and_split. Split to a memory load from 0 in
19746 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19747 using PTR mode iterator.
19748 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19749 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19750 (*add_tp_<mode>): Redefine as define_insn_and_split.
19751 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19752 address space. Merge with *add_tp_x32 using PTR mode iterator.
19753 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19754 Split to an add with a memory load from 0 in
19755 DEFAULT_TLS_SEG_REG address space.
19757 2017-08-12 Andrew Pinski <apinski@cavium.com>
19759 * config/aarch64/aarch64-option-extensions.def (rdma):
19760 Fix feature string to what Linux prints out in /proc/cpuinfo.
19762 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19765 * dwarf2out.c (modified_type_die): For C typedef types that have
19766 an ultimate origin, process the ultimate origin instead of the
19768 (gen_typedef_die): Assert that input DECLs have no ultimate
19770 (gen_type_die_with_usage): For typedef variants that have an
19771 ultimate origin, just call gen_decl_die on the original DECL.
19772 (process_scope_var): Avoid creating DIEs for local typedefs and
19773 concrete static variables.
19775 2017-08-12 Alan Modra <amodra@gmail.com>
19779 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19780 match gnu-user.h startfile.
19781 (ENDFILE_LINUX_SPEC): Similarly.
19783 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19786 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19788 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19790 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19791 * config/aarch64/aarch64.md (mov<mode>): Change.
19792 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19793 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19794 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19796 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19798 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19799 for storage order barriers.
19801 2017-08-11 Martin Liska <mliska@suse.cz>
19804 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19805 variables of void type.
19807 2017-08-11 Martin Liska <mliska@suse.cz>
19809 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19810 TARGET_SUPPORTS_ALIASES.
19811 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19812 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19813 (optimize_weakref): Likewise.
19814 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19815 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19816 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19818 2017-08-11 Martin Liska <mliska@suse.cz>
19821 * config/i386/i386.c (make_resolver_func): Do complete
19822 refactoring of the function.
19824 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19827 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19828 * config/i386/i386.c (ix86_stack_protect_guard): Use
19829 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19830 * doc/invoke.texi (x86 Options): Document
19831 -mstack-protector-guard-symbol= option.
19833 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19835 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19836 * config/i386/i386.c (ix86_split_stack_guard): New function.
19837 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19838 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19839 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19840 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19841 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19842 (split_stack_space_check): Call ix86_split_stack_guard.
19844 2017-08-10 Martin Sebor <msebor@redhat.com>
19846 * print-tree.c (print_node): Print location using the established
19848 Replace spaces with colons.
19849 (debug_raw, debug): Ditto.
19851 2017-08-10 Martin Sebor <msebor@redhat.com>
19854 * pretty-print.c (pp_format): Correct the handling of %s precision.
19856 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19859 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19861 (ix86_finalize_stack_frame_flags): This. Also clear
19862 frame_pointer_needed if -fno-omit-frame-pointer is used without
19864 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19865 with ix86_finalize_stack_frame_flags.
19866 (ix86_expand_epilogue): Likewise.
19867 (ix86_expand_split_stack_prologue): Likewise.
19868 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19870 2017-08-10 Martin Liska <mliska@suse.cz>
19873 * c-attribs.c (handle_target_attribute):
19874 Report warning for an empty string argument of target attribute.
19876 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19879 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19881 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19882 or DECL_NONLOCAL labels.
19883 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19884 or DECL_NONLOCAL labels here.
19886 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19888 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19889 to indicate when early gimple folding has been disabled.
19890 (rs6000_gimple_fold_builtin): Add debug content.
19891 (rs6000_invalid_builtin): Fix whitespace.
19892 (rs6000_expand_builtin): Fix whitespace.
19893 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19895 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19898 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19899 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19901 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19903 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19904 (qdf24xx): Likewise.
19905 * config/aarch64/aarch64.md: Include falkor.md.
19906 * config/aarch64/falkor.md: New.
19908 2017-08-09 Marek Polacek <polacek@redhat.com>
19911 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19912 * diagnostic.c (emit_diagnostic): Add a comment.
19913 (emit_diagnostic_valist): New function.
19915 2017-08-09 Marek Polacek <polacek@redhat.com>
19918 * input.c (make_location): New overload.
19919 * input.h (make_location): Declare.
19921 2017-08-08 Alan Modra <amodra@gmail.com>
19922 H.J. Lu <hongjiu.lu@intel.com>
19925 * gcc.c (NO_PIE_SPEC): Delete.
19926 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19928 (LINK_PIE_SPEC): ..to here.
19929 (LINK_COMMAND_SPEC): Support -no-pie.
19930 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19931 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19933 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19934 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19935 (ENDFILE_CRTEND_SPEC): Similarly.
19937 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19940 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19941 (mstack-protector-guard-offset=): Ditto.
19942 * config/i386/i386.c (ix86_option_override): Handle
19943 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19945 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19946 ix86_stack_protect_guard_offset variables.
19947 (TARGET_STACK_PROTECT_GUARD): Always define.
19948 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19949 and -mstack-protector-guard-offset= options.
19951 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19953 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19954 boundary case for the last candidate.
19956 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19958 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19959 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19961 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19963 PR middle-end/19706
19964 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19965 * config/aarch64/aarch64-builtins.c
19966 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19967 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19968 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19970 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19971 Andrew Pinski <pinskia@gmail.com>
19973 PR middle-end/19706
19974 * internal-fn.def (XORSIGN): New.
19975 * optabs.def (xorsign_optab): New.
19976 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19977 (convert_expand_mult_copysign): New.
19978 (pass_optimize_widening_mul::execute): Call
19979 convert_expand_mult_copysign.
19981 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19983 PR tree-optimization/81354
19984 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19985 Insert on edges rather than explicitly creating landing pads.
19986 (analyze_candidates_and_replace): Commit edge inserts.
19988 2017-08-08 Richard Biener <rguenther@suse.de>
19990 PR middle-end/81719
19991 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19992 (expand_simple_operations): Also look through ADDR_EXPRs with
19993 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19995 2017-08-08 Richard Biener <rguenther@suse.de>
19997 PR tree-optimization/81723
19998 * tree-vect-slp.c (struct bst_traits): New hash traits.
19999 (bst_fail): New global.
20000 (vect_build_slp_tree_2): New worker, split out from ...
20001 (vect_build_slp_tree): ... this now wrapping it with using
20002 bst_fail set to cache SLP tree build fails. Properly handle
20004 (vect_analyze_slp_instance): Allocate and free bst_fail.
20006 2017-08-08 Martin Liska <mliska@suse.cz>
20009 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20010 LABEL_DECLs that can be from a different function.
20012 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20014 PR tree-optimization/81744
20015 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20016 loop's number of iterations.
20018 2017-08-08 Martin Liska <mliska@suse.cz>
20020 * asan.c: Include header files.
20021 * attribs.c (build_decl_attribute_variant): New function moved
20023 (build_type_attribute_qual_variant): Likewise.
20024 (cmp_attrib_identifiers): Likewise.
20025 (simple_cst_list_equal): Likewise.
20026 (omp_declare_simd_clauses_equal): Likewise.
20027 (attribute_value_equal): Likewise.
20028 (comp_type_attributes): Likewise.
20029 (build_type_attribute_variant): Likewise.
20030 (lookup_ident_attribute): Likewise.
20031 (remove_attribute): Likewise.
20032 (merge_attributes): Likewise.
20033 (merge_type_attributes): Likewise.
20034 (merge_decl_attributes): Likewise.
20035 (merge_dllimport_decl_attributes): Likewise.
20036 (handle_dll_attribute): Likewise.
20037 (attribute_list_equal): Likewise.
20038 (attribute_list_contained): Likewise.
20039 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20040 (lookup_attribute_by_prefix): Likewise.
20041 * bb-reorder.c: Include header files.
20042 * builtins.c: Likewise.
20043 * calls.c: Likewise.
20044 * cfgexpand.c: Likewise.
20045 * cgraph.c: Likewise.
20046 * cgraphunit.c: Likewise.
20047 * convert.c: Likewise.
20048 * dwarf2out.c: Likewise.
20049 * final.c: Likewise.
20050 * fold-const.c: Likewise.
20051 * function.c: Likewise.
20052 * gimple-expr.c: Likewise.
20053 * gimple-fold.c: Likewise.
20054 * gimple-pretty-print.c: Likewise.
20055 * gimple.c: Likewise.
20056 * gimplify.c: Likewise.
20057 * hsa-common.c: Likewise.
20058 * hsa-gen.c: Likewise.
20059 * internal-fn.c: Likewise.
20060 * ipa-chkp.c: Likewise.
20061 * ipa-cp.c: Likewise.
20062 * ipa-devirt.c: Likewise.
20063 * ipa-fnsummary.c: Likewise.
20064 * ipa-inline.c: Likewise.
20065 * ipa-visibility.c: Likewise.
20067 * lto-cgraph.c: Likewise.
20068 * omp-expand.c: Likewise.
20069 * omp-general.c: Likewise.
20070 * omp-low.c: Likewise.
20071 * omp-offload.c: Likewise.
20072 * omp-simd-clone.c: Likewise.
20073 * opts-global.c: Likewise.
20074 * passes.c: Likewise.
20075 * predict.c: Likewise.
20076 * sancov.c: Likewise.
20077 * sanopt.c: Likewise.
20078 * symtab.c: Likewise.
20079 * toplev.c: Likewise.
20080 * trans-mem.c: Likewise.
20081 * tree-chkp.c: Likewise.
20082 * tree-eh.c: Likewise.
20083 * tree-into-ssa.c: Likewise.
20084 * tree-object-size.c: Likewise.
20085 * tree-parloops.c: Likewise.
20086 * tree-profile.c: Likewise.
20087 * tree-ssa-ccp.c: Likewise.
20088 * tree-ssa-live.c: Likewise.
20089 * tree-ssa-loop.c: Likewise.
20090 * tree-ssa-sccvn.c: Likewise.
20091 * tree-ssa-structalias.c: Likewise.
20092 * tree-ssa.c: Likewise.
20093 * tree-streamer-in.c: Likewise.
20094 * tree-vectorizer.c: Likewise.
20095 * tree-vrp.c: Likewise.
20096 * tsan.c: Likewise.
20097 * ubsan.c: Likewise.
20098 * varasm.c: Likewise.
20099 * varpool.c: Likewise.
20100 * tree.c: Remove functions moved to attribs.[ch].
20101 * tree.h: Likewise.
20102 * config/aarch64/aarch64.c: Add attrs.h header file.
20103 * config/alpha/alpha.c: Likewise.
20104 * config/arc/arc.c: Likewise.
20105 * config/arm/arm.c: Likewise.
20106 * config/avr/avr.c: Likewise.
20107 * config/bfin/bfin.c: Likewise.
20108 * config/c6x/c6x.c: Likewise.
20109 * config/cr16/cr16.c: Likewise.
20110 * config/cris/cris.c: Likewise.
20111 * config/darwin.c: Likewise.
20112 * config/epiphany/epiphany.c: Likewise.
20113 * config/fr30/fr30.c: Likewise.
20114 * config/frv/frv.c: Likewise.
20115 * config/ft32/ft32.c: Likewise.
20116 * config/h8300/h8300.c: Likewise.
20117 * config/i386/winnt.c: Likewise.
20118 * config/ia64/ia64.c: Likewise.
20119 * config/iq2000/iq2000.c: Likewise.
20120 * config/lm32/lm32.c: Likewise.
20121 * config/m32c/m32c.c: Likewise.
20122 * config/m32r/m32r.c: Likewise.
20123 * config/m68k/m68k.c: Likewise.
20124 * config/mcore/mcore.c: Likewise.
20125 * config/microblaze/microblaze.c: Likewise.
20126 * config/mips/mips.c: Likewise.
20127 * config/mmix/mmix.c: Likewise.
20128 * config/mn10300/mn10300.c: Likewise.
20129 * config/moxie/moxie.c: Likewise.
20130 * config/msp430/msp430.c: Likewise.
20131 * config/nds32/nds32-isr.c: Likewise.
20132 * config/nds32/nds32.c: Likewise.
20133 * config/nios2/nios2.c: Likewise.
20134 * config/nvptx/nvptx.c: Likewise.
20135 * config/pa/pa.c: Likewise.
20136 * config/pdp11/pdp11.c: Likewise.
20137 * config/powerpcspe/powerpcspe.c: Likewise.
20138 * config/riscv/riscv.c: Likewise.
20139 * config/rl78/rl78.c: Likewise.
20140 * config/rx/rx.c: Likewise.
20141 * config/s390/s390.c: Likewise.
20142 * config/sh/sh.c: Likewise.
20143 * config/sol2.c: Likewise.
20144 * config/sparc/sparc.c: Likewise.
20145 * config/spu/spu.c: Likewise.
20146 * config/stormy16/stormy16.c: Likewise.
20147 * config/tilegx/tilegx.c: Likewise.
20148 * config/tilepro/tilepro.c: Likewise.
20149 * config/v850/v850.c: Likewise.
20150 * config/vax/vax.c: Likewise.
20151 * config/visium/visium.c: Likewise.
20152 * config/xtensa/xtensa.c: Likewise.
20154 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20157 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20158 constraints since the -mupper-regs-* switches have been
20160 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20161 into a vector from a double word element that was extracted from
20162 another vector, and eliminate extra XXPERMDI instructions.
20163 (vsx_concat_<mode>_2): Likewise.
20164 (vsx_concat_<mode>_3): Likewise.
20165 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20166 concat to allow optimizing inserts from previous extracts.
20168 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20170 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20171 memory reference to a SSP offset in TLS address space.
20172 (ix86_print_operand) <case '@'>: Remove.
20173 (ix86_print_operand_punct_valid_p): Remove '@' code.
20174 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20175 UNSPEC_SP_TLS_TEST.
20176 (stack_tls_protect_set_<mode>): Remove.
20177 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20178 (stack_tls_protect_test_<mode>): Remove.
20179 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20181 2017-08-07 Olivier Hainque <hainque@adacore.com>
20184 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20186 2017-08-07 Douglas Rupp <rupp@adacore.com>
20188 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20189 variable was referenced as multidir in command.
20191 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20194 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20197 2017-08-07 Martin Liska <mliska@suse.cz>
20199 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20200 * config/rl78/rl78.c: Add include of attribs.h.
20201 * config/sh/sh.c: Likewise.
20202 * config/v850/v850.c: Likewise.
20204 2017-08-07 Tom de Vries <tom@codesourcery.com>
20206 PR middle-end/78266
20207 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20209 2017-08-07 Martin Liska <mliska@suse.cz>
20211 * config/mips/mips.c: Include attribs.h.
20213 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20216 * doc/invoke.texi: Document change in behvaior for -Ofast for
20219 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20221 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20223 (aarch64_pop_regs): Likewise.
20224 (aarch64_gen_load_pair): Likewise.
20225 (aarch64_save_callee_saves): Likewise.
20226 (aarch64_restore_callee_saves): Likewise.
20228 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20230 * config/i386/i386.c: Revert the last change.
20232 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20235 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20237 (ix86_finalize_stack_frame_flags): This. Also clear
20238 frame_pointer_needed if -fno-omit-frame-pointer is used without
20240 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20241 with ix86_finalize_stack_frame_flags.
20242 (ix86_expand_epilogue): Likewise.
20243 (ix86_expand_split_stack_prologue): Likewise.
20245 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20248 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20249 to P_AES for Westmere.
20251 2017-08-07 Jonathan Yong <10walls@gmail.com>
20253 * config/i386/mingw.opt (fset-stack-executable): Removed.
20254 * config/i386/cygming.opt (fset-stack-executable): Moved
20256 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20258 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20260 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20262 2017-08-07 Marek Polacek <polacek@redhat.com>
20264 PR middle-end/81737
20265 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20267 2017-08-07 Martin Liska <mliska@suse.cz>
20269 * attribs.h (canonicalize_attr_name): New function.
20270 (cmp_attribs): Move from c-format.c and adjusted.
20271 (is_attribute_p): Moved from tree.h.
20272 * tree-inline.c: Add new includes.
20273 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20274 (private_is_attribute_p): Remove.
20275 (private_lookup_attribute): Likewise.
20276 (private_lookup_attribute_by_prefix): Simplify.
20277 (remove_attribute): Use is_attribute_p.
20278 * tree.h: Remove removed declarations.
20280 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20282 PR middle-end/81698
20283 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20284 instead of computing it in the function. Formatting fix.
20285 (expand_case): Don't rely on default_edge being the first edge,
20286 clear it if removing it, pass default_edge to
20287 emit_case_dispatch_table.
20288 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20291 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20293 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20294 insn in the function, emit NOP after the insn.
20296 2017-08-06 Tom de Vries <tom@codesourcery.com>
20298 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20301 2017-08-06 Tom de Vries <tom@codesourcery.com>
20303 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20306 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20308 PR tree-optimization/57371
20309 * match.pd: New pattern.
20311 2017-08-04 Marek Polacek <polacek@redhat.com>
20313 PR middle-end/81695
20314 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20315 perform the computation in offset_int.
20317 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20319 PR tree-optimization/81136
20320 * tree-vectorizer.h: Include tree-hash-traits.h.
20321 (vec_base_alignments): New typedef.
20322 (vec_info): Add a base_alignments field.
20323 (vect_record_base_alignments): Declare.
20324 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20326 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20327 (create_data_ref): Add an is_conditional_in_stmt argument.
20328 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20329 the is_conditional_in_stmt field.
20330 (data_ref_loc): Add an is_conditional_in_stmt field.
20331 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20332 (find_data_references_in_stmt): Update call to create_data_ref.
20333 (graphite_find_data_references_in_stmt): Likewise.
20334 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20335 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20336 (vect_record_base_alignment): New function.
20337 (vect_record_base_alignments): Likewise.
20338 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20339 for nested statements even if we fail to compute a misalignment.
20340 Use pooled base alignments for unconditional references.
20341 (vect_find_same_alignment_drs): Compare base addresses instead
20343 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20344 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20346 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20348 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20349 Add an explicit name for the enum. Use auto_vec for slp_instances
20350 and grouped_stores.
20351 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20353 (_bb_vec_info): Add a constructor and destructor.
20354 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20355 (destroy_loop_vec_info): Delete.
20356 (vect_destroy_datarefs): Likewise.
20357 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20358 (vec_info::vec_info): New function.
20359 (vec_info::~vec_info): Likewise.
20360 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20361 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20362 destroy_loop_vec_info.
20363 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20364 (_loop_vec_info::_loop_vec_info): ...this.
20365 (destroy_loop_vec_info): Replace with...
20366 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20367 the stmt_vec_infos. Leave handling of vec_info information to its
20368 destructor. Remove explicit vector releases.
20369 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20370 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20371 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20372 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20373 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20374 (destroy_bb_vec_info): Replace with...
20375 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20376 information to its destructor.
20377 (vect_slp_analyze_bb_1): Use new and delete instead of
20378 new_bb_vec_info and destroy_bb_vec_info.
20379 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20382 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20384 * tree-data-ref.h (subscript): Add access_fn field.
20385 (data_dependence_relation): Add could_be_independent_p.
20386 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20387 (same_access_functions): Move to tree-data-ref.c.
20388 * tree-data-ref.c (ref_contains_union_access_p): New function.
20389 (access_fn_component_p): Likewise.
20390 (access_fn_components_comparable_p): Likewise.
20391 (dr_analyze_indices): Add a reference to access_fn_component_p.
20392 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20394 (constant_access_functions): Likewise.
20395 (add_other_self_distances): Likewise.
20396 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20397 (initialize_data_dependence_relation): Use XCNEW and remove
20398 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20399 of access functions that have the same type. Allow the
20400 subsequence to end with different bases in some circumstances.
20401 Record the chosen access functions in SUB_ACCESS_FN.
20402 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20403 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20404 (subscript_dependence_tester_1): Likewise dra and drb.
20405 (build_classic_dist_vector): Update calls accordingly.
20406 (subscript_dependence_tester): Likewise.
20407 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20408 DDR_COULD_BE_INDEPENDENT_P.
20409 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20410 comp_alias_ddrs instead of may_alias_ddrs.
20411 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20413 (vect_analyze_data_ref_dependence): Use it if
20414 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20415 distance vectors if that fails.
20416 (dependence_distance_ge_vf): New function.
20417 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20418 LOOP_VINFO_MAY_ALIAS_DDRS.
20420 2017-08-04 Richard Biener <rguenther@suse.de>
20422 PR middle-end/81705
20423 * fold-const.c (fold_binary_loc): Properly restrict
20424 minus_var0 && minus_var1 case when associating undefined overflow
20427 2017-08-04 Tom de Vries <tom@codesourcery.com>
20429 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20431 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20433 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20434 Don't start diagnostic messages with a capital letter.
20435 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20437 (rs6000_invalid_builtin): Likewise.
20438 (rs6000_trampoline_init): Likewise.
20440 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20443 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20444 after setting changeable df flags.
20446 2017-08-03 Richard Biener <rguenther@suse.de>
20448 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20449 up if the use is in USE - X.
20451 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20453 * toplev.c (dumpfile.h): New include.
20454 (internal_error_reentered): New static function. Use it...
20455 (internal_error_function): ...here to handle reentered internal_error.
20457 2017-08-03 Richard Biener <rguenther@suse.de>
20459 PR middle-end/81148
20460 * fold-const.c (split_tree): Add minus_var and minus_con
20461 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20462 here but always use minus_*.
20463 (associate_trees): Assert we never associate with MINUS_EXPR
20464 and NULL first operand. Do not recurse for PLUS_EXPR operands
20465 when associating as MINUS_EXPR either.
20466 (fold_binary_loc): Track minus_var and minus_con.
20468 2017-08-03 Tom de Vries <tom@codesourcery.com>
20471 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20472 ACCEL_COMPILER, apply finish_options on
20473 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20475 2017-08-03 Tom de Vries <tom@codesourcery.com>
20478 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20479 function_entry_patch_area_size > 0.
20481 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20484 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20485 instead of 10??LU, perform unit multiplication in wide_int,
20486 don't change alloc_object_size_limit if the limit is larger
20489 PR tree-optimization/81655
20490 PR tree-optimization/81588
20491 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20492 the case when ranges[i].low and high are 1 for unsigned type with
20495 PR middle-end/81052
20496 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20497 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20499 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20501 * tree-vrp.h: Add include guard.
20503 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20506 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20507 (ud2): New insn pattern.
20508 * config/i386/i386.c (ix86_expand_epilogue):
20509 For naked functions, generate ud2 instead of trap insn.
20511 2017-08-02 Marek Polacek <polacek@redhat.com>
20514 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20516 2017-08-02 Tom de Vries <tom@codesourcery.com>
20517 Cesar Philippidis <cesar@codesourcery.com>
20519 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20520 Add missing edge probabilities.
20522 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20524 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20525 Correct endianness.
20527 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20529 PR middle-end/79499
20530 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20531 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20532 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20533 prologue_seq sequences - if any.
20535 2017-08-02 Richard Biener <rguenther@suse.de>
20537 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20538 via vectors if supported, integer extracts via punning if supported
20539 or otherwise vector extracts.
20541 2017-08-02 Richard Biener <rguenther@suse.de>
20543 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20545 (bitmap_insert_into_set): ... this.
20547 2017-08-02 Richard Biener <rguenther@suse.de>
20549 PR tree-optimization/81633
20551 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20553 PR tree-optimization/71752
20554 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20556 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20558 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20559 (machine_function::call_ms2sysv_pad_out): Remove field.
20560 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20561 (ix86_compute_frame_layout): Likewise.
20563 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20566 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20567 attribute with interrupt attribute.
20569 2017-08-01 Andrew Pinski <apinski@cavium.com>
20571 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20572 BIT_INSERT_EXPR's operand 1
20573 to see if the types precision matches.
20575 2017-08-01 Martin Liska <mliska@suse.cz>
20577 PR middle-end/70140
20578 * builtins.c (expand_builtin_memcpy_args): Remove.
20579 (expand_builtin_memcpy): Call newly added function
20580 expand_builtin_memory_copy_args.
20581 (expand_builtin_memcpy_with_bounds): Likewise.
20582 (expand_builtin_mempcpy): Remove last argument.
20583 (expand_builtin_mempcpy_with_bounds): Likewise.
20584 (expand_builtin_memory_copy_args): New function created from
20585 expand_builtin_mempcpy_args with small modifications.
20586 (expand_builtin_mempcpy_args): Remove.
20587 (expand_builtin_stpcpy): Remove unused argument.
20588 (expand_builtin): Likewise.
20589 (expand_builtin_with_bounds): Likewise.
20591 2017-08-01 Martin Liska <mliska@suse.cz>
20594 Make mempcpy more optimal (PR middle-end/70140).
20596 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20599 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20600 __builtin_vec_cmpne verify both arguments are compatible vectors
20601 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20602 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20603 move computation of aligned to after checking the argument types.
20607 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20608 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20611 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20612 Doug Rupp <rupp@adacore.com>
20613 Olivier Hainque <hainque@adacore.com>
20615 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20616 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20618 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20619 for TARGET_OS_CPP_BUILTIN.
20620 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20621 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20623 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20624 passing required abi options to the assembler for EABI configurations.
20625 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20626 of .text.hot and .text.unlikely sections for kernel modules when
20627 using ARM style exceptions.
20628 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20629 options. Add EXTRA_CC1_SPEC.
20630 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20632 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20634 (ARM_TARGET2_DWARF_FORMAT): Define.
20635 * config/arm/t-vxworks: Adjust multilib control to removal of the
20636 Diab command line options.
20638 2017-08-01 Martin Liska <mliska@suse.cz>
20640 PR gcov-profile/81561
20641 * gcov.c (unblock): Make unblocking safe as we need to preserve
20642 index correspondence of blocks and block_lists.
20644 2017-08-01 Richard Biener <rguenther@suse.de>
20646 PR tree-optimization/81181
20647 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20648 (compute_antic): ... end of iteration here.
20650 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20652 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20653 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20654 (ftree-slp-vectorize): Likewise.
20655 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20656 can no longer be set independent of flag_tree_loop_vectorize.
20657 * omp-general.c (emp_max_vf): Likewise.
20658 * opts.c (enable_fdo_optimizations): Remove references to
20659 flag_tree_vectorize, these are now implicit.
20660 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20661 and leave it for the options machinery.
20663 2017-08-01 Martin Liska <mliska@suse.cz>
20665 PR middle-end/70140
20666 * builtins.c (expand_builtin_memcpy_args): Remove.
20667 (expand_builtin_memcpy): Call newly added function
20668 expand_builtin_memory_copy_args.
20669 (expand_builtin_memcpy_with_bounds): Likewise.
20670 (expand_builtin_mempcpy): Remove last argument.
20671 (expand_builtin_mempcpy_with_bounds): Likewise.
20672 (expand_builtin_memory_copy_args): New function created from
20673 expand_builtin_mempcpy_args with small modifications.
20674 (expand_builtin_mempcpy_args): Remove.
20675 (expand_builtin_stpcpy): Remove unused argument.
20676 (expand_builtin): Likewise.
20677 (expand_builtin_with_bounds): Likewise.
20679 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20682 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20683 print "ds:" only for immediates in generic address space.
20685 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20688 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20689 (ix86_function_ok_for_sibcall): Return false for naked functions.
20691 2017-08-01 Richard Biener <rguenther@suse.de>
20693 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20694 (compute_antic): Seed worklist with exit block predecessors.
20695 * cfganal.c (dfs_find_deadend): For a cycle return the source
20696 of the edge closing it.
20698 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20700 * config/aarch64/aarch64.c
20701 (aarch64_can_const_movi_rtx_p): Move 0 check.
20703 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20705 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20706 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20708 * match.pd: Ditto in address comparison pattern.
20710 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20712 PR tree-optimization/81627
20713 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20714 closed ssa form for store-store chain.
20716 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20718 PR tree-optimization/81620
20719 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20720 for store-store chain.
20722 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20724 PR tree-optimization/81588
20725 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20726 ranges[i].in_p, invert comparison code ccode. For >/>=,
20727 swap rhs1 and rhs2 and comparison code unconditionally,
20728 for </<= don't do that. Don't swap rhs1/rhs2 again if
20729 ranges[i].in_p, instead invert comparison code ccode if
20730 opcode or oe->rank is BIT_IOR_EXPR.
20733 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20734 a direct optab to conversion optab.
20735 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20736 with GET_MODE_INNER as last argument instead of optab_handler.
20737 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20738 vector extraction if possible and optab is available.
20739 * expr.c (store_constructor): Use convert_optab_handler instead
20740 of optab_handler. Use vector initialization from smaller
20741 vectors if possible and optab is available.
20742 * tree-vect-stmts.c (vectorizable_load): Likewise.
20743 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20744 now have two modes.
20745 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20746 of vec_init from half-sized vectors with the same element mode.
20747 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20748 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20749 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20750 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20751 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20752 after mode in gen_vec_extract* calls.
20753 (vec_extract<mode>): Renamed to ...
20754 (vec_extract<mode><ssescalarmodelower>): ... this.
20755 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20756 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20757 element mode after mode in gen_vec_init* calls.
20758 (VEC_INIT_HALF_MODE): New mode iterator.
20759 (vec_init<mode>): Renamed to ...
20760 (vec_init<mode><ssescalarmodelower>): ... this.
20761 (vec_init<mode><ssehalfvecmodelower>): New expander.
20762 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20763 (vec_extractv2sfsf): ... this.
20764 (vec_initv2sf): Renamed to ...
20765 (vec_initv2sfsf): ... this.
20766 (vec_extractv2si): Renamed to ...
20767 (vec_extractv2sisi): ... this.
20768 (vec_initv2si): Renamed to ...
20769 (vec_initv2sisi): ... this.
20770 (vec_extractv4hi): Renamed to ...
20771 (vec_extractv4hihi): ... this.
20772 (vec_initv4hi): Renamed to ...
20773 (vec_initv4hihi): ... this.
20774 (vec_extractv8qi): Renamed to ...
20775 (vec_extractv8qiqi): ... this.
20776 (vec_initv8qi): Renamed to ...
20777 (vec_initv8qiqi): ... this.
20778 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20779 (vec_init<mode>): Renamed to ...
20780 (vec_init<mode><VEC_base_l>): ... this.
20781 (vec_extract<mode>): Renamed to ...
20782 (vec_extract<mode><VEC_base_l>): ... this.
20783 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20784 (vec_initv2sfsf): ... this.
20785 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20786 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20787 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20788 element mode after mode in gen_vec_init* calls.
20789 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20790 (vec_init<mode><Vel>): ... this.
20791 (vec_extract<mode>): Renamed to ...
20792 (vec_extract<mode><Vel>): ... this.
20793 * config/aarch64/iterators.md (Vel): New mode attribute.
20794 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20795 Add element mode after mode in gen_vec_extract* calls.
20796 * config/s390/vector.md (non_vec_l): New mode attribute.
20797 (vec_extract<mode>): Renamed to ...
20798 (vec_extract<mode><non_vec_l>): ... this.
20799 (vec_init<mode>): Renamed to ...
20800 (vec_init<mode><non_vec_l>): ... this.
20801 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20802 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20804 * config/arm/iterators.md (V_elem_l): New mode attribute.
20805 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20806 (vec_extract<mode><V_elem_l>): ... this.
20807 (vec_extractv2di): Renamed to ...
20808 (vec_extractv2didi): ... this.
20809 (vec_init<mode>): Renamed to ...
20810 (vec_init<mode><V_elem_l>): ... this.
20811 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20812 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20813 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20814 Add element mode after gen_vec_extract* calls.
20815 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20816 (vec_init<mode><unitmode>): ... this.
20817 (vec_extract<mode>): Renamed to ...
20818 (vec_extract<mode><unitmode>): ... this.
20819 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20820 (vec_init<mode><unitmode>): ... this.
20821 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20822 (vec_initv2sfsf): ... this.
20823 (vec_extractv2sf): Renamed to ...
20824 (vec_extractv2sfsf): ... this.
20825 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20826 Add element mode after gen_vec_extract* calls.
20827 * config/mips/mips.md (unitmode): New mode iterator.
20828 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20829 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20830 * config/spu/spu.md (inner_l): New mode attribute.
20831 (vec_init<mode>): Renamed to ...
20832 (vec_init<mode><inner_l>): ... this.
20833 (vec_extract<mode>): Renamed to ...
20834 (vec_extract<mode><inner_l>): ... this.
20835 * config/sparc/sparc.md (veltmode): New mode iterator.
20836 (vec_init<VMALL:mode>): Renamed to ...
20837 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20838 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20839 (vec_initv2sisi): ... this.
20840 (vec_initv2sf): Renamed to ...
20841 (vec_initv2sfsf): ... this.
20842 (vec_extractv2sf): Renamed to ...
20843 (vec_extractv2sfsf): ... this.
20844 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20845 (vec_init<mode>): Renamed to ...
20846 (vec_init<mode><VEC_base_l>): ... this.
20847 (vec_extract<mode>): Renamed to ...
20848 (vec_extract<mode><VEC_base_l>): ... this.
20849 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20850 (vec_initv2sfsf): ... this.
20851 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20852 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20853 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20854 gen_vec_init* calls.
20856 2017-08-01 Richard Biener <rguenther@suse.de>
20858 PR tree-optimization/81297
20859 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20860 TREE_OVERFLOW from INTEGER_CSTs.
20862 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20864 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20866 2017-07-31 Carl Love <cel@us.ibm.com>
20868 * config/rs6000/rs6000-c: Add support for built-in functions
20869 vector signed char vec_xl_be (signed long long, signed char *);
20870 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20871 vector signed int vec_xl_be (signed long long, signed int *);
20872 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20873 vector signed long long vec_xl_be (signed long long, signed long long *);
20874 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20875 vector signed short vec_xl_be (signed long long, signed short *);
20876 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20877 vector double vec_xl_be (signed long long, double *);
20878 vector float vec_xl_be (signed long long, float *);
20879 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20880 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20881 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20883 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20884 (altivec_expand_builtin): Add switch statement to call
20885 altivec_expand_xl_be for each builtin.
20886 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20887 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20888 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20889 __builtin_vsx_le_be_v16qi.
20890 * doc/extend.texi: Update the built-in documentation file for the
20891 new built-in functions.
20893 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20896 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20898 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20900 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20902 * config.gcc: Add z14.
20903 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20904 CPU model numbers for z13s and z14.
20905 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20907 * config/s390/s390-opts.h (enum processor_type): Rename
20908 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20909 * config/s390/s390.c (processor_table): Add field for CPU name to
20910 be passed to Binutils.
20911 (s390_asm_output_machine_for_arch): Use the new field in
20912 processor_table for Binutils.
20913 (s390_expand_builtin): Replace arch12 with z14.
20914 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20915 (s390_get_sched_attrmask): Likewise.
20916 (s390_get_unit_mask): Likewise.
20917 * config/s390/s390.opt: Add z14 to processor_type enum.
20919 2017-07-31 Martin Jambor <mjambor@suse.cz>
20922 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20923 regardless of optimization level.
20925 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20926 Martin Liska <mliska@suse.cz>
20928 * predict.def: Remove old comment and adjust probability.
20929 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20930 PREDICT statements.
20932 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20935 * config/i386/i386.c (ix86_function_naked): New function.
20936 (ix86_can_use_return_insn_p): Return false for naked functions.
20937 (ix86_expand_prologue): Skip prologue for naked functions.
20938 (ix86_expand_epilogue): Skip epilogue for naked functions
20939 and emit trap instruction.
20940 (ix86_warn_func_return): New function.
20941 (ix86_attribute_table): Add "naked" attribute specification.
20942 (TARGET_WARN_FUNC_RETURN): Define.
20943 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20945 2017-07-31 Martin Liska <mliska@suse.cz>
20947 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20948 (dump_gimple_bb_header): Always dump BB info.
20949 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20951 2017-07-31 Martin Liska <mliska@suse.cz>
20954 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20955 also with current_function_decl non-null equality.
20957 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20960 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20961 change type to the element type, instead add eltype variable and
20962 use it where we are interested in the element type.
20964 PR tree-optimization/81603
20965 * ipa-polymorphic-call.c
20966 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20967 offset arithmetic in offset_int, bail out if the resulting bit offset
20968 doesn't fit into shwi.
20970 2017-07-31 Martin Liska <mliska@suse.cz>
20972 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20973 (gimplify_save_expr): Fix comment.
20975 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20978 * config/i386/i386.c (ix86_function_arg): Update arguments for
20980 (ix86_compute_frame_layout): Set the initial stack offset to
20981 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20982 INCOMING_FRAME_SP_OFFSET.
20983 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20984 stack before exception handler returns.
20985 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20986 the 'ERROR_CODE' for exception handler.
20988 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20990 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20991 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20992 (ASM_OUTPUT_REG_POP): Ditto.
20993 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20994 instead of asm_fprintf to output pure string.
20996 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20998 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20999 to imported_module_or_decl hook.
21000 (debug_nothing_tree_tree_tree_bool): Remove.
21001 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21002 * debug.c (do_nothing_debug_hooks): Use
21003 debug_nothing_tree_tree_tree_bool_bool instead of
21004 debug_nothing_tree_tree_tree_bool.
21005 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21006 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21007 * sdbout.c (sdb_debug_hooks): Likewise.
21008 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21009 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21011 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21012 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21013 attribute, don't add anything.
21015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21017 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21018 (fold_build2_stat_loc): Likewise.
21019 (fold_build3_stat_loc): Likewise.
21020 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21021 (fold_build1_loc): Remove macro.
21022 (fold_build2_loc): Likewise.
21023 (fold_build3_loc): Likewise.
21025 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21027 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21028 (gimple_build_debug_bind_source_stat): Likewise.
21029 * gimple.h (gimple_build_debug_bind): Remove macro.
21030 (gimple_build_debug_bind_source): Likewise.
21032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21034 * bitmap.c (bitmap_alloc): Adjust.
21035 (bitmap_gc_alloc): Likewise.
21036 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21040 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21041 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21042 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21043 (bitmap_gc_alloc_stat): Likewise.
21044 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21046 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21048 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21049 * rtl.h (shallow_copy_rtx): Remove macro.
21051 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21053 * emit-rtl.c (gen_raw_REG): Adjust.
21054 * gengenrtl.c (gendef): Likewise.
21055 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21056 * rtl.h (rtx_alloc): Remove macro.
21058 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21060 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21061 (build_tree_list_stat): Likewise.
21062 * tree.h (build_tree_list): Remove macro.
21063 (build_tree_list_vec): Likewise.
21065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21067 * tree.c (make_vector_stat): Remove _stat from name.
21068 (build_vector_stat): Likewise.
21069 * tree.h (make_vector_stat): Remove macro.
21070 (build_vector_stat): Likewise.
21072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21074 * tree.h (build_var_debug_value): Remove prototype.
21076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21078 * tree.c (tree_cons_stat): Remove _stat from name.
21079 * tree.h (tree_cons): Remove macro.
21081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21083 * tree.c (build_vl_exp_stat): Remove _stat from name.
21084 * tree.h (build_vl_exp): Remove macro.
21086 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21088 * tree.c (build_decl_stat): Remove _stat from name.
21089 * tree.h (build_decl): Remove macro.
21091 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21093 * gimple.c (gimple_build_with_ops_stat): Adjust.
21094 (gimple_alloc_stat): Remove _stat from name.
21095 * gimple.h (gimple_alloc): Remove macro.
21097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21099 * tree.c (make_tree_vec_stat): Remove _stat from name.
21100 (grow_tree_vec_stat): Likewise.
21101 * tree.h (make_tree_vec_stat): Adjust prototype.
21102 (grow_tree_vec_stat): Likewise.
21103 (make_tree_vec): Remove macro.
21104 (grow_tree_vec): Likewise.
21106 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21108 * fold-const.c (fold_build1_stat_loc): Adjust.
21109 (fold_build2_stat_loc): Likewise.
21110 (fold_build3_stat_loc): Likewise.
21111 * tree.c (build0_stat): Remove _stat from name.
21112 (build1_stat): Likewise.
21113 (build2_stat): Likewise.
21114 (build3_stat): Likewise.
21115 (build4_stat): Likewise.
21116 (build5_stat): Likewise.
21117 * tree.h (build1_loc): Remove macro, and rename _stat function
21119 (build2_loc): Likewise.
21120 (build3_loc): Likewise.
21121 (build4_loc): Likewise.
21122 (build5_loc): Likewise.
21124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21126 * tree.c (make_int_cst_stat): Remove _stat from name.
21127 * tree.h (make_int_cst_stat): Adjust prototype.
21128 (make_int_cst): Remove macro.
21130 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21132 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21133 * tree.h (make_tree_binfo_stat): Adjust prototype.
21134 (make_tree_binfo): Remove.
21136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21138 * tree.c (copy_node_stat): Rename to copy_node.
21139 (build_distinct_type_copy): Adjust.
21140 * tree.h (copy_node_stat): Adjust prototype.
21141 (copy_node): Remove macro.
21143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21145 * tree.c (make_node_stat): rename to make_node.
21146 (build_tree_list_stat): Adjust.
21147 (build0_stat): Likewise.
21148 (build2_stat): Likewise.
21149 (build3_stat): Likewise.
21150 (build4_stat): Likewise.
21151 (build5_stat): Likewise.
21152 (build_decl_stat): Likewise.
21153 * tree.h (make_node_stat): Adjust prototype.
21154 (make_node): remove macro.
21156 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21158 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21159 (PPC_FEATURE2_SCV): Likewise.
21160 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21162 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21164 * config/aarch64/aarch64.c
21165 (aarch64_internal_mov_immediate): Add new special pattern.
21166 * config/aarch64/aarch64.md (*movdi_aarch64):
21167 Add reg/32bit const mov case.
21169 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21170 Richard Sandiford <richard.sandiford@linaro.org>
21172 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21173 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21174 Add integer and movi cases.
21175 (movi-split-hf-df-sf split, fp16): New.
21176 (enabled): Added TARGET_FP_F16INST.
21177 * config/aarch64/iterators.md (GPF_HF): New.
21178 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21180 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21182 * config/aarch64/aarch64.c
21183 (aarch64_simd_container_mode): Add prototype.
21184 (aarch64_expand_mov_immediate): Add HI support.
21185 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21186 (aarch64_can_const_movi_rtx_p): New.
21187 (aarch64_preferred_reload_class):
21188 Remove restrictions of using FP registers for certain SIMD operations.
21189 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21190 (aarch64_valid_floating_const): Add integer move validation.
21191 (aarch64_simd_imm_scalar_p): Remove.
21192 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21193 (aarch64_legitimate_constant_p): Expand list of supported cases.
21194 * config/aarch64/aarch64-protos.h
21195 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21196 (aarch64_reinterpret_float_as_int): New.
21197 (aarch64_simd_imm_scalar_p): Remove.
21198 * config/aarch64/constraints.md (Uvi): New.
21199 (Dd): Split into Ds and new Dd.
21200 * config/aarch64/aarch64.md (*movsi_aarch64):
21202 (*movdi_aarch64): Add SIMD mov case.
21204 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21206 * tree-predcom.c: (struct chain): Handle store-store chain in which
21207 stores for elimination only store loop invariant values.
21208 (execute_pred_commoning_chain): Ditto.
21209 (prepare_initializers_chain_store_elim): Ditto.
21210 (prepare_finalizers): Ditto.
21211 (is_inv_store_elimination_chain): New function.
21212 (initialize_root_vars_store_elim_1): New function.
21214 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21216 * tree-predcom.c: Revise general description of the pass.
21217 (enum chain_type): New enum type for store elimination.
21218 (struct chain): New field supporting store elimination.
21219 (struct component): Ditto.
21220 (dump_chain): Dump store-stores chain.
21221 (release_chain): Release resources.
21222 (split_data_refs_to_components): Compute and create component
21223 contains only stores for elimination.
21224 (get_chain_last_ref_at): New function.
21225 (make_invariant_chain): Initialization.
21226 (make_rooted_chain): Specify chain type in parameter and record it.
21227 (add_looparound_copies): Skip for store-stores chain.
21228 (determine_roots_comp): Compute type of chain and pass it to
21230 (initialize_root_vars_store_elim_2): New function.
21231 (finalize_eliminated_stores): New function.
21232 (remove_stmt): Handle store for elimination.
21233 (execute_pred_commoning_chain): Execute predictive commoning on
21234 store-store chains.
21235 (determine_unroll_factor): Skip unroll for store-stores chain.
21236 (prepare_initializers_chain_store_elim): New function.
21237 (prepare_initializers_chain): Hanlde store-store chain.
21238 (prepare_finalizers_chain, prepare_finalizers): New function.
21239 (tree_predictive_commoning_loop): Return integer value indicating
21240 if loop is unrolled or lcssa form is corrupted.
21241 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21243 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21245 * tree-predcom.c (initialize_root): Delete.
21246 (execute_pred_commoning_chain): Initialize root vars and replace
21247 reference of non-combined chain directly, rather than call above
21250 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21252 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21253 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21255 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21257 * tree-predcom.c (struct chain): New field init_seq.
21258 (release_chain): Release init_seq.
21259 (prepare_initializers_chain): Record intialization stmts in above
21261 (insert_init_seqs): New function.
21262 (tree_predictive_commoning_loop): Call insert_init_seqs.
21264 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21266 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21268 2017-07-28 Richard Biener <rguenther@suse.de>
21270 * match.pd: Remove superfluous :c.
21271 * genmatch.c (simplify::id): Add member.
21272 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21274 (current_id): New global.
21275 (dt_node::parent): Move from ...
21276 (dt_operand::parent): ... here. Add for_id member.
21277 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21278 (decision_tree::find_node): Relax order requirement when
21279 merging DT_TRUE nodes to ones inbetween the current simplify
21280 and the one we try to merge with. Add diagnostic whenever
21281 we need to enforce pattern order by not merging.
21282 (decision_tree::insert): Set current_id.
21283 (decision_tree::print_node): Dump parent node and for_id.
21284 (parser::last_id): Add member.
21285 (parser::push_simplify): Assign unique id.
21286 (parser::parser): Initialize last_id.
21288 2017-07-28 Martin Liska <mliska@suse.cz>
21291 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21292 gimple_build_debug_bind.
21294 2017-07-28 Richard Biener <rguenther@suse.de>
21296 PR tree-optimization/81502
21297 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21299 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21301 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21302 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21303 for BIT_FIELD_REF args.
21304 * fold-const.c (make_bit_field_ref): Likewise.
21305 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21307 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21310 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21311 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21312 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21313 Or it into SANITIZER_UNDEFINED.
21314 * ubsan.c: Include gimple-fold.h and varasm.h.
21315 (ubsan_expand_ptr_ifn): New function.
21316 (instrument_pointer_overflow): New function.
21317 (maybe_instrument_pointer_overflow): New function.
21318 (instrument_object_size): Formatting fix.
21319 (pass_ubsan::execute): Call instrument_pointer_overflow
21320 and maybe_instrument_pointer_overflow.
21321 * internal-fn.c (expand_UBSAN_PTR): New function.
21322 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21323 * sanitizer.def (__ubsan_handle_pointer_overflow,
21324 __ubsan_handle_pointer_overflow_abort): New builtins.
21325 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21326 * internal-fn.def (UBSAN_PTR): New internal function.
21327 * opts.c (sanitizer_opts): Add pointer-overflow.
21328 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21329 * fold-const.c (build_range_check): Compute pointer range check in
21330 integral type if pointer arithmetics would be needed. Formatting
21333 2017-07-28 Martin Liska <mliska@suse.cz>
21336 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21337 parameters that are of a variable-length.
21339 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21341 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21343 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21344 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21345 (CRT_CALL_STATIC_FUNCTION): Likewise.
21346 (ASM_DEFAULT_SPEC): New define.
21347 (ASM_SPEC32): Likewise.
21348 (ASM_SPEC64): Likewise.
21349 (ASM_SPEC_COMMON): Likewise.
21350 (ASM_SPEC): Likewise.
21351 (INVALID_64BIT): Likewise.
21352 (LINK_OS_DEFAULT_SPEC): Likewise.
21353 (LINK_OS_SPEC32): Likewise.
21354 (LINK_OS_SPEC64): Likewise.
21355 (POWERPC_LINUX): Likewise.
21356 (PTRDIFF_TYPE): Likewise.
21357 (RESTORE_FP_PREFIX): Likewise.
21358 (RESTORE_FP_SUFFIX): Likewise.
21359 (SAVE_FP_PREFIX): Likewise.
21360 (SAVE_FP_SUFFIX): Likewise.
21361 (SIZE_TYPE): Likewise.
21362 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21363 (TARGET_64BIT): Likewise.
21364 (TARGET_64BIT): Likewise.
21365 (TARGET_AIX): Likewise.
21366 (WCHAR_TYPE_SIZE): Likewise.
21367 (WCHAR_TYPE): Undefine.
21368 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21369 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21370 (CPP_OS_RTEMS_SPEC): Delete.
21371 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21372 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21374 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21376 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21378 PR tree-optimization/81578
21379 * tree-parloops.c (build_new_reduction): Bail out if
21380 reduction_code isn't one of the standard OpenMP reductions.
21381 Move the details printing after that decision.
21383 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21385 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21386 related to reload_in_progress.
21387 (splat_input_operand): Likewise.
21388 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21390 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21392 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21393 (TARGET_INSTANTIATE_DECLS): Likewise.
21394 (legitimate_indexed_address_p): Delete reload_in_progress code.
21395 (rs6000_debug_legitimate_address_p): Likewise.
21396 (rs6000_eliminate_indexed_memrefs): Likewise.
21397 (rs6000_emit_le_vsx_store): Likewise.
21398 (rs6000_emit_move_si_sf_subreg): Likewise.
21399 (rs6000_emit_move): Likewise.
21400 (register_to_reg_type): Likewise.
21401 (rs6000_pre_atomic_barrier): Likewise.
21402 (rs6000_machopic_legitimize_pic_address): Likewise.
21403 (rs6000_allocate_stack_temp): Likewise.
21404 (rs6000_address_for_fpconvert): Likewise.
21405 (rs6000_address_for_altivec): Likewise.
21406 (rs6000_secondary_memory_needed_rtx): Delete function.
21407 (rs6000_check_sdmode): Likewise.
21408 (rs6000_alloc_sdmode_stack_slot): Likewise.
21409 (rs6000_instantiate_decls): Likewise.
21410 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21411 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21412 Delete reload_in_progress.
21413 (*vec_reload_and_plus_<mptrsize>): Likewise.
21414 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21415 (vsx_div_v2di): Likewise.
21416 (vsx_udiv_v2di): Likewise.
21418 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21420 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21421 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21422 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21423 (rs6000_debug_reg_global): Delete print of LRA status.
21424 (rs6000_option_override_internal): Delete dead LRA related code.
21425 (rs6000_lra_p): Delete function.
21426 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21428 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21430 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21431 * config/riscv/rtems.h: New file.
21433 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21434 Sudakshina Das <sudi.das@arm.com>
21436 * config/aarch64/aarch64.md
21437 (define_split for and<mode>3nr_compare): Move
21438 non aarch64_logical_operand to a register.
21439 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21440 register immediate operand to a register.
21441 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21443 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21445 PR middle-end/81564
21446 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21448 2017-07-27 Richard Biener <rguenther@suse.de>
21450 PR tree-optimization/81573
21451 PR tree-optimization/81494
21452 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21453 multi defuse cycle case.
21455 2017-07-27 Richard Biener <rguenther@suse.de>
21457 PR tree-optimization/81571
21458 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21461 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21463 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21464 earlier and only if MASK_FPU is set. Adjust formatting.
21466 2017-07-27 Martin Liska <mliska@suse.cz>
21468 * opt-functions.awk: Add validation of value of Init.
21469 * optc-gen.awk: Pass new argument.
21471 2017-07-27 Martin Liska <mliska@suse.cz>
21473 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21474 Fix wrong condition.
21476 2017-07-27 Martin Liska <mliska@suse.cz>
21478 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21479 BBs and edges seen by autoFDO.
21481 2017-07-27 Richard Biener <rguenther@suse.de>
21483 PR tree-optimization/81502
21484 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21485 with incompatible but same sized type.
21486 (execute_update_addresses_taken): Likewise.
21488 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21490 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21491 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21493 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21496 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21497 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21498 Change s_operand to memory_operand.
21500 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21502 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21503 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21504 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21505 Emit instructions rather than returning an expression. Handle TFmode
21506 and KFmode by casting to TImode.
21507 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21508 (rs6000_emit_le_vsx_store): Likewise.
21509 * config/rs6000/vsx.md (VSX_TI): New iterator.
21510 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21511 (*vsx_le_undo_permute_<mode>): Likewise.
21512 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21513 emit the split sequence.
21514 (*vsx_le_perm_store_<mode>): Likewise.
21516 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21518 PR tree-optimization/81555
21519 PR tree-optimization/81556
21520 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21521 if true, force CHANGED for the recursive invocation.
21522 (reassociate_bb): Remember original length of ops array, pass
21523 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21525 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21526 attributes for noipa attribute. For naked attribute use
21527 lookup_attribute first before lookup_attribute_spec.
21528 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21530 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21531 for functions with noipa attribute.
21532 (cgraph_externally_visible_p): Return true for functions with noipa
21534 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21535 for functions with noipa attribute.
21536 * doc/extend.texi: Document noipa function attribute.
21537 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21538 also for functions with noipa attribute.
21539 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21541 2017-07-26 Andrew Pinski <apinski@cavium.com>
21543 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21544 vec_unalign_load_cost and vec_unalign_store_cost.
21546 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21548 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21549 -mvsx-small-integer option.
21550 (ISA_3_0_MASKS_IEEE): Likewise.
21551 (OTHER_VSX_VECTOR_MASKS): Likewise.
21552 (POWERPC_MASKS): Likewise.
21553 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21554 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21555 code, only testing for DImode being allowed in non-VSX floating
21557 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21558 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21560 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21561 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21562 TARGET_P8_VECTOR test.
21563 (rs6000_secondary_reload_simple_move): Likewise.
21564 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21565 since TARGET_P9_VECTOR was already tested.
21566 (rs6000_opt_masks): Remove -mvsx-small-integer.
21567 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21568 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21570 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21571 test for TARGET_VEXTRACTUB was used, and that uses
21573 (p9 extract splitter): Likewise.
21574 (vsx_extract_<mode>_di_p9): Likewise.
21575 (vsx_extract_<mode>_store_p9): Likewise.
21576 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21577 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21578 the elimination of TARGET_VSX_SMALL_INTEGER.
21579 (vsx_extract_<mode>_p8): Likewise.
21580 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21581 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21582 (vsx_set_<mode>_p9): Likewise.
21583 (vsx_set_v4sf_p9): Likewise.
21584 (vsx_set_v4sf_p9_zero): Likewise.
21585 (vsx_insert_extract_v4sf_p9): Likewise.
21586 (vsx_insert_extract_v4sf_p9_2): Likewise.
21587 * config/rs6000/rs6000.md (sign extend splitter): Change
21588 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21589 (floatsi<mode>2_lfiwax_mem): Likewise.
21590 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21591 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21592 since a test for TARGET_P9_VECTOR was used.
21593 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21594 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21595 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21596 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21597 TARGET_P8_VECTOR test.
21598 (fix_trunc<mode>si2_stfiwx): Likewise.
21599 (fix_trunc<mode>si2_internal): Likewise.
21600 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21601 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21603 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21604 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21605 TARGET_P8_VECTOR test.
21606 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21607 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21608 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21610 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21611 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21612 since a test for TARGET_P9_VECTOR was used.
21613 (splitter for loading small constants): Likewise.
21615 2017-07-26 Andrew Pinski <apinski@cavium.com>
21617 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21620 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21623 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21624 (fp_valid_at): Likewise.
21626 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21628 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21629 (qdf24xx_addrcost_table): Likewise.
21630 (cortexa57_tunings): Update to use generic_branch_cost.
21631 (cortexa72_tunings): Likewise.
21632 (cortexa73_tunings): Likewise.
21633 (qdf24xx_tunings): Likewise.
21635 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21637 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21638 (thunderx2t99_branch_cost): Likewise.
21639 (cortexa35_tunings): Update to use generic_branch_cost.
21640 (cortexa53_tunings): Likewise.
21641 (cortexa57_tunings): Likewise.
21642 (cortexa72_tunings): Likewise.
21643 (cortexa73_tunings): Likewise.
21644 (thunderx2t99_tunings): Likewise.
21646 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21648 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21649 (sparc_option_override): Honour MASK_FSMULD.
21650 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21651 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21652 * config/sparc/sparc.opt (mfsmuld): New option.
21653 * doc/invoke.texi (mfsmuld): Document option.
21655 2017-07-26 Marek Polacek <polacek@redhat.com>
21657 PR middle-end/70992
21658 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21660 2017-07-26 Richard Biener <rguenther@suse.de>
21662 * gimple-match-head.c (do_valueize): Return OP if valueize
21664 (get_def): New helper to get at the def stmt of a SSA name
21665 if valueize allows.
21666 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21667 do_valueize to get at the def stmt.
21668 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21670 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21672 PR middle-end/46932
21673 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21675 2017-07-26 Martin Liska <mliska@suse.cz>
21678 * function.c (expand_function_start): Make expansion of
21679 nonlocal_goto_save_area after parm_birth_insn.
21681 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21683 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21684 from all CPU target flags enable members.
21686 2017-07-26 Richard Biener <rguenther@suse.de>
21688 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21689 (decision_tree::gen): Make 'type' const.
21690 (write_predicate): Likewise.
21692 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21694 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21695 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21696 (rs6000_option_override_internal): Likewise.
21697 (rs6000_expand_vector_set): Likewise.
21698 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21699 (TARGET_UPPER_REGS_SF): Likewise.
21700 (TARGET_UPPER_REGS_DI): Likewise.
21701 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21702 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21703 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21704 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21705 (Splitters for DI constants in Altivec registers): Likewise.
21706 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21707 (vsx_set_v4sf_p9): Likewise.
21708 (vsx_set_v4sf_p9_zero): Likewise.
21709 (vsx_insert_extract_v4sf_p9): Likewise.
21710 (vsx_insert_extract_v4sf_p9_2): Likewise.
21712 2017-07-25 Carl Love <cel@us.ibm.com>
21714 * doc/extend.texi: Update the built-in documentation file for the
21715 existing built-in functions
21716 vector signed char vec_cnttz (vector signed char);
21717 vector unsigned char vec_cnttz (vector unsigned char);
21718 vector signed short vec_cnttz (vector signed short);
21719 vector unsigned short vec_cnttz (vector unsigned short);
21720 vector signed int vec_cnttz (vector signed int);
21721 vector unsigned int vec_cnttz (vector unsigned int);
21722 vector signed long long vec_cnttz (vector signed long long);
21723 vector unsigned long long vec_cnttz (vector unsigned long long);
21725 2017-07-25 Andrew Pinski <apinski@cavium.com>
21727 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21728 accesses where the use is for the first operand of a BIT_INSERT.
21730 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21733 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21734 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21736 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21738 * config/i386/gstabs.h: Delete.
21739 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21741 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21743 * config/i386/i386.c (ix86_decompose_address): Do not check for
21744 register RTX when looking at index_reg or base_reg.
21745 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21747 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21749 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21750 to update EH info here.
21752 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21754 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21756 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21758 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21760 2017-07-25 Torsten Duwe <duwe@suse.de>
21762 * common.opt: Introduce -fpatchable-function-entry
21763 command line option, and its variables function_entry_patch_area_size
21764 and function_entry_patch_area_start.
21765 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21766 including a two-value parser.
21767 * target.def (print_patchable_function_entry): New target hook.
21768 * targhooks.h (default_print_patchable_function_entry): New function.
21769 * targhooks.c (default_print_patchable_function_entry): Likewise.
21770 * toplev.c (process_options): Switch off IPA-RA if
21771 patchable function entries are being generated.
21772 * varasm.c (assemble_start_function): Look at the
21773 patchable-function-entry command line switch and current
21774 function attributes and maybe generate NOP instructions by
21775 calling the print_patchable_function_entry hook.
21776 * doc/extend.texi: Document patchable_function_entry attribute.
21777 * doc/invoke.texi: Document -fpatchable_function_entry
21778 command line option.
21779 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21781 * doc/tm.texi: Re-generate.
21783 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21786 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21787 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21789 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21791 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21793 2017-07-25 Richard Biener <rguenther@suse.de>
21795 PR tree-optimization/81455
21796 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21797 not walk in cycles when looking for guards.
21799 2017-07-25 Richard Biener <rguenther@suse.de>
21801 PR tree-optimization/81529
21802 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21803 when optimizing backedge uses.
21805 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21807 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21809 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21810 to dl_section_ref. On AIX, append an expression to subtract
21811 the size of the section length to dl_section_ref.
21813 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21815 * configure.ac: If any of the config.* scripts fail, exit 1.
21816 * configure: Regenerate.
21818 2017-07-25 Richard Biener <rguenther@suse.de>
21820 PR middle-end/81546
21821 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21822 of immediate uses, be more verbose on errors.
21824 2017-07-25 Richard Biener <rguenther@suse.de>
21826 PR tree-optimization/81510
21827 * tree-vect-loop.c (vect_is_simple_reduction): When the
21828 reduction stmt is not inside the loop bail out.
21830 2017-07-25 Richard Biener <rguenther@suse.de>
21832 PR tree-optimization/81303
21833 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21834 profitability check against LOOP_VINFO_NITERSM1.
21836 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21838 * domwalk.c (cmp_bb_postorder): Simplify.
21839 (sort_bbs_postorder): New function. Use it...
21840 (dom_walker::walk): ...here to optimize common cases.
21842 2017-07-25 Martin Liska <mliska@suse.cz>
21845 * ipa-visibility.c (function_and_variable_visibility): Make the
21846 redirection just on target that supports aliasing.
21847 Fix GNU coding style.
21849 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21852 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21854 * config/arm/rtems.h: Likewise.
21855 * config/bfin/rtems.h: Likewise.
21856 * config/i386/rtemself.h: Likewise.
21857 * config/lm32/rtems.h: Likewise.
21858 * config/m32c/rtems.h: Likewise.
21859 * config/m68k/rtemself.h: Likewise.
21860 * config/microblaze/rtems.h: Likewise.
21861 * config/mips/rtems.h: Likewise.
21862 * config/moxie/rtems.h: Likewise.
21863 * config/nios2/rtems.h: Likewise.
21864 * config/powerpcspe/rtems.h: Likewise.
21865 * config/rs6000/rtems.h: Likewise.
21866 * config/rtems.h: Likewise.
21867 * config/sh/rtems.h: Likewise.
21868 * config/sh/rtemself.h: Likewise.
21869 * config/sparc/rtemself.h: Likewise.
21871 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21874 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21875 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21876 * tree-ssa-structalias.c (alias_get_name): Same.
21878 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21881 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21882 instructions if no du chain is found.
21884 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21886 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21888 2017-07-25 Richard Biener <rguenther@suse.de>
21890 PR middle-end/81505
21891 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21894 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21896 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21897 upper-regs options.
21898 (ISA_2_7_MASKS_SERVER): Likewise.
21899 (ISA_3_0_MASKS_IEEE): Likewise.
21900 (OTHER_P8_VECTOR_MASKS): Likewise.
21901 (OTHER_VSX_VECTOR_MASKS): Likewise.
21902 (POWERPC_MASKS): Likewise.
21903 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21904 duplicate list of options.
21905 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21906 explicit -mupper-regs options.
21907 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21908 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21909 alias for -mupper-regs-df.
21910 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21911 (rs6000_init_hard_regno_mode_ok): Likewise.
21912 (rs6000_option_override_internal): Likewise.
21913 (rs6000_opt_masks): Likewise.
21914 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21915 options in terms of whether -mvsx or -mpower8-vector was used.
21916 (TARGET_UPPER_REGS_DI): Likewise.
21917 (TARGET_UPPER_REGS_SF): Likewise.
21918 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21919 -mupper-regs-* options.
21921 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21923 * passes.c (emergency_dump_function): Print some empty lines and a
21924 header before the RTL dump.
21926 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21928 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21930 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21933 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21934 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21936 2017-07-24 Carl Love <cel@us.ibm.com>
21938 * config/rs6000/rs6000-c.c: Add support for built-in functions
21939 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21940 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21941 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21942 vec_extract_fp_from_shortl): Add defines for the two builtins.
21943 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21944 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21946 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21947 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21948 * doc/extend.texi: Update the built-in documentation file for the
21949 new built-in function.
21951 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21954 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21956 * doc/generic.texi: Likewise.
21957 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21958 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21960 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21962 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21963 (aarch64_mls_elt_merge<mode>): Likewise.
21965 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21967 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21968 having __cxa_atexit.
21970 2017-07-23 Michael Collison <michael.collison@arm.com>
21972 * config/arm/arm.c (arm_option_override): Deprecate
21973 use of -mstructure-size-boundary.
21974 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21975 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21977 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21980 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21981 Reduce cost estimate for direct moves.
21983 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21986 * config/i386/i386.c (ix86_option_override_internal): Disable
21987 BMI, BMI2 and TBM instructions for -m16.
21989 2017-07-21 Carl Love <cel@us.ibm.com>
21991 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21992 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21993 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21994 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21995 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21996 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21997 VMULOSW): New enum "unspec" values.
21998 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21999 altivec_vmulosw): New patterns.
22000 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22001 VMULOSW): Add definitions.
22003 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22005 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22006 (qdf24xx): Likewise.
22007 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22008 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22009 (AARCH64_FL_V8_1): Renumber.
22010 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22011 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22012 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22013 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22014 rdma to feature modifiers list.
22016 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22018 PR middle-end/56727
22019 * ipa-visibility (function_and_variable_visibility): Convert
22020 recursive PLT call to direct call if appropriate.
22022 2017-07-21 Andrew Pinski <apinski@cavium.com>
22024 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22025 operand 1 to see if the types precision matches.
22026 * fold-const.c (operand_equal_p): Likewise.
22028 2017-07-21 Richard Biener <rguenther@suse.de>
22030 PR tree-optimization/81303
22031 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22032 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22033 (vect_peeling_hash_get_lowest_cost): Adjust.
22034 (vect_enhance_data_refs_alignment): Likewise. Use
22035 vect_get_peeling_costs_all_drs to compute the penalty for no
22036 peeling to match up costs.
22038 2017-07-21 Richard Biener <rguenther@suse.de>
22040 PR tree-optimization/81500
22041 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22042 we didn't identify a reduction path.
22044 2017-07-21 Tom de Vries <tom@codesourcery.com>
22045 Cesar Philippidis <cesar@codesourcery.com>
22047 PR gcov-profile/81442
22048 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22051 2017-07-21 Tom de Vries <tom@codesourcery.com>
22054 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22056 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22057 nvptx_override_options_after_change.
22059 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22061 * dwarf2out.c (output_file_names): Avoid double testing for
22062 dwarf_version >= 5.
22064 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22066 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22068 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22070 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22072 (try_crossjump_to_edge): Do not punt on partitioned functions.
22074 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22076 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22077 Put all BBs reachable only via paths crossing cold region to cold
22079 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22081 2016-07-21 Richard Biener <rguenther@suse.de>
22083 PR tree-optimization/81303
22084 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22085 into account prologue and epilogue iterations when raising
22086 min_profitable_iters to sth at least covering one vector iteration.
22088 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22090 * config/arm/arm.c (arm_test_cpu_arch_dat):
22093 2017-07-20 Nathan Sidwell <nathan@acm.org>
22095 Remove TYPE_METHODS.
22096 * tree.h (TYPE_METHODS): Delete.
22097 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22098 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22099 (dbxout_type_methods): Scan TYPE_FIELDS.
22100 (dbxout_type): Don't check TYPE_METHODS here.
22101 * function.c (use_register_for_decl): Always ignore register for
22102 class types when not optimizing.
22103 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22104 * tree.c (free_lang_data_in_type): Stitch out member functions and
22105 templates from TYPE_FIELDS.
22106 (build_distinct_type_copy, verify_type_variant,
22107 verify_type): Member fns are on TYPE_FIELDS.
22108 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22109 * tree-pretty-print.c (dump_generic_node): Likewise.
22111 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22114 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22115 V2TImode and V4TImode.
22116 (ix86_expand_vector_extract): Likewise.
22117 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22118 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22119 (ssescalarmode): Handle V4TImode and V2TImode.
22120 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22121 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22122 (VEXTRACTI128_MODE): New mode iterator.
22123 (splitter for *vec_extractv?ti first element): New.
22124 (VEC_INIT_MODE): New mode iterator.
22125 (vec_init<mode>): Consolidate 3 expanders into one using
22126 VEC_INIT_MODE mode iterator.
22128 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22130 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22131 non_spilled_static_chain_regno_p.
22133 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22135 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22137 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22139 * bb-reorder.c (connect_traces): Allow copying of blocks within
22142 2017-07-20 Richard Biener <rguenther@suse.de>
22144 * gimple.h (gimple_phi_result): Add gphi * overload.
22145 (gimple_phi_result_ptr): Likewise.
22146 (gimple_phi_arg): Likewise. Adjust index assert to only
22147 allow actual argument accesses rather than all slots available
22149 (gimple_phi_arg_def): Add gphi * overload.
22150 * tree-phinodes.c (make_phi_node): Initialize only actual
22152 (resize_phi_node): Clear memory not covered by old node,
22153 do not initialize excess argument slots.
22154 (reserve_phi_args_for_new_edge): Initialize new argument slot
22157 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22159 PR tree-optimization/81388
22161 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22163 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22164 by removing computation of may_be_zero.
22166 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22167 Tom de Vries <tom@codesourcery.com>
22169 PR middle-end/81030
22170 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22171 when gimple level profile disagrees with what RTL expander did.
22173 2017-07-20 Richard Biener <rguenther@suse.de>
22175 PR tree-optimization/61171
22176 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22177 (vect_analyze_stmt): Add slp instance parameter.
22178 (vectorizable_reduction): Likewise.
22179 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22180 (vect_is_simple_reduction): Deal with chains not detected
22181 as SLP reduction chain, specifically not properly associated
22182 chains containing a mix of plus/minus.
22183 (get_reduction_op): Remove.
22184 (get_initial_defs_for_reduction): Simplify, pass in whether
22185 this is a reduction chain, pass in the SLP node for the PHIs.
22186 (vect_create_epilog_for_reduction): Get the SLP instance as
22188 (vectorizable_reduction): Get the SLP instance as arg.
22189 During analysis remember the SLP node with the PHIs in the
22190 instance. Simplify getting at the vectorized reduction PHIs.
22191 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22192 through SLP instance.
22193 (vect_slp_analyze_operations): Likewise.
22194 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22195 (vect_transform_stmt): Likewise.
22197 2017-07-20 Tom de Vries <tom@codesourcery.com>
22199 PR tree-optimization/81489
22200 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22201 read of phi arg location to before loop that modifies phi.
22203 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22205 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22208 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22210 PR middle-end/81331
22211 * except.c (execute): Fix ordering issue.
22213 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22215 PR rtl-optimization/81423
22216 * combine.c (make_compound_operation_int): Don't try to optimize
22217 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22219 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22221 PR rtl-optimization/81423
22222 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22223 with a constant that is -1 in the truncated to mode.
22225 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22227 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22228 (determine_unlikely_bbs): ... here.
22229 * predict.h (propagate_unlikely_bbs_forward): Declare.
22230 * cfgexpand.c (pass_expand::execute): Use it.
22231 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22233 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22234 propagate_unlikely_bbs_forward.
22236 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22238 PR middle-end/81331
22239 * except.c (maybe_add_nop_after_section_switch): New function.
22242 2017-07-19 Tom de Vries <tom@codesourcery.com>
22244 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22246 2017-07-19 Tom de Vries <tom@codesourcery.com>
22248 * gimple.h (gimple_phi_arg): Make assert more strict.
22250 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22252 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22253 * config/rs6000/mmintrin.h: New file.
22254 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22256 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22258 PR tree-optimization/81346
22259 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22261 2017-07-19 Tom de Vries <tom@codesourcery.com>
22263 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22265 2017-07-19 Tom de Vries <tom@codesourcery.com>
22267 * config/nvptx/nvptx-modes.def: Add V2DImode.
22268 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22269 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22270 (nvptx_output_mov_insn): Handle lack of mov.b128.
22271 (nvptx_print_operand): Handle 'H' and 'L' codes.
22272 (nvptx_vector_mode_supported): Allow V2DImode.
22273 (nvptx_preferred_simd_mode): New function.
22274 (nvptx_data_alignment): New function.
22275 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22276 nvptx_preferred_simd_mode.
22277 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22279 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22281 2017-07-19 Tom de Vries <tom@codesourcery.com>
22283 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22284 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22285 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22286 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22287 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22288 (mov<VECIM>_insn): New define_insn.
22289 (define_expand "mov<VECIM>): New define_expand.
22291 2017-07-19 Tom de Vries <tom@codesourcery.com>
22293 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22295 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22297 PR tree-optimization/81346
22298 * fold-const.h (fold_div_compare, range_check_type): Declare.
22299 * fold-const.c (range_check_type): New function.
22300 (build_range_check): Use range_check_type.
22301 (fold_div_compare): No longer static, rewritten into
22302 a match.pd helper function.
22303 (fold_comparison): Don't call fold_div_compare here.
22304 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22305 as helper function.
22307 2017-07-19 Nathan Sidwell <nathan@acm.org>
22309 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22310 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22311 * tree.c (find_decls_types_r, verify_type): Use
22312 TYPE_{MIN,MAX}_VALUE_RAW.
22313 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22314 (hash_tree): Likewise.
22315 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22317 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22320 2017-07-18 Tom de Vries <tom@codesourcery.com>
22322 PR middle-end/81464
22323 * omp-expand.c (expand_omp_for_static_chunk): Handle
22324 equal-argument loop exit phi.
22326 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22329 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22330 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22331 operand 2 predicate.
22332 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22333 operand 2 predicate.
22334 (ror,rol -> rorx splitters): Use const_int_operand as
22335 operand 2 predicate.
22337 2017-06-18 Richard Biener <rguenther@suse.de>
22339 PR tree-optimization/81410
22340 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22341 the gap in the ! slp_perm SLP case after each group.
22343 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22345 PR middle-end/81463
22346 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22349 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22351 PR middle-end/81462
22352 * predict.c (set_even_probabilities): Cleanup; do not affect
22353 probabilities that are already known.
22354 (combine_predictions_for_bb): Call even when count is set.
22356 2017-07-18 Nathan Sidwell <nathan@acm.org>
22358 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22361 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22364 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22365 optimization for loop niter analysis.
22367 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22370 * config/avr/avr.c (avr_optimize_casesi): Don't use
22371 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22373 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22375 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22376 body_cost_vec from _vect_peel_extended_info.
22377 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22378 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22381 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22383 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22385 2017-07-18 Richard Biener <rguenther@suse.de>
22387 PR tree-optimization/80620
22388 PR tree-optimization/81403
22389 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22390 info when re-using a VN table entry.
22392 2017-07-18 Richard Biener <rguenther@suse.de>
22394 PR tree-optimization/81418
22395 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22396 vectype_in. Verify that with lane-reducing reduction operations
22397 we have a single def-use cycle.
22399 2017-07-17 Carl Love <cel@us.ibm.com>
22401 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22403 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22404 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22405 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22406 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22407 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22408 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22409 VMULOSW): New enum "unspec" values.
22410 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22411 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22412 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22413 altivec_vmulosw): New patterns.
22414 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22415 VMULOSW): Add definitions.
22417 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22419 * config/alpha/alpha.c: Include predict.h.
22421 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22423 * tree-vrp.c (compare_assert_loc): Fix comparison function
22424 to return predictable results.
22426 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22428 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22430 (subdi3): Likewise.
22431 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22432 * doc/invoke.texi (mexpand-adddi): Update text.
22434 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22436 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22437 that also clobbers the CC register. The old expand code is moved
22439 (*arc_clzsi2): ... here.
22440 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22441 the CC register. The old expand code is moved to ...
22442 (arc_ctzsi2): ... here.
22444 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22446 * config/arc/arc.opt (mindexed-loads): Use initial value
22447 TARGET_INDEXED_LOADS_DEFAULT.
22448 (mauto-modify-reg): Use initial value
22449 TARGET_AUTO_MODIFY_REG_DEFAULT.
22450 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22451 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22452 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22453 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22455 2017-07-17 Martin Liska <mliska@suse.cz>
22458 * opts.c (finish_options): Do not allow -fgnu-tm
22459 w/ -fsanitize={kernel-,}address. Say sorry.
22461 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22464 * tree-loop-distribution.c (classify_partition): Only assert on
22465 numer of iterations.
22466 (merge_dep_scc_partitions): Delete prameter. Update function call.
22467 (distribute_loop): Remove code handling loop with unknown niters.
22468 (pass_loop_distribution::execute): Skip loop with unknown niters.
22470 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22473 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22474 function sort_partitions_by_post_order.
22476 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22478 PR tree-optimization/81374
22479 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22480 the max index of basic blocks, rather than number of basic blocks.
22482 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22484 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22486 (arc_legitimate_pic_operand_p): Likewise.
22487 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22489 (arc_needs_pcl_p): Likewise.
22490 (arc_legitimate_pc_offset_p): Likewise.
22491 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22492 function is also used in constrains.md.
22493 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22494 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22495 PLUS. Only return true/false in known cases, otherwise assert.
22496 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22497 is already called in arc_legitimate_constant_p.
22498 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22500 (LEGITIMATE_PIC_OPERAND_P): Use
22501 arc_raw_symbolic_reference_mentioned_p function.
22502 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22507 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22508 Andrew Burgess <andrew.burgess@embecosm.com>
22510 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22511 (arc_return_address_register): New function.
22512 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22513 (arc_handle_fndecl_attribute): Add naked attribute.
22514 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22515 (TARGET_WARN_FUNC_RETURN): Likewise.
22516 (arc_allocate_stack_slots_for_args): New function.
22517 (arc_warn_func_return): Likewise.
22518 (machine_function): Change type fn_type.
22519 (arc_compute_function_type): Consider new naked function type,
22520 change function return type.
22521 (arc_must_save_register): Adapt to handle new
22522 arc_compute_function_type's return type.
22523 (arc_expand_prologue): Likewise.
22524 (arc_expand_epilogue): Likewise.
22525 (arc_return_address_regs): Delete.
22526 (arc_return_address_register): New function.
22527 (arc_epilogue_uses): Use above function.
22528 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22529 (arc_function_type): Change encoding, add naked type.
22530 (ARC_INTERRUPT_P): Change to handle the new encoding.
22531 (ARC_FAST_INTERRUPT_P): Likewise.
22532 (ARC_NORMAL_P): Define.
22533 (ARC_NAKED_P): Likewise.
22534 (arc_compute_function_type): Delete prototype.
22535 * config/arc/arc.md (in_ret_delay_slot): Use
22536 arc_return_address_register function.
22537 (simple_return): Likewise.
22538 (p_return_i): Likewise.
22540 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22542 PR tree-optimization/81428
22543 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22544 can't be built for those types.
22546 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22548 Remove stuff dead since r239246.
22550 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22551 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22552 (avr_inform_devices): Remove dead stuff.
22554 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22556 * config/arm/arm_neon.h: Fix softp typo.
22558 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22560 PR tree-optimization/81365
22561 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22562 aggregate moves onto bb predecessor edges, make sure there are no
22563 loads that could alias the lhs in between the start of bb and the
22566 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22569 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22570 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22571 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22573 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22575 PR tree-optimization/81396
22576 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22577 (init_symbolic_number): Initialize it to 1.
22578 (perform_symbolic_merge): Add n_ops from both operands into the new
22580 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22581 without base_addr as useless if they need more than one operation.
22582 (bswap_replace): Handle !bswap case for NULL base_addr.
22584 2017-07-17 Tom de Vries <tom@codesourcery.com>
22587 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22590 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22592 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22593 conditional builtin define __FIX_LEON3FT_B2BST.
22595 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22597 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22598 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22601 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22603 PR rtl-optimization/81424
22604 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22605 to remove potential trapping from operands if -fnon-call-exceptions.
22607 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22609 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22610 profile_proability for scalling.
22611 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22613 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22615 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22617 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22619 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22620 fixpoint arithmetics.
22622 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22624 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22625 fixpoint arithmetics.
22627 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22629 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22630 fixpoint arithmetics.
22632 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22634 * profile-count.h (profile_probability::from_reg_br_prob_note,
22635 profile_probability::to_reg_br_prob_note): New functions.
22636 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22637 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22638 * predict.c (probability_reliable_p): Update.
22639 (edge_probability_reliable_p): Update.
22640 (br_prob_note_reliable_p): Update.
22641 (invert_br_probabilities): Update.
22642 (add_reg_br_prob_note): New function.
22643 (combine_predictions_for_insn): Update.
22644 * asan.c (asan_clear_shadow): Update.
22645 * cfgbuild.c (compute_outgoing_frequencies): Update.
22646 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22647 (update_br_prob_note): Update.
22648 (rtl_verify_edges): Update.
22649 (purge_dead_edges): Update.
22650 (fixup_reorder_chain): Update.
22651 * emit-rtl.c (try_split): Update.
22652 * ifcvt.c (cond_exec_process_insns): Update.
22653 (cond_exec_process_if_block): Update.
22654 (dead_or_predicable): Update.
22655 * internal-fn.c (expand_addsub_overflow): Update.
22656 (expand_neg_overflow): Update.
22657 (expand_mul_overflow): Update.
22658 * loop-doloop.c (doloop_modify): Update.
22659 * loop-unroll.c (compare_and_jump_seq): Update.
22660 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22661 * predict.h: Update.
22662 * reorg.c (mostly_true_jump): Update.
22664 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22665 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22666 * config/arc/arc.c: (emit_unlikely_jump): Update.
22667 * config/arm/arm.c: (emit_unlikely_jump): Update.
22668 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22669 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22670 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22671 (ix86_print_operand): Update.
22672 (ix86_split_fp_branch): Update.
22673 (predict_jump): Update.
22674 * config/ia64/ia64.c (ia64_print_operand): Update.
22675 * config/mmix/mmix.c (mmix_print_operand): Update.
22676 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22677 (rs6000_expand_split_stack_prologue): Update.
22678 * config/rs6000/rs6000.c: Update.
22679 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22680 (s390_expand_vec_movstr): Update.
22681 (s390_expand_cs_tdsi): Update.
22682 (s390_expand_split_stack_prologue): Update.
22683 * config/sh/sh.c (sh_print_operand): Update.
22684 (expand_cbranchsi4): Update.
22685 (expand_cbranchdi4): Update.
22686 * config/sparc/sparc.c (output_v9branch): Update.
22687 * config/spu/spu.c (get_branch_target): Update.
22688 (ea_load_store_inline): Update.
22689 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22690 * config/tilepro/tilepro.c: Update.
22692 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22694 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22695 (gimplify_save_expr): Likewise.
22697 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22699 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22701 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22703 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22705 * timevar.def (TV_IPA_FNSUMMARY): Define.
22707 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22709 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22710 to back store errata sensitive sequence from being generated.
22711 (sqrtdf2_fix): Likewise.
22713 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22715 * tree-ssa-threadupdate.c (compute_path_counts,
22716 update_joiner_offpath_counts): Use profile_probability.
22718 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22721 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22723 * config/arm/arm-c.c (arm_cpu_builtins): Define
22724 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22726 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22728 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22729 array entries to represent __ieee128 versions of the
22730 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22731 scalar_extract_sig, and scalar_insert_exp built-in functions.
22732 (altivec_resolve_overloaded_builtin): Add special case handling
22733 for the __builtin_scalar_insert_exp function, as represented by
22734 the P9V_BUILTIN_VEC_VSIEDP constant.
22735 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22736 exponent support for __ieee128 argument.
22737 (VSESQP): Add scalar extract signature support for __ieee128
22739 (VSTDCNQP): Add scalar test negative support for __ieee128
22741 (VSIEQP): Add scalar insert exponent support for __int128 argument
22742 with __ieee128 result.
22743 (VSIEQPF): Add scalar insert exponent support for __ieee128
22744 argument with __ieee128 result.
22745 (VSTDCQP): Add scalar test data class support for __ieee128
22747 (VSTDCNQP): Add overload support for scalar test negative with
22748 __ieee128 argument.
22749 (VSTDCQP): Add overload support for scalar test data class
22750 __ieee128 argument.
22751 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22752 UNSPEC_VSX_SXSIGDP.
22753 (UNSPEC_VSX_SIEXPQP): New constant.
22754 (xsxexpqp): New insn for VSX scalar extract exponent quad
22756 (xsxsigqp): New insn for VSX scalar extract significand quad
22758 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22759 precision with floating point argument.
22760 (xststdcqp): New expand for VSX scalar test data class quad
22762 (xststdcnegqp): New expand for VSX scalar test negative quad
22764 (xststdcqp): New insn to match expansions for VSX scalar test data
22765 class quad precision and VSX scalar test negative quad precision.
22766 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22767 special case operand checking to enforce that second operand of
22768 VSX scalar test data class with quad precision argument is a 7-bit
22770 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22771 prototypes and descriptions of __ieee128 versions of
22772 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22773 scalar_test_data_class, and scalar_test_neg built-in functions.
22775 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22777 PR tree-optimization/81162
22778 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22779 replace a negate with an add.
22781 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22783 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22785 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22787 * config/arm/arm-c.c (arm_cpu_builtins): Define
22788 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22790 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22792 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22793 (armv8-r): Set ARM Cortex-R52 as default CPU.
22794 * config/arm/arm-tables.opt: Regenerate.
22795 * config/arm/arm-tune.md: Regenerate.
22796 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22798 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22799 extension for -mcpu=cortex-r52.
22801 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22803 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22804 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22805 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22806 (fp-armv8): Define it as FP_ARMv8 only.
22807 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22808 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22810 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22811 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22812 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22813 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22814 than TARGET_FPU_ARMV8.
22815 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22816 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22817 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22819 * config/arm/neon.md (neon_vrint): Likewise.
22820 (neon_vcvt): Likewise.
22821 (neon_<fmaxmin_op><mode>): Likewise.
22822 (<fmaxmin><mode>3): Likewise.
22823 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22824 * config/arm/predicates.md (arm_cond_move_operator): Check against
22825 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22827 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22829 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22830 to top of function.
22832 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22834 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22835 loop in comment with memset.
22837 2017-07-14 Martin Liska <mliska@suse.cz>
22839 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22840 * dwarf2out.c (is_java): Remove the function.
22841 (output_pubname): Remove usage of the function.
22842 (lower_bound_default): Remove usage of DW_LANG_Java.
22843 (gen_compile_unit_die): Likewise.
22844 * gcc.c: Remove compiler defaults for .java and .zip files.
22845 * gimple-expr.c (remove_suffix): Change as there's no longer
22846 extension than 4-letter one.
22847 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22848 (gimplify_save_expr): Likewise.
22849 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22850 as it's possible even for other languages than Java.
22851 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22852 * lto-opts.c (lto_write_options): Remove reference to Java.
22853 * opts.c (strip_off_ending): Update file extension handling.
22854 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22855 * tree-eh.c (lower_resx): Likewise.
22856 * tree.c (free_lang_data_in_type): Remove dead code.
22857 (find_decls_types_r): Likewise.
22858 (build_common_builtin_nodes): Remove Java from a comment.
22859 (verify_type): Remove dead code.
22860 * varasm.c (assemble_external): Remove Java from a comment.
22862 2017-07-14 Martin Liska <mliska@suse.cz>
22864 * opts.c (finish_options): Add quotes.
22865 (common_handle_option): Likewise.
22867 2017-07-14 Martin Liska <mliska@suse.cz>
22869 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22870 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22871 Remove N_SO_PASCAL.
22872 * dwarf2out.c (lower_bound_default): Do not handle
22874 (gen_compile_unit_die): Likewise.
22875 * gcc.c: Remove default extension binding for GNU Pascal.
22876 * stmt.c: Remove Pascal language from a comment.
22877 * xcoffout.c: Likewise.
22879 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22882 * diagnostic-show-locus.c (fixit_cmp): New function.
22883 (layout::layout): Sort m_fixit_hints.
22884 (column_range::column_range): Assert that the values are valid.
22885 (struct char_span): New struct.
22886 (correction::overwrite): New method.
22887 (struct source_line): New struct.
22888 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22889 calls in terms of classes source_line and char_span, and
22890 correction::overwrite.
22891 (selftest::test_overlapped_fixit_printing_2): New function.
22892 (selftest::diagnostic_show_locus_c_tests): Call it.
22894 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22897 early if there is no lhs.
22899 2017-07-13 Martin Liska <mliska@suse.cz>
22901 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22902 (gen_reference_type_die): Likewise.
22903 * stor-layout.c: Remove Pascal-related comment.
22905 2017-07-13 Martin Liska <mliska@suse.cz>
22907 * opts.c (finish_options): Add quotes to error messages.
22908 (parse_sanitizer_options): Likewise.
22910 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22912 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22914 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22916 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22918 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22920 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22922 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22924 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22927 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22928 provides the hardware capability bits, define the macro
22929 __BUILTIN_CPU_SUPPORTS__.
22930 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22931 if GLIBC does not provide the hardware capability bits. Add a
22932 gcc_unreachable call if the built-in cpu function is neither
22933 __builtin_cpu_is nor __builtin_cpu_supports.
22934 (rs6000_get_function_versions_dispatcher): Change the warning
22935 that an old GLIBC is used which does not export the capability
22936 bits to be an error.
22937 * doc/extend.texi (target_clones attribute): Document the
22938 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22939 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22940 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22941 the macros defined by GCC if the newer GLIBC is available.
22943 2017-07-12 Jeff Law <law@redhat.com>
22945 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22946 remaining includes slightly.
22947 * config/riscv/riscv-builtins.c: Include profile-count.h.
22949 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22952 * config/avr/avr.c (avr_set_current_function): In diagnostic
22953 messages: Quote keywords and (parts of) identifiers.
22954 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22957 2017-07-12 Carl Love <cel@us.ibm.com>
22959 * config/rs6000/rs6000-c.c: Add support for built-in functions
22960 vector bool char vec_revb (vector bool char);
22961 vector bool short vec_revb (vector short char);
22962 vector bool int vec_revb (vector bool int);
22963 vector bool long long vec_revb (vector bool long long);
22964 * doc/extend.texi: Update the built-in documentation file for the
22965 new built-in functions.
22967 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22969 * config/s390/s390.md: Remove movcc splitter.
22971 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22973 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22974 load/store on condition.
22976 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22979 * config/avr/avr.c (avr_encode_section_info)
22980 [progmem && !TREE_READONLY]: Error if progmem object needs
22983 2017-07-11 Michael Collison <michael.collison@arm.com>
22985 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22988 2017-07-11 Carl Love <cel@us.ibm.com>
22990 * config/rs6000/rs6000-c.c: Add support for builtins
22991 vector unsigned int vec_parity_lsbb (vector signed int);
22992 vector unsigned int vec_parity_lsbb (vector unsigned int);
22993 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22994 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22995 vector unsigned long long vec_parity_lsbb (vector signed long long);
22996 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22997 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22998 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22999 * doc/extend.texi: Update the built-in documentation file for the
23000 new built-in functions.
23002 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23004 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23005 (layout::m_primary_loc): New field.
23006 (layout::layout): Initialize new field. Move location filtering
23007 logic from here to...
23008 (layout::maybe_add_location_range): ...this new method. Add
23009 support for filtering to just the lines already specified by other
23011 (layout::will_show_line_p): New method.
23012 (gcc_rich_location::add_location_if_nearby): New method.
23013 (selftest::test_add_location_if_nearby): New test function.
23014 (selftest::diagnostic_show_locus_c_tests): Call it.
23015 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23018 2017-07-11 Tom de Vries <tom@codesourcery.com>
23020 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23021 (bb_first_real_insn): New function.
23022 (nvptx_single): Add extra initialization of broadcasted condition
23025 2017-07-11 Nathan Sidwell <nathan@acm.org>
23027 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23029 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23031 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23032 Binutils doc as TEXI will mess them up.
23033 * doc/invoke.texi (AVR Options): Same here.
23035 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23037 * config/sparc/sparc.opt (mfix-ut700): New option.
23038 (mfix-gr712rc): Likewise.
23039 (sparc_fix_b2bst): New variable.
23040 * doc/invoke.texi (SPARC options): Document them.
23041 (ARM options): Fix warnings.
23042 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23043 instructions to prevent sequences that can trigger the store-store
23044 errata for certain LEON3FT processors.
23045 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23046 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23047 * config/sparc/sparc.md (fix_b2bst): New attribute.
23048 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23050 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23053 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23055 (*rsqrtsf2_sse): Ditto.
23057 (div<mode>3): Macroize insn from divdf3 and divsf3
23058 using MODEF mode iterator.
23060 2017-07-10 Martin Sebor <msebor@redhat.com>
23062 PR tree-optimization/80397
23063 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23064 instead of testing for equality to INTEGER_TYPE.
23066 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23068 * config.gcc: Remove uclibc from arc target spec.
23070 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23072 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23074 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23077 * lto-wrapper.c (remove_option): New function.
23078 (merge_and_complain): Merge PIC/PIE options more realistically.
23080 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23082 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23086 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23087 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23088 * config.in: Regenerate.
23089 * configure: Regenerate.
23090 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23091 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23092 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23093 (TARGET_GASISR_PROLOGUES): ...target mask.
23094 * common/config/avr/avr-common.c
23095 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23096 Set -mgas-isr-prologues.
23097 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23098 INSERT_PASS_BEFORE for it.
23099 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23100 * config/avr/avr.c (avr_option_override)
23101 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23102 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23103 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23104 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23105 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23107 (make_avr_pass_pre_proep): New function.
23108 (emit_push_sfr) <treg>: Add argument to function and use it
23109 instead of TMP_REG.
23110 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23111 and set machine->gasisr.yes.
23112 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23113 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23114 __gcc_isr.n_pushed to .L__stack_usage.
23115 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23116 (avr_asm_final_postscan_insn): ...this new static function.
23117 * config/avr/avr.h (machine_function)
23118 <is_no_gccisr, use_L__stack_usage>: New fields.
23119 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23120 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23121 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23122 (gasisr, *gasisr): New expander and insn.
23123 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23124 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23125 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23127 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23129 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23132 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23134 Move jump-tables out of .text again.
23137 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23138 (ASM_OUTPUT_ADDR_VEC): New function.
23139 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23140 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23141 INSN_ADDRESSes as asm comment.
23142 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23143 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23144 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23145 * config/avr/avr.md (*tablejump): Adjust comment.
23146 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23147 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23149 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23150 (avr_output_addr_vec): New proto.
23151 (avr_log_t) <insn_addresses>: New field.
23153 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23156 * config/i386/i386.c (ix86_function_arg_advance): Set
23157 outgoing_args_on_stack to true if there are outgoing arguments
23159 (ix86_function_arg): Likewise.
23160 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23161 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23162 * config/i386/i386.h (machine_function): Add
23163 outgoing_args_on_stack.
23165 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23167 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23168 supporting pthreds.
23169 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23171 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23173 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23174 (REAL_H): Remove $(MACHMODE_H).
23175 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23177 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23178 $(MACHMODE_H) and double-int.h.
23179 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23181 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23184 2017-07-07 Andrew Pinski <apinski@cavium.com>
23186 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23187 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23189 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23191 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23192 Add warning if GCC was not configured to link against a GLIBC that
23193 exports the hardware capability bits.
23194 (make_resolver_func): Make resolver function private and not a
23195 COMDAT function. Create the name with clone_function_name instead
23196 of make_unique_name.
23199 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23200 correct operand in doing the split.
23202 2017-07-07 Carl Love <cel@us.ibm.com>
23204 * config/rs6000/rs6000-c: Add support for built-in function
23205 vector unsigned short vec_pack_to_short_fp32 (vector float,
23207 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23208 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23209 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23210 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23211 (convert_4f32_8i16): Add define_expand.
23212 * doc/extend.texi: Update the built-in documentation file for the
23213 new built-in function.
23215 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23217 * config/sparc/m8.md: New file.
23218 * config/sparc/sparc.md: Include m8.md.
23220 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23222 * config/sparc/sparc.opt: New option -mvis4b.
23223 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23224 (sparc_option_override): Handle VIS4B.
23225 (enum sparc_builtins): Define
23226 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23227 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23228 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23229 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23230 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23231 (check_constant_argument): New function.
23232 (sparc_vis_init_builtins): Define builtins
23233 __builtin_vis_dictunpack{8,16,32},
23234 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23235 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23236 __builtin_vis_fpcmpde{8,16,32}shl and
23237 __builtin_vis_fpcmpur{8,16,32}shl.
23238 (sparc_expand_builtin): Check that the constant operands to
23239 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23240 constant and in range.
23241 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23243 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23244 (SPARC_IMM5_P): Likewise.
23245 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23246 (enabled): Handle vis4b.
23247 (UNSPEC_DICTUNPACK): New unspec.
23248 (UNSPEC_FPCMPSHL): Likewise.
23249 (UNSPEC_FPUCMPSHL): Likewise.
23250 (UNSPEC_FPCMPDESHL): Likewise.
23251 (UNSPEC_FPCMPURSHL): Likewise.
23252 (cpu_feature): New CPU feature `vis4b'.
23253 (dictunpack{8,16,32}): New insns.
23254 (FPCSMODE): New mode iterator.
23255 (fpcscond): New code iterator.
23256 (fpcsucond): Likewise.
23257 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23258 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23259 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23260 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23261 * config/sparc/constraints.md: Define constraints `q' for unsigned
23262 2-bit integer constants and `t' for unsigned 5-bit integer
23264 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23266 (imm5_operand_dictunpack16): Likewise.
23267 (imm5_operand_dictunpack32): Likewise.
23268 (imm2_operand): Likewise.
23269 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23270 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23271 ditunpack* and fpcmp*shl builtins.
23273 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23275 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23276 * config.in: Add HAVE_AS_SPARC6 define.
23277 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23279 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23281 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23282 (CPP_CPU_SPEC): Handle m8.
23283 (ASM_CPU_SPEC): Likewise.
23284 * config/sparc/sparc-opts.h (enum processor_type): Add
23286 * config/sparc/sparc.c (m8_costs): New struct.
23287 (sparc_option_override): Handle TARGET_CPU_m8.
23288 (sparc32_initialize_trampoline): Likewise.
23289 (sparc64_initialize_trampoline): Likewise.
23290 (sparc_issue_rate): Likewise.
23291 (sparc_register_move_cost): Likewise.
23292 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23293 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23294 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23295 (CPP_CPU_SPEC): Handle M8.
23296 (ASM_CPU_SPEC): Likewise.
23297 (AS_M8_FLAG): Define.
23298 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23299 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23300 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23302 * configure: Regenerate.
23303 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23306 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23308 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23310 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23311 ("*movdi_insn_sp32"): Do not set v3pipe.
23312 ("*movsi_insn"): Likewise.
23313 ("*movdi_insn_sp64"): Likewise.
23314 ("*movsf_insn"): Likewise.
23315 ("*movdf_insn_sp32"): Likewise.
23316 ("*movdf_insn_sp64"): Likewise.
23317 ("*zero_extendsidi2_insn_sp64"): Likewise.
23318 ("*sign_extendsidi2_insn"): Likewise.
23319 ("*mov<VM32:mode>_insn"): Likewise.
23320 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23321 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23322 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23323 ("<vlop:code><VL:mode>3"): Likewise.
23324 ("*not_<vlop:code><VL:mode>3"): Likewise.
23325 ("*nand<VL:mode>_vis"): Likewise.
23326 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23327 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23328 ("one_cmpl<VL:mode>2"): Likewise.
23329 ("faligndata<VM64:mode>_vis"): Likewise.
23330 ("alignaddrsi_vis"): Likewise.
23331 ("alignaddrdi_vis"): Likweise.
23332 ("alignaddrlsi_vis"): Likewise.
23333 ("alignaddrldi_vis"): Likewise.
23334 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23335 ("bmaskdi_vis"): Likewise.
23336 ("bmasksi_vis"): Likewise.
23337 ("bshuffle<VM64:mode>_vis"): Likewise.
23338 ("cmask8<P:mode>_vis"): Likewise.
23339 ("cmask16<P:mode>_vis"): Likewise.
23340 ("cmask32<P:mode>_vis"): Likewise.
23341 ("pdistn<P:mode>_vis"): Likewise.
23342 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23344 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23346 * config/sparc/sparc.md ("subtype"): New insn attribute.
23347 ("*wrgsr_sp64"): Set insn subtype.
23348 ("*rdgsr_sp64"): Likewise.
23349 ("alignaddrsi_vis"): Likewise.
23350 ("alignaddrdi_vis"): Likewise.
23351 ("alignaddrlsi_vis"): Likewise.
23352 ("alignaddrldi_vis"): Likewise.
23353 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23354 ("fexpand_vis"): Likewise.
23355 ("fpmerge_vis"): Likewise.
23356 ("faligndata<VM64:mode>_vis"): Likewise.
23357 ("bshuffle<VM64:mode>_vis"): Likewise.
23358 ("cmask8<P:mode>_vis"): Likewise.
23359 ("cmask16<P:mode>_vis"): Likewise.
23360 ("cmask32<P:mode>_vis"): Likewise.
23361 ("fchksm16_vis"): Likewise.
23362 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23363 ("fmean16_vis"): Likewise.
23364 ("fp<plusminus_insn>64_vis"): Likewise.
23365 ("<plusminus_insn>v8qi3"): Likewise.
23366 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23367 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23368 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23369 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23370 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23371 ("*movqi_insn"): Likewise.
23372 ("*movhi_insn"): Likewise.
23373 ("*movsi_insn"): Likewise.
23374 ("movsi_pic_gotdata_op"): Likewise.
23375 ("*movdi_insn_sp32"): Likewise.
23376 ("*movdi_insn_sp64"): Likewise.
23377 ("movdi_pic_gotdata_op"): Likewise.
23378 ("*movsf_insn"): Likewise.
23379 ("*movdf_insn_sp32"): Likewise.
23380 ("*movdf_insn_sp64"): Likewise.
23381 ("*zero_extendhisi2_insn"): Likewise.
23382 ("*zero_extendqihi2_insn"): Likewise.
23383 ("*zero_extendqisi2_insn"): Likewise.
23384 ("*zero_extendqidi2_insn"): Likewise.
23385 ("*zero_extendhidi2_insn"): Likewise.
23386 ("*zero_extendsidi2_insn_sp64"): Likewise.
23387 ("ldfsr"): Likewise.
23388 ("prefetch_64"): Likewise.
23389 ("prefetch_32"): Likewise.
23390 ("tie_ld32"): Likewise.
23391 ("tie_ld64"): Likewise.
23392 ("*tldo_ldub_sp32"): Likewise.
23393 ("*tldo_ldub1_sp32"): Likewise.
23394 ("*tldo_ldub2_sp32"): Likewise.
23395 ("*tldo_ldub_sp64"): Likewise.
23396 ("*tldo_ldub1_sp64"): Likewise.
23397 ("*tldo_ldub2_sp64"): Likewise.
23398 ("*tldo_ldub3_sp64"): Likewise.
23399 ("*tldo_lduh_sp32"): Likewise.
23400 ("*tldo_lduh1_sp32"): Likewise.
23401 ("*tldo_lduh_sp64"): Likewise.
23402 ("*tldo_lduh1_sp64"): Likewise.
23403 ("*tldo_lduh2_sp64"): Likewise.
23404 ("*tldo_lduw_sp32"): Likewise.
23405 ("*tldo_lduw_sp64"): Likewise.
23406 ("*tldo_lduw1_sp64"): Likewise.
23407 ("*tldo_ldx_sp64"): Likewise.
23408 ("*mov<VM32:mode>_insn"): Likewise.
23409 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23410 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23412 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23414 * config/sparc/sparc.md ("type"): New insn type viscmp.
23415 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23417 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23418 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23419 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23420 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23422 ("n7_vis_logical_11cycle"): Likewise.
23423 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23424 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23425 * config/sparc/niagara.md ("niag_vis"): Likewise.
23426 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23427 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23429 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23431 * config/sparc/sparc.md: New instruction type `bmask'.
23432 (bmaskdi_vis): Use the `bmask' type.
23433 (bmasksi_vis): Likewise.
23434 * config/sparc/ultra3.md (us3_array): Likewise.
23435 * config/sparc/niagara7.md (n7_array): Likewise.
23436 * config/sparc/niagara4.md (n4_array): Likewise.
23437 * config/sparc/niagara2.md (niag2_vis): Likewise.
23438 (niag3_vis): Likewise.
23439 * config/sparc/niagara.md (niag_vis): Likewise.
23441 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23443 * ipa-comdats.c: Remove optimize check from gate.
23444 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23445 for functions not optimized.
23446 (ipa_fn_summary_read): Skip optimize check.
23447 (ipa_fn_summary_write): Likewise.
23448 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23450 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23452 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23453 (check_callers): Give up on caller which is not optimized.
23454 (inline_small_functions): Likewise.
23455 (ipa_inline): Do not give up when not optimizing.
23456 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23457 away unoptimizes cdtors.
23458 (whole_program_function_and_variable_visibility): Do
23459 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23460 * ipa.c (process_references): Do not check optimize.
23461 (symbol_table::remove_unreachable_nodes): Update optimize check.
23462 (set_writeonly_bit): Update optimize check.
23463 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23464 (pass_ipa_single_use::gate): Remove.
23466 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23468 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23469 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23470 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23471 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23472 permute_load, permute_store, adjust_extract, adjust_splat,
23473 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23474 replace_swap_with_copy, dump_swap_insn_table,
23475 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23476 recombine_lvx_pattern, recombine_stvx_pattern,
23477 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23478 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23479 to file rs6000-p8swap.c.
23480 * config/rs6000/rs6000-p8swap.c: New file.
23481 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23482 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23483 and rs6000*-*-* targets.
23485 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23487 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23489 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23491 * lto-wrapper.c (merge_and_complain): Do not merge
23492 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23493 fsigned_zeros, ftrapping_math, fwrapv.
23494 (append_compiler_options): Do not track these options.
23495 (append_linker_options): Likewie
23497 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23499 * cgraphunit.c (cgraph_node::finalize_function): When
23500 !flag_toplevel_reorde set no_reorder flag.
23501 (varpool_node::finalize_decl): Likewise.
23502 (symbol_table::compile): Drop no toplevel reorder path.
23504 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23506 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23507 edges; zero probability is not better than uninitialized.
23509 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23511 * asan.h (asan_sanitize_allocas_p): Declare.
23512 * asan.c (asan_sanitize_allocas_p): New function.
23513 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23514 (handle_builtin_alloca): Likewise.
23515 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23516 if !asan_sanitize_allocas_p.
23517 * params.def (asan-instrument-allocas): Add new option.
23518 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23519 * opts.c (common_handle_option): Disable allocas sanitization for
23522 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23524 * asan.c: Include gimple-fold.h.
23525 (get_last_alloca_addr): New function.
23526 (handle_builtin_stackrestore): Likewise.
23527 (handle_builtin_alloca): Likewise.
23528 (asan_emit_allocas_unpoison): Likewise.
23529 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23530 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23531 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23532 (instrument_builtin_call): Pass gimple iterator to
23533 get_mem_refs_of_builtin_call.
23534 (last_alloca_addr): New global.
23535 * asan.h (asan_emit_allocas_unpoison): Declare.
23536 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23537 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23538 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23539 if function calls alloca.
23540 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23541 * gimple-fold.h (replace_call_with_value): Declare.
23542 * internal-fn.c: Include asan.h.
23543 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23544 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23546 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23548 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23549 (C_SELFTEST_FLAGS): New.
23550 (CPP_SELFTEST_FLAGS): New.
23551 (SELFTEST_DEPS): New, from deps of s-selftest.
23552 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23553 (CPP_SELFTEST_DEPS): New.
23554 (selftest): Add dependency on s-selftest-c++.
23555 (s-selftest): Rename to...
23556 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23557 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23558 than SELFTEST_FLAGS.
23559 (selftest-gdb): Rename to...
23560 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23562 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23563 (selftest-valgrind): Rename to...
23564 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23566 (selftest-valgrind): Reintroduce as an alias for
23567 selftest-c-valgrind.
23568 (s-selftest-c++): New.
23569 (selftest-c++-gdb): New.
23570 (selftest-c++-valgrind): New.
23572 2017-07-06 Olivier Hainque <hainque@adacore.com>
23574 * gcc.c (process_command): When deciding if undefined variables
23575 should be ignored when processing specs, accept "gcc -v" as well.
23577 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23579 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23580 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23582 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23584 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23585 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23586 * config/arm/arm-tables.opt: Regenerate.
23587 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23589 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23591 2017-07-06 Carl Love <cel@us.ibm.com>
23593 * ChangeLog: Clean up from mid air collision
23595 2017-07-06 Carl Love <cel@us.ibm.com>
23597 * config/rs6000/rs6000-c.c: Add support for built-in functions
23598 vector signed int vec_subc (vector signed int, vector signed int);
23599 vector signed __int128 vec_subc (vector signed __int128,
23600 vector signed __int128);
23601 vector unsigned __int128 vec_subc (vector unsigned __int128,
23602 vector unsigned __int128);
23603 vector signed int vec_sube (vector signed int, vector signed int,
23604 vector signed int);
23605 vector unsigned int vec_sube (vector unsigned int,
23606 vector unsigned int,
23607 vector unsigned int);
23608 vector signed __int128 vec_sube (vector signed __int128,
23609 vector signed __int128,
23610 vector signed__int128);
23611 vector unsigned __int128 vec_sube (vector unsigned __int128,
23612 vector unsigned __int128,
23613 vector unsigned __int128);
23614 vector signed int vec_subec (vector signed int, vector signed int,
23615 vector signed int);
23616 vector unsigned int vec_subec (vector unsigned int,
23617 vector unsigned int,
23618 vector unsigned int);
23619 vector signed __int128 vec_subec (vector signed __int128,
23620 vector signed __int128,
23621 vector signed__int128);
23622 vector unsigned __int128 vec_subec (vector unsigned __int128,
23623 vector unsigned __int128,
23624 vector unsigned __int128);
23625 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23626 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23627 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23628 BU_ALTIVEC_OVERLOAD_X definitions.
23629 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23630 * doc/extend.texi: Update the built-in documentation file for the new
23631 built-in functions.
23633 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23636 * diagnostic-show-locus.c (layout::layout): Use start and finish
23637 spelling location for the start and finish of each range.
23638 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23639 Add unused aspect param.
23640 * input.c (expand_location_1): Add "aspect" param, and use it
23641 to access the correct part of the location.
23642 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23644 (expand_location_to_spelling_point): Likewise.
23645 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23646 param, and pass it to expand_location_1.
23648 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23650 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23651 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23652 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23653 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23654 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23655 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23656 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23657 _mm_maskz_getmant_ss): New intrinsics.
23658 (__builtin_ia32_getexpss128_mask): Changed to ...
23659 __builtin_ia32_getexpss128_round ... this.
23660 (__builtin_ia32_getexpsd128_mask): Changed to ...
23661 __builtin_ia32_getexpsd128_round ... this.
23662 * config/i386/i386-builtin-types.def
23663 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23664 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23665 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23666 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23667 __builtin_ia32_getmantss_mask_round): New builtins.
23668 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23669 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23670 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23671 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23672 * config/i386/sse.md
23673 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23674 avx512f_sgetexp<mode><mask_scalar_name>
23675 <round_saeonly_scalar_name> ... this.
23676 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23677 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23678 vgetexp<ssescalarmodesuffix>
23679 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23680 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23681 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23682 avx512f_vgetmant<mode><mask_scalar_name>
23683 <round_saeonly_scalar_name> ... this.
23684 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23685 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23686 vgetmant<ssescalarmodesuffix>
23687 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23688 %0<mask_scalar_operand4>, %1, %2
23689 <round_saeonly_scalar_mask_op4>, %3} ... this.
23690 * config/i386/subst.md (mask_scalar_operand4,
23691 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23692 round_saeonly_scalar_nimm_predicate): New subst attributes.
23694 2017-07-06 Julia Koval <julia.koval@intel.com>
23696 * config/i386/i386.c (ix86_erase_embedded_rounding):
23697 Remove code for old rounding pattern.
23699 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23701 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23703 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23705 * doc/sourcebuild.texi (Test Directives, Variants of
23706 dg-require-support): Add documentation for dg-require-stack-check.
23708 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23710 * config/i386/subst.md (mask_scalar, round_scalar,
23711 round_saeonly_scalar): New meta-templates.
23712 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23713 round_scalar_mask_operand3, round_scalar_mask_op3,
23714 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23715 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23716 round_saeonly_scalar_constraint,
23717 round_saeonly_scalar_prefix): New subst attribute.
23718 * config/i386/sse.md
23719 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23720 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23721 <round_scalar_name> ... this.
23722 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23723 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23724 <round_scalar_name> ... this.
23725 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23726 <sse>_vm<code><mode>3<mask_scalar_name>
23727 <round_saeonly_scalar_name> ... this.
23728 (v<plusminus_mnemonic><ssescalarmodesuffix>
23729 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23730 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23731 v<plusminus_mnemonic><ssescalarmodesuffix>
23732 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23733 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23734 (v<multdiv_mnemonic><ssescalarmodesuffix>
23735 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23736 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23737 v<multdiv_mnemonic><ssescalarmodesuffix>
23738 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23739 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23740 (v<maxmin_float><ssescalarmodesuffix>
23741 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23742 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23743 v<maxmin_float><ssescalarmodesuffix>
23744 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23745 %0<mask_scalar_operand3>, %1, %<iptr>2
23746 <round_saeonly_scalar_mask_op3>} ... this.
23748 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23750 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23751 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23753 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23754 Alan Hayward <alan.hayward@arm.com>
23755 David Sherwood <david.sherwood@arm.com>
23757 * combine.c (simplify_if_then_else): Remove "enum" before
23759 * compare-elim.c (can_eliminate_compare): Likewise.
23760 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23762 (aarch64_lookup_simd_builtin_type): Likewise.
23763 (aarch64_simd_builtin_type): Likewise.
23764 (aarch64_init_simd_builtin_types): Likewise.
23765 (aarch64_simd_expand_args): Likewise.
23766 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23768 (aarch64_reverse_mask): Likewise.
23769 (aarch64_simd_emit_reg_reg_move): Likewise.
23770 (aarch64_gen_adjusted_ldpstp): Likewise.
23771 (aarch64_ccmp_mode_to_code): Likewise.
23772 (aarch64_operands_ok_for_ldpstp): Likewise.
23773 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23774 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23776 (aarch64_min_divisions_for_recip_mul): Likewise.
23777 (aarch64_reassociation_width): Likewise.
23778 (aarch64_get_condition_code_1): Likewise.
23779 (aarch64_simd_emit_reg_reg_move): Likewise.
23780 (aarch64_simd_attr_length_rglist): Likewise.
23781 (aarch64_reverse_mask): Likewise.
23782 (aarch64_operands_ok_for_ldpstp): Likewise.
23783 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23784 (aarch64_gen_adjusted_ldpstp): Likewise.
23785 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23787 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23788 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23789 (arm_lookup_simd_builtin_type): Likewise.
23790 (arm_simd_builtin_type): Likewise.
23791 (arm_init_simd_builtin_types): Likewise.
23792 (arm_expand_builtin_args): Likewise.
23793 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23794 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23795 (ft32_setup_incoming_varargs): Likewise.
23796 (ft32_function_arg): Likewise.
23797 (ft32_function_arg_advance): Likewise.
23798 (ft32_pass_by_reference): Likewise.
23799 (ft32_arg_partial_bytes): Likewise.
23800 (ft32_valid_pointer_mode): Likewise.
23801 (ft32_addr_space_pointer_mode): Likewise.
23802 (ft32_addr_space_legitimate_address_p): Likewise.
23803 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23805 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23806 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23807 (iamcu_alignment): Likewise.
23808 (canonicalize_vector_int_perm): Likewise.
23809 (ix86_noce_conversion_profitable_p): Likewise.
23810 (ix86_mpx_bound_mode): Likewise.
23811 (ix86_operands_ok_for_move_multiple): Likewise.
23812 * config/microblaze/microblaze-protos.h
23813 (microblaze_expand_conditional_branch_reg): Likewise.
23814 * config/microblaze/microblaze.c
23815 (microblaze_expand_conditional_branch_reg): Likewise.
23816 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23818 (rs6000_reassociation_width): Likewise.
23819 (rs6000_invalid_binary_op): Likewise.
23820 (fusion_p9_p): Likewise.
23821 (emit_fusion_p9_load): Likewise.
23822 (emit_fusion_p9_store): Likewise.
23823 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23825 (riscv_hard_regno_mode_ok_p): Likewise.
23826 (riscv_address_insns): Likewise.
23827 (riscv_split_symbol): Likewise.
23828 (riscv_legitimize_move): Likewise.
23829 (riscv_function_value): Likewise.
23830 (riscv_hard_regno_nregs): Likewise.
23831 (riscv_expand_builtin): Likewise.
23832 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23833 (riscv_build_integer): Likewise.
23834 (riscv_split_integer): Likewise.
23835 (riscv_legitimate_constant_p): Likewise.
23836 (riscv_cannot_force_const_mem): Likewise.
23837 (riscv_regno_mode_ok_for_base_p): Likewise.
23838 (riscv_valid_base_register_p): Likewise.
23839 (riscv_valid_offset_p): Likewise.
23840 (riscv_valid_lo_sum_p): Likewise.
23841 (riscv_classify_address): Likewise.
23842 (riscv_legitimate_address_p): Likewise.
23843 (riscv_address_insns): Likewise.
23844 (riscv_load_store_insns): Likewise.
23845 (riscv_force_binary): Likewise.
23846 (riscv_split_symbol): Likewise.
23847 (riscv_force_address): Likewise.
23848 (riscv_legitimize_address): Likewise.
23849 (riscv_move_integer): Likewise.
23850 (riscv_legitimize_const_move): Likewise.
23851 (riscv_legitimize_move): Likewise.
23852 (riscv_address_cost): Likewise.
23853 (riscv_subword): Likewise.
23854 (riscv_output_move): Likewise.
23855 (riscv_canonicalize_int_order_test): Likewise.
23856 (riscv_emit_int_order_test): Likewise.
23857 (riscv_function_arg_boundary): Likewise.
23858 (riscv_pass_mode_in_fpr_p): Likewise.
23859 (riscv_pass_fpr_single): Likewise.
23860 (riscv_pass_fpr_pair): Likewise.
23861 (riscv_get_arg_info): Likewise.
23862 (riscv_function_arg): Likewise.
23863 (riscv_function_arg_advance): Likewise.
23864 (riscv_arg_partial_bytes): Likewise.
23865 (riscv_function_value): Likewise.
23866 (riscv_pass_by_reference): Likewise.
23867 (riscv_setup_incoming_varargs): Likewise.
23868 (riscv_print_operand): Likewise.
23869 (riscv_elf_select_rtx_section): Likewise.
23870 (riscv_save_restore_reg): Likewise.
23871 (riscv_for_each_saved_reg): Likewise.
23872 (riscv_register_move_cost): Likewise.
23873 (riscv_hard_regno_mode_ok_p): Likewise.
23874 (riscv_hard_regno_nregs): Likewise.
23875 (riscv_class_max_nregs): Likewise.
23876 (riscv_memory_move_cost): Likewise.
23877 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23878 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23879 (rl78_addr_space_address_mode): Likewise.
23880 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23882 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23883 (rs6000_reassociation_width): Likewise.
23884 (rs6000_invalid_binary_op): Likewise.
23885 (fusion_p9_p): Likewise.
23886 (emit_fusion_p9_load): Likewise.
23887 (emit_fusion_p9_store): Likewise.
23888 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23889 (ok_for_simple_move_operands): Likewise.
23890 (ok_for_simple_move_strict_operands): Likewise.
23891 (ok_for_simple_arith_logic_operands): Likewise.
23892 (visium_legitimize_reload_address): Likewise.
23893 (visium_select_cc_mode): Likewise.
23894 (output_cbranch): Likewise.
23895 (visium_split_double_move): Likewise.
23896 (visium_expand_copysign): Likewise.
23897 (visium_expand_int_cstore): Likewise.
23898 (visium_expand_fp_cstore): Likewise.
23899 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23900 (visium_function_arg): Likewise.
23901 (visium_function_arg_advance): Likewise.
23902 (visium_libcall_value): Likewise.
23903 (visium_setup_incoming_varargs): Likewise.
23904 (visium_legitimate_constant_p): Likewise.
23905 (visium_legitimate_address_p): Likewise.
23906 (visium_legitimize_address): Likewise.
23907 (visium_secondary_reload): Likewise.
23908 (visium_register_move_cost): Likewise.
23909 (visium_memory_move_cost): Likewise.
23910 (prepare_move_operands): Likewise.
23911 (ok_for_simple_move_operands): Likewise.
23912 (ok_for_simple_move_strict_operands): Likewise.
23913 (ok_for_simple_arith_logic_operands): Likewise.
23914 (visium_function_value_1): Likewise.
23915 (rtx_ok_for_offset_p): Likewise.
23916 (visium_legitimize_reload_address): Likewise.
23917 (visium_split_double_move): Likewise.
23918 (visium_expand_copysign): Likewise.
23919 (visium_expand_int_cstore): Likewise.
23920 (visium_expand_fp_cstore): Likewise.
23921 (visium_split_cstore): Likewise.
23922 (visium_select_cc_mode): Likewise.
23923 (visium_split_cbranch): Likewise.
23924 (output_cbranch): Likewise.
23925 (visium_print_operand_address): Likewise.
23926 * expmed.c (flip_storage_order): Likewise.
23927 * expmed.h (emit_cstore): Likewise.
23928 (flip_storage_order): Likewise.
23929 * genrecog.c (validate_pattern): Likewise.
23930 * hsa-gen.c (gen_hsa_addr): Likewise.
23931 * internal-fn.c (expand_arith_overflow): Likewise.
23932 * ira-color.c (allocno_copy_cost_saving): Likewise.
23933 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23934 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23935 (process_invariant_for_inheritance): Likewise.
23936 * lra-eliminations.c (move_plus_up): Likewise.
23937 * omp-low.c (lower_oacc_reductions): Likewise.
23938 * simplify-rtx.c (simplify_subreg): Likewise.
23939 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23940 (TARGET_CHKP_BOUND_MODE): Likewise..
23941 * targhooks.c (default_chkp_bound_mode): Likewise.
23942 (default_setup_incoming_vararg_bounds): Likewise.
23943 * targhooks.h (default_chkp_bound_mode): Likewise.
23944 (default_setup_incoming_vararg_bounds): Likewise.
23945 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23946 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23947 (have_whole_vector_shift): Likewise.
23948 * tree-vect-stmts.c (vectorizable_load): Likewise.
23949 * doc/tm.texi: Regenerate.
23951 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23953 Graceful degrade if Binutils PR21472 is not available.
23956 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23957 .rodata in flash test fails.
23958 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23959 * confgure: Regenerate.
23960 * config.in: Regenerate.
23961 * config/avr/avr.c (avr_asm_named_section)
23962 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23963 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23964 (avr_asm_init_sections): Same.
23966 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23968 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23969 (fma<VH:mode>4_intrinsic): Likewise.
23970 (*fmsub<VCVTF:mode>4): Likewise.
23971 (*fmsub<VH:mode>4_intrinsic): Likewise.
23973 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23976 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23977 Don't depend on "optimize > 0".
23978 (out_movhi_r_mr, out_movqi_mr_r): Same.
23979 (out_movhi_mr_r, out_movqi_r_mr): Same.
23980 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23981 io_address_operand on "optimize > 0".
23983 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23985 * tree-loop-distribution.c: Add general explanantion on the pass.
23986 (generate_loops_for_partition): Mark distributed loop.
23987 (pg_add_dependence_edges): New parameter. Handle alias data
23988 dependence specially and record it in the parameter if asked.
23989 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23990 (init_partition_graph_vertices, add_partition_graph_edge): New.
23991 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23992 (free_partition_graph_vdata, build_partition_graph): New.
23993 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23994 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23995 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23996 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23997 (version_for_distribution_p, finalize_partitions): New.
23998 (distribute_loop): Handle alias data dependence specially. Factor
23999 out loop fusion code as functions and call these functions.
24001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24003 * tree-loop-distribution.c (classify_partition): New parameter and
24004 better handle reduction statement.
24005 (rdg_build_partitions): Revise comment.
24006 (distribute_loop): Compute statements in all partitions and pass it
24007 to classify_partition.
24009 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24011 * tree-loop-distribution.c (enum partition_type): New.
24012 (struct partition): New field type.
24013 (partition_merge_into): Add parameter. Update partition type.
24014 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24015 (build_rdg_partition_for_vertex): Compute partition type.
24016 (rdg_build_partitions): Dump partition type.
24017 (distribute_loop): Update calls to partition_merge_into.
24019 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24021 * tree-loop-distribution.c (struct ddr_hasher): New.
24022 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24024 (classify_partition): Call get_data_dependence.
24025 (pg_add_dependence_edges): Ditto.
24026 (distribute_loop): Release data dependence hash table.
24028 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24030 * tree-loop-distribution.c (ref_base_address): Delete.
24031 (similar_memory_accesses): Rename ...
24032 (share_memory_accesses): ... to this. Check if partitions access
24033 the same memory reference.
24034 (distribute_loop): Call share_memory_accesses.
24036 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24038 * tree-loop-distribution.c (struct partition): New field recording
24039 its data reference.
24040 (partition_alloc, partition_free): Init and release data refs.
24041 (partition_merge_into): Merge data refs.
24042 (build_rdg_partition_for_vertex): Collect data refs for partition.
24043 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24045 (distribute_loop): Remve data refs from vertice data of partition
24048 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24050 * tree-loop-distribution.c (params.h): Include header file.
24051 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24052 (datarefs_vec): New global var.
24053 (create_rdg_vertices): Use datarefs_vec directly.
24054 (free_rdg): Don't free data references.
24055 (build_rdg): Update use. Don't free data references.
24056 (distribute_loop): Compute global variable for data references.
24057 Bail out if there are too many data references.
24059 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24061 * tree-loop-distribution.c (loop_nest): New global var.
24062 (build_rdg): Use loop directly, rather than loop nest.
24063 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24065 (distribute_loop): Compute global variable loop nest. Update use.
24067 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24069 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24070 (partition_merge_into): New parameter. Dump reason for fusion.
24071 (distribute_loop): Update use of partition_merge_into.
24073 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24075 * tree-loop-distribution.c (bb_top_order_index): New.
24076 (bb_top_order_index_size, bb_top_order_cmp): New.
24077 (stmts_from_loop): Use topological order.
24078 (pass_loop_distribution::execute): Compute and release topological
24079 order for basic blocks.
24081 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24083 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24086 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24088 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24089 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24090 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24091 * internal-fn.def (LOOP_DIST_ALIAS): New.
24092 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24093 (fold_loop_internal_call): ... this.
24094 (vect_loop_dist_alias_call): New function.
24095 (set_uid_loop_bbs): Call fold_loop_internal_call.
24096 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24099 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24102 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24103 Require dead FLAGS_REG at the beginning of a peephole.
24105 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24108 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24109 arguments in the call to __builtin_ia32_sbb_u32.
24110 (_subborrow_u64): Swap _X and _Y arguments in the call to
24111 __builtin_ia32_sbb_u64.
24113 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24116 * tree-vrp.c (compare_assert_loc): Turn into a function template
24117 with stable template parameter. Only test if a->e is NULL,
24118 !a->e == !b->e has been verified already. Use e == NULL or
24119 e != NULL instead of e or ! e tests. If stable is true, don't use
24120 iterative_hash_expr, on the other side allow a or b or both NULL
24121 and sort the NULLs last.
24122 (process_assert_insertions): Sort using compare_assert_loc<false>
24123 instead of compare_assert_loc, later sort using
24124 compare_assert_loc<true> before calling process_assert_insertions_for
24125 in a loop. Use break instead of continue once seen NULL pointer.
24127 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24129 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24130 Cortex-R7 and Cortex-R8 processors.
24132 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24134 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24135 uninitialized while src is not.
24137 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24139 * common/config/arm/arm-common.c: Adjust include path for
24141 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24142 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24143 (arm-cpu-data.h): Likewise.
24144 (arm-cpu-cdata.h): Likewise.
24145 * config/arm/arm-cpu.h: Delete.
24146 * config/arm/arm-cpu-cdata.h: Delete.
24147 * config/arm/arm-cpu-data.h: Delete.
24149 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24151 * config/arm/arm-cpus.in (cortex-a55): New.
24152 (cortex-a75): Likewise.
24153 (cortex-a75.cortex-a55): Likewise.
24154 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24156 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24157 * config/arm/arm-cpu-cdata.h: Regenerate.
24158 * config/arm/arm-cpu-data.h: Regenerate.
24159 * config/arm/arm-cpu.h: Regenerate.
24160 * config/arm/arm-tables.opt: Regenerate.
24161 * config/arm/arm-tune.md: Regenerate.
24163 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24165 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24167 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24169 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24172 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24174 PR tree-optimization/81292
24175 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24176 full_string_p, also call adjust_related_strinfos if the adjustment
24177 is simple, otherwise invalidate related strinfos.
24179 2017-07-04 Martin Liska <mliska@suse.cz>
24182 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24183 newly created variable as DECL_IGNORED_P.
24185 2017-07-04 Martin Liska <mliska@suse.cz>
24188 * ipa-inline.c (inline_small_functions):
24189 Use xstrdup_for_dump.
24191 2017-07-04 Tom de Vries <tom@codesourcery.com>
24193 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24195 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24198 * config/darwin.c (darwin_function_switched_text_sections):
24201 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24203 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24205 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24207 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24209 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24211 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24212 min_profitable_iters, and th as inclusive lower bounds.
24213 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24214 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24215 for min_profitable_iters and min_profitable_estimate.
24216 (vect_transform_loop): Treat th as an inclusive lower bound.
24217 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24219 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24222 * config/darwin.c (darwin_function_switched_text_sections):
24223 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24224 in two pieces, and suppress the use of buf.
24226 2017-07-03 Nathan Sidwell <nathan@acm.org>
24228 * hash-table.h (hash_table_mod1): Fix indentation.
24230 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24232 PR middle-end/81290
24233 * predict.c (force_edge_cold): Be more careful about propagation
24235 * profile-count.h (profile_probability::guessed,
24236 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24238 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24240 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24242 * doc/invoke.texi (rcpc architecture extension): Document it.
24244 2017-07-03 Richard Biener <rguenther@suse.de>
24246 PR tree-optimization/60510
24247 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24248 the scalar reduction PHI and use it.
24249 (vectorizable_reduction): Properly guard the single_defuse_cycle
24250 path for non-SLP reduction chains where we cannot use it.
24251 Rework reduc_def/index and vector type deduction. Rework
24252 vector operand gathering during reduction op code-gen.
24253 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24254 chains dissolve the chain and leave it to non-SLP reduction
24257 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24259 * tree-data-ref.h (dr_alignment): Declare.
24260 * tree-data-ref.c (dr_alignment): New function.
24261 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24262 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24264 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24266 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24268 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24269 and base_misalignment fields.
24270 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24271 * tree-data-ref.c: Include builtins.h.
24272 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24273 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24274 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24275 * tree-vect-data-refs.c: Include tree-cfg.h.
24276 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24277 fields instead of calculating an alignment here.
24278 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24279 innermost_loop_behavior fields.
24281 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24283 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24285 (DR_STEP_ALIGNMENT): New macro.
24286 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24287 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24288 (create_data_ref): Print it.
24289 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24290 to tell whether the step preserves vector (mis)alignment.
24291 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24292 Move the check for an integer step and generalise to all INTEGER_CST.
24293 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24294 Print the outer step alignment.
24296 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24298 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24299 with offset_alignment.
24300 (DR_ALIGNED_TO): Delete.
24301 (DR_OFFSET_ALIGNMENT): New macro.
24302 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24303 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24304 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24305 (create_data_ref): Likewise.
24306 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24307 (vect_analyze_data_refs): Likewise.
24308 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24309 creating dummy innermost behavior.
24311 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24313 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24314 with a "innermost_loop_behavior *" and refeence tree.
24315 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24316 (create_data_ref): Update call accordingly.
24317 * tree-predcom.c (find_looparound_phi): Likewise.
24319 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24321 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24322 fields with dr_wrt_vec_loop.
24323 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24324 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24325 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24326 (vect_dr_behavior): New function.
24327 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24328 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24329 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24330 track whether the step preserves the misalignment.
24331 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24332 Use vect_dr_behavior.
24333 (vect_setup_realignment): Update call accordingly.
24334 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24335 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24336 call to vect_create_addr_base_for_vector_ref.
24337 (vect_create_cond_for_align_checks): Likewise.
24338 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24339 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24340 (vect_recog_mask_conversion_pattern): Likewise.
24341 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24342 (new_stmt_vec_info): Remove redundant zeroing.
24344 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24346 * common/config/arm/arm-common.c (arm_be8_option): New function.
24347 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24348 (ISA_ARMv6): Add isa_bit_be8.
24349 * config/arm/arm.h (arm_be8_option): Add prototype.
24350 (BE8_SPEC_FUNCTION): New define.
24351 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24352 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24353 (mlittle-endian): Similarly.
24354 (mbe8, mbe32): New options.
24355 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24356 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24358 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24360 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24362 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24364 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24365 (cleanup_tree_cfg_bb): Use it.
24366 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24368 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24370 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24373 * loop-doloop.c (add_test): Update profile.
24375 2017-07-03 Martin Liska <mliska@suse.cz>
24378 * sanopt.c (rewrite_usage_of_param): New function.
24379 (sanitize_rewrite_addressable_params): Likewise.
24380 (pass_sanopt::execute): Call rewrite_usage_of_param.
24382 2017-07-03 Richard Biener <rguenther@suse.de>
24384 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24385 back to using VIEW_CONVERT_EXPR.
24387 2017-07-03 Martin Liska <mliska@suse.cz>
24390 * doc/extend.texi: Document when a resolver function is
24391 generated for target_clones.
24393 2017-07-03 Martin Liska <mliska@suse.cz>
24395 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24396 and shadow memory of auto variables which are subject of
24397 use-after-scope sanitization.
24398 (asan_expand_mark_ifn): Add do set only when is_poison.
24400 2016-07-03 Richard Biener <rguenther@suse.de>
24402 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24404 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24405 (vectorizable_reduction): Perform reduction PHI creation when
24406 visiting a reduction PHI and adjust and simplify code generation
24407 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24408 (vect_transform_loop): Visit reduction PHIs.
24409 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24410 defs into the SLP tree.
24411 (vect_build_slp_tree): Reduction defs terminate the recursion.
24412 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24414 (vect_get_vec_defs_for_stmt_copy): Export.
24415 (vect_get_vec_defs): Likewise.
24416 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24418 (vect_get_vec_defs_for_stmt_copy): Declare.
24419 (vect_get_vec_defs): Likewise.
24421 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24423 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24424 parameter with a "loop" parameter and use it instead of the
24425 loop containing DR_STMT. Don't check simple_iv when doing
24426 BB analysis. Describe the two analysis modes in the comment.
24428 2017-07-03 Tom de Vries <tom@codesourcery.com>
24430 PR tree-optimization/69468
24431 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24432 (find_same_succ_bb): Handle ignore_edge_flags.
24434 2017-07-03 Tom de Vries <tom@codesourcery.com>
24436 PR tree-optimization/81192
24437 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24439 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24441 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24443 2017-07-03 Tom de Vries <tom@codesourcery.com>
24445 PR tree-optimization/81192
24446 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24447 BB_SAME_SUCC (bb) == NULL.
24449 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24451 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24454 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24456 * dumpfile.c: Include profile-count.h
24457 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24459 (insert_cond_bb): Update profile.
24460 * tree-cfg.h (insert_cond_bb): Update prototype.
24461 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24462 * tree-dump.c: Do not include tree-cfg.
24464 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24466 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24468 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24470 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24473 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24475 * tree-complex.c (expand_complex_div_wide): update profile.
24477 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24478 Alan Hayward <alan.hayward@arm.com>
24479 David Sherwood <david.sherwood@arm.com>
24481 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24482 (CORETYPES_H): New define.
24483 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24484 (insn-modes-inline.h, s-modes-inline-h): New rules.
24485 (generated_files): Add insn-modes-inline.h.
24486 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24487 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24488 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24489 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24490 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24491 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24492 (build/gencodes.o, build/genconditions.o): Likewise.
24493 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24494 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24495 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24496 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24497 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24498 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24499 * coretypes.h: Include everything up to real.h for generators.
24500 Include insn-modes.h first. Include wide-int-print.h after
24501 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24502 * machmode.h: Don't include insn-modes.h here.
24503 * function-tests.c: Remove includes of signop.h, machmode.h,
24504 double-int.h and wide-int.h.
24506 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24508 * optc-save-gen.awk: Likewise.
24509 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24510 * godump.c: Remove include of wide-int-print.h.
24511 * pretty-print.h: Likewise.
24512 * wide-int-print.cc: Likewise.
24513 * wide-int.cc: Likewise.
24514 * hash-map-tests.c: Remove include of signop.h.
24515 * hash-set-tests.c: Likewise.
24516 * rtl-tests.c: Likewise.
24517 * mkconfig.sh: Remove include of machmode.h.
24518 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24520 (emit_insn_modes_inline_h): ...this new function. Emit the code
24521 into an insn-modes-inline.h header file, adding appropriate
24522 include guards and end comments.
24523 (emit_insn_modes_c_header): Remove include of machmode.h.
24524 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24526 (main): Handle -i flag and call emit_insn_modes_inline_h when
24529 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24531 * tree-ssa-strlen.c (strinfo): Rename the length field to
24532 nonzero_chars. Add a full_string_p field.
24533 (compare_nonzero_chars, zero_length_string_p): New functions.
24534 (get_addr_stridx): Add an offset_out parameter.
24535 Use compare_nonzero_chars.
24536 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24537 (new_strinfo): Update after above changes to strinfo.
24538 (set_endptr_and_length): Set full_string_p.
24539 (get_string_length): Update after above changes to strinfo.
24540 (unshare_strinfo): Update call to new_strinfo.
24541 (maybe_invalidate): Likewise.
24542 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24543 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24544 as a uhwi instead of an shwi. Update after above changes to
24545 strinfo and new_strinfo.
24546 (zero_length_string): Assert that chainsi contains full strings.
24547 Use zero_length_string_p. Update call to new_strinfo.
24548 (adjust_related_strinfos): Update after above changes to strinfo.
24549 Copy full_string_p from origsi.
24550 (adjust_last_stmt): Use zero_length_string_p.
24551 (handle_builtin_strlen): Update after above changes to strinfo and
24552 new_strinfo. Install the lhs as the string length if the previous
24553 entry didn't describe a full string.
24554 (handle_builtin_strchr): Update after above changes to strinfo
24556 (handle_builtin_strcpy): Likewise.
24557 (handle_builtin_strcat): Likewise.
24558 (handle_builtin_malloc): Likewise.
24559 (handle_pointer_plus): Likewise.
24560 (handle_builtin_memcpy): Likewise. Track nonzero characters
24561 that aren't necessarily followed by a nul terminator.
24562 (handle_char_store): Likewise.
24564 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24566 PR tree-optimization/80769
24567 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24568 for malloc and calloc. Document the new invariant that all related
24569 strinfos have delayed lengths or none do.
24570 (verify_related_strinfos): Move earlier in file.
24571 (set_endptr_and_length): New function, split out from...
24572 (get_string_length): ...here. Also set the lengths of related
24574 (zero_length_string): Assert that chainsi has known (rather than
24576 (adjust_related_strinfos): Likewise.
24578 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24580 PR tree-optimization/81136
24581 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24582 assert that two references with the same misalignment have the same
24583 compile-time misalignment if those compile-time misalignments
24586 2017-07-01 Andi Kleen <ak@linux.intel.com>
24588 * print-tree.c (print_node): Print all attributes.
24590 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24592 * cfg.c (scale_bbs_frequencies): New function.
24593 * cfg.h (scale_bbs_frequencies): Declare it.
24594 * cfgloopanal.c (single_likely_exit): Cleanup.
24595 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24597 (scale_loop_profile): Likewise.
24598 (loop_version): Likewise.
24599 (create_empty_loop_on_edge): Update.
24600 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24601 scale_loop_frequencies, scale_loop_profile, loopify,
24602 loop_version): Update prototypes.
24603 * modulo-sched.c (sms_schedule): Update.
24604 * predict.c (unlikely_executed_edge_p): Also check probability.
24605 (probably_never_executed_edge_p): Fix typo.
24606 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24607 * tree-parloops.c (gen_parallel_loop): Update.
24608 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24609 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24610 * tree-ssa-loop-split.c (split_loop): Update.
24611 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24612 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24613 (vect_loop_versioning): Update.
24614 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24616 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24618 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24620 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24622 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24623 to keep profile consistent.
24625 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24627 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24628 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24629 * profile-count.h (max_safe_multiplier): Make unsigned.
24630 (profile_count::guessed_zero): New.
24632 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24634 * bb-reorder.c (fix_up_crossing_landing_pad,
24635 fix_crossing_conditional_branches): Use make_single_succ_edge
24636 to keep profile consistent.
24638 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24640 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24643 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24646 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24647 the right scopes, make sure cond_jump isn't preserved between multiple
24648 iterations. Search for fallthru edge whenever there are 3+ edges and
24649 use find_fallthru_edge for it.
24651 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24653 Patch by Alexander Monakov <amonakov@ispras.ru>
24654 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24655 probabilities consistently.
24657 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24659 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24660 emit_cmp_and_jump_insns.
24662 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24665 * tree-inline.c (expand_call_inline): Combine profile statuses.
24667 2017-06-30 Andrew Pinski <apinski@cavium.com>
24669 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24670 fold_stmt returned true.
24672 2017-06-30 Nathan Sidwell <nathan@acm.org>
24674 * ggc.h (empty_string): Delete.
24675 * cfgexpand.c (expand_asm_stmt): Use plain "".
24676 * optabs.c (expand_asm_memory_barrier): Likewise.
24677 * stringpool.c (empty_string): Delete.
24678 (digit_vector, digit_string): Delete.
24679 (ggc_alloc_string): Use plain "", don't optimize single digit
24680 strings. Use ggc_alloc_atomic.
24682 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24684 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24685 comparison set and one other set, use the cost of the non-comparison
24688 2017-06-30 Nathan Sidwell <nathan@acm.org>
24690 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24693 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24695 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24696 loops. Remove now unneeded calls to gimple_switch_set_label() that
24697 just set removed labels to NULL_TREE.
24699 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24701 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24702 (set_range_info): ...here. Only call set_range_info_raw if domain
24704 (set_nonzero_bits): Call set_range_info_raw.
24705 * tree-ssanames.h (set_range_info_raw): New.
24707 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24710 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24711 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24712 of nonimmediate_operand and <store_mask_constraint> instead of m
24713 for the input operand. For V8FI iterator, always split if input
24714 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24715 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24716 <store_mask_predicate> instead of register_operand and
24717 <store_mask_constraint> instead of v for the input operand. Make
24718 sure both operands aren't MEMs for if not <mask_applied>.
24720 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24722 * lto-wrapper.c (copy_file) Close both file descriptors before
24725 2017-06-30 Martin Liska <mliska@suse.cz>
24728 * multiple_target.c (create_dispatcher_calls): Make ifunc
24729 also for function that don't have calls or are not referenced.
24731 2017-06-30 Richard Biener <rguenther@suse.de>
24733 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24734 analyze the first scalar stmt. Move vector type computation
24735 for the BB case here from ...
24736 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24737 live operation processing in the SLP case properly.
24739 2017-06-30 Richard Biener <rguenther@suse.de>
24741 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24743 2017-06-30 Martin Liska <mliska@suse.cz>
24746 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24747 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24749 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24751 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24754 2017-06-30 Richard Biener <rguenther@suse.de>
24756 PR tree-optimization/81249
24757 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24758 condition reduction result to original scalar type.
24760 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24762 * profile-count.h (enum profile_quality): Fix typos and whitespace
24765 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24767 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24768 type for branch probabilities.
24770 2017-06-29 Julian Brown <julian@codesourcery.com>
24771 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24773 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24774 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24775 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24776 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24778 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24780 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24781 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24782 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24783 CC usage from generic code to here.
24784 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24785 CC usage into the target macros.
24787 2017-06-29 Maya Rashish <coypu@sdf.org>
24789 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24792 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24794 * arm/arm-builtins.c: Include profile-count.h
24795 * except.c (sjlj_emit_function_enter): Use
24796 profile_probability::unlikely.
24798 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24800 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24801 and tocrel_offset be pointer args rather than implicitly using
24803 (legitimate_constant_pool_address_p, rs6000_emit_move,
24804 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24805 tocrel_offset and use in toc_relative_expr_p call.
24806 (print_operand, print_operand_address): Use static tocrel_base_oac
24807 and tocrel_offset_oac.
24808 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24811 2017-06-29 Maya Rashish <coypu@sdf.org>
24813 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24815 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24817 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24818 objects, take into account only the alignment of 'op0' and 'mode1' if
24821 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24823 * ccmp.c (ccmp_tree_comparison_p): New function.
24824 (ccmp_candidate_p): Update to use above function.
24825 (get_compare_parts): New function.
24826 (expand_ccmp_next): Update to use new functions.
24827 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24829 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24830 take mode as argument.
24831 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24832 * expr.c (expand_expr_real_1): Pass mode as argument.
24834 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24836 * combine.c (combine_instructions): Print insns to dump_file, together
24839 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24841 * asan.c (asan_emit_stack_protection): Update.
24842 (create_cond_insert_point): Update.
24843 * auto-profile.c (afdo_propagate_circuit): Update.
24844 * basic-block.h (struct edge_def): Turn probability to
24845 profile_probability.
24846 (EDGE_FREQUENCY): Update.
24847 * bb-reorder.c (find_traces_1_round): Update.
24848 (better_edge_p): Update.
24849 (sanitize_hot_paths): Update.
24850 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24851 (make_single_succ_edge): Update.
24852 (check_bb_profile): Update.
24853 (dump_edge_info): Update.
24854 (update_bb_profile_for_threading): Update.
24855 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24856 probabilitycount to 0.
24857 * cfgbuild.c (compute_outgoing_frequencies): Update.
24858 * cfgcleanup.c (try_forward_edges): Update.
24859 (outgoing_edges_match): Update.
24860 (try_crossjump_to_edge): Update.
24861 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24862 (expand_gimple_tailcall): Update.
24863 (construct_init_block): Use make_single_succ_edge.
24864 (construct_exit_block): Use make_single_succ_edge.
24865 * cfghooks.c (verify_flow_info): Update.
24866 (redirect_edge_succ_nodup): Update.
24867 (split_edge): Update.
24868 (account_profile_record): Update.
24869 * cfgloopanal.c (single_likely_exit): Update.
24870 * cfgloopmanip.c (scale_loop_profile): Update.
24871 (set_zero_probability): Remove.
24872 (duplicate_loop_to_header_edge): Update.
24873 * cfgloopmanip.h (loop_version): Update prototype.
24874 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24875 (force_nonfallthru_and_redirect): Update.
24876 (update_br_prob_note): Update.
24877 (rtl_verify_edges): Update.
24878 (purge_dead_edges): Update.
24879 (rtl_lv_add_condition_to_bb): Update.
24880 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24881 * cgraphunit.c (init_lowered_empty_function): Update.
24882 (cgraph_node::expand_thunk): Update.
24883 * cilk-common.c: Include profile-count.h
24884 * dojump.c (inv): Remove.
24885 (jumpifnot): Update.
24886 (jumpifnot_1): Update.
24887 (do_jump_1): Update.
24889 (do_jump_by_parts_greater_rtx): Update.
24890 (do_compare_rtx_and_jump): Update.
24891 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24892 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24893 * dwarf2cfi.c: Include profile-count.h
24894 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24895 (sjlj_emit_dispatch_table): Likewise.
24896 * explow.c: Include profile-count.h
24897 * expmed.c (emit_store_flag_force): Update.
24898 (do_cmp_and_jump): Update.
24899 * expr.c (compare_by_pieces_d::generate): Update.
24900 (compare_by_pieces_d::finish_mode): Update.
24901 (emit_block_move_via_loop): Update.
24902 (store_expr_with_bounds): Update.
24903 (store_constructor): Update.
24904 (expand_expr_real_2): Update.
24905 (expand_expr_real_1): Update.
24906 * expr.h (try_casesi, try_tablejump): Update prototypes.
24907 * gimple-pretty-print.c (dump_probability): Update.
24908 (dump_profile): New.
24909 (dump_gimple_label): Update.
24910 (dump_gimple_bb_header): Update.
24911 * graph.c (draw_cfg_node_succ_edges): Update.
24912 * hsa-gen.c (convert_switch_statements): Update.
24913 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24914 (find_if_case_1): Update.
24915 (find_if_case_2): Update.
24916 * internal-fn.c (expand_arith_overflow_result_store): Update.
24917 (expand_addsub_overflow): Update.
24918 (expand_neg_overflow): Update.
24919 (expand_mul_overflow): Update.
24920 (expand_vector_ubsan_overflow): Update.
24921 * ipa-cp.c (good_cloning_opportunity_p): Update.
24922 * ipa-split.c (split_function): Use make_single_succ_edge.
24923 * ipa-utils.c (ipa_merge_profiles): Update.
24924 * loop-doloop.c (add_test): Update.
24925 (doloop_modify): Update.
24926 * loop-unroll.c (compare_and_jump_seq): Update.
24927 (unroll_loop_runtime_iterations): Update.
24928 * lra-constraints.c (lra_inheritance): Update.
24929 * lto-streamer-in.c (input_cfg): Update.
24930 * lto-streamer-out.c (output_cfg): Update.
24931 * mcf.c (adjust_cfg_counts): Update.
24932 * modulo-sched.c (sms_schedule): Update.
24933 * omp-expand.c (expand_omp_for_init_counts): Update.
24934 (extract_omp_for_update_vars): Update.
24935 (expand_omp_ordered_sink): Update.
24936 (expand_omp_for_ordered_loops): Update.
24937 (expand_omp_for_generic): Update.
24938 (expand_omp_for_static_nochunk): Update.
24939 (expand_omp_for_static_chunk): Update.
24940 (expand_cilk_for): Update.
24941 (expand_omp_simd): Update.
24942 (expand_omp_taskloop_for_outer): Update.
24943 (expand_omp_taskloop_for_inner): Update.
24944 * omp-simd-clone.c (simd_clone_adjust): Update.
24945 * optabs.c (expand_doubleword_shift): Update.
24946 (expand_abs): Update.
24947 (emit_cmp_and_jump_insn_1): Update.
24948 (expand_compare_and_swap_loop): Update.
24949 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24950 * predict.c (predictable_edge_p): Update.
24951 (edge_probability_reliable_p): Update.
24952 (set_even_probabilities): Update.
24953 (combine_predictions_for_insn): Update.
24954 (combine_predictions_for_bb): Update.
24955 (propagate_freq): Update.
24956 (estimate_bb_frequencies): Update.
24957 (force_edge_cold): Update.
24958 * profile-count.c (profile_count::dump): Add missing space into dump.
24959 (profile_count::debug): Add newline.
24960 (profile_count::differs_from_p): Explicitly convert to unsigned.
24961 (profile_count::stream_in): Update.
24962 (profile_probability::dump): New member function.
24963 (profile_probability::debug): New member function.
24964 (profile_probability::differs_from_p): New member function.
24965 (profile_probability::differs_lot_from_p): New member function.
24966 (profile_probability::stream_in): New member function.
24967 (profile_probability::stream_out): New member function.
24968 * profile-count.h (profile_count_quality): Rename to ...
24969 (profile_quality): ... this one.
24970 (profile_probability): New.
24971 (profile_count): Update.
24972 * profile.c (compute_branch_probabilities): Update.
24973 * recog.c (peep2_attempt): Update.
24974 * sched-ebb.c (schedule_ebbs): Update.
24975 * sched-rgn.c (find_single_block_region): Update.
24976 (compute_dom_prob_ps): Update.
24977 (schedule_region): Update.
24978 * sel-sched-ir.c (compute_succs_info): Update.
24979 * stmt.c (struct case_node): Update.
24980 (do_jump_if_equal): Update.
24981 (get_outgoing_edge_probs): Update.
24982 (conditional_probability): Update.
24983 (emit_case_dispatch_table): Update.
24984 (expand_case): Update.
24985 (expand_sjlj_dispatch_table): Update.
24986 (emit_case_nodes): Update.
24987 * targhooks.c: Update.
24988 * tracer.c (better_p): Update.
24989 (find_best_successor): Update.
24990 * trans-mem.c (expand_transaction): Update.
24991 * tree-call-cdce.c: Update.
24992 * tree-cfg.c (gimple_split_edge): Upate.
24993 (move_sese_region_to_fn): Upate.
24994 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24995 * tree-eh.c (lower_resx): Upate.
24996 (cleanup_empty_eh_move_lp): Upate.
24997 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24998 * tree-inline.c (copy_edges_for_bb): Update.
24999 (copy_cfg_body): Update.
25000 * tree-parloops.c (gen_parallel_loop): Update.
25001 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25002 (gimple_gen_time_profiler): Update.
25003 * tree-ssa-dce.c (remove_dead_stmt): Update.
25004 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25005 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25006 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25007 (unloop_loops): Update.
25008 (try_peel_loop): Update.
25009 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25010 * tree-ssa-loop-split.c (connect_loops): Update.
25011 (split_loop): Update.
25012 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25013 (hoist_guard): Update.
25014 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25015 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25016 (value_replacement): Update.
25017 * tree-ssa-reassoc.c (branch_fixup): Update.
25018 * tree-ssa-tail-merge.c (replace_block_by): Update.
25019 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25020 (create_edge_and_update_destination_phis): Update.
25021 (compute_path_counts): Update.
25022 (recompute_probabilities): Update.
25023 (update_joiner_offpath_counts): Update.
25024 (freqs_to_counts_path): Update.
25025 (duplicate_thread_path): Update.
25026 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25027 (struct switch_conv_info): Update.
25028 (gen_inbound_check): Update.
25029 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25030 (vect_do_peeling): Update.
25031 (vect_loop_versioning): Update.
25032 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25033 (optimize_mask_stores): Update.
25034 * ubsan.c (ubsan_expand_null_ifn): Update.
25035 * value-prof.c (gimple_divmod_fixed_value): Update.
25036 (gimple_divmod_fixed_value_transform): Update.
25037 (gimple_mod_pow2): Update.
25038 (gimple_mod_pow2_value_transform): Update.
25039 (gimple_mod_subtract): Update.
25040 (gimple_mod_subtract_transform): Update.
25041 (gimple_ic): Update.
25042 (gimple_stringop_fixed_value): Update.
25043 (gimple_stringops_transform): Update.
25044 * value-prof.h: Update.
25046 2017-06-29 Carl Love <cel@us.ibm.com>
25048 * config/rs6000/rs6000-c.c: Add support for built-in functions
25049 vector signed int vec_signed (vector float);
25050 vector signed long long vec_signed (vector double);
25051 vector signed int vec_signed2 (vector double, vector double);
25052 vector signed int vec_signede (vector double);
25053 vector signed int vec_signedo (vector double);
25054 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25055 instruction generator.
25056 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25057 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25058 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25060 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25061 vunsignede_v2df): Add define_expands.
25062 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25063 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25064 VEC_UNSIGNEDO): Add definitions.
25065 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25066 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25067 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25068 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25069 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25070 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25071 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25072 vec_unsignede, vec_unsignedo): Add builtin defines.
25073 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25075 * doc/extend.texi: Update the built-in documentation file for the
25076 new built-in functions.
25078 2017-06-29 Richard Biener <rguenther@suse.de>
25080 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25081 reduction chains to LOOP_VINFO_REDUCTIONS.
25082 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25083 SLP reductions after processing reduction chains.
25085 2017-06-29 Nathan Sidwell <nathan@acm.org>
25087 * builtins.c (fold_builtin_FUNCTION): Use
25088 lang_hooks.decl_printable_name.
25090 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25092 PR middle-end/81194
25093 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25094 with only one label.
25095 * stmt.c (expand_case): Assert NCASES is greater than one.
25097 2017-06-29 Richard Biener <rguenther@suse.de>
25099 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25101 (group_case_labels): Likewise.
25102 (find_taken_edge): Push sanity checking on val to workers...
25103 (find_taken_edge_cond_expr): ... here
25104 (find_taken_edge_switch_expr): ... and here, handle cases
25105 with just a default label.
25106 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25107 (group_case_labels): Likewise.
25108 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25109 group_case_labels does anything cleanup the CFG again.
25111 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25113 PR tree-optimization/81196
25114 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25115 exit condition comparing two IVs.
25117 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25119 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25120 profile to the dummy entry at the end of the list of architectures.
25121 * config/arm/arm-cpu-cdata.h: Regenerated.
25123 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25124 Michael Collison <michael.collison@arm.com>
25127 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25129 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25130 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25131 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25132 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25133 of shift when the shift amount is masked with constant equal to
25134 the size of the mode.
25135 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25138 2017-06-29 Martin Liska <mliska@suse.cz>
25140 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25142 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25145 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25148 (edge_growth_cache_entry::edge_growth_cache_entry): New
25150 (reset_edge_growth_cache): Update to use constructor.
25152 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25154 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25155 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25156 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25158 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25160 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25161 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25163 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25165 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25166 (*-linux-uclibc*): Add t-uclibc tmake_file.
25167 * config/t-musl: New.
25168 * config/t-uclibc: New.
25170 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25172 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25174 (gen_comm_data): Emit architectural setting of arch_prof.
25175 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25177 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25178 (armv8-m.base, armv8-m.main): Likewise.
25179 * arm-protos.h (arm_build_target): Add profile field.
25180 (arch_option): Likewise.
25181 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25183 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25184 arm_active_target.profile.
25186 2017-06-28 Richard Biener <rguenther@suse.de>
25188 PR middle-end/81227
25189 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25190 TYPE_OVERFLOW_WRAPS.
25191 * match.pd (negate_expr_p): Likewise.
25192 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25193 fold_build2, not fold_binary.
25195 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25197 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25198 Convert memory address to Pmode.
25199 (aarch64_print_operand): Assert MEM operands are always Pmode.
25201 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25204 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25205 Remove redundant if.
25206 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25207 * config/arm/aarch-common-protos.h
25208 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25209 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25211 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25214 * multiple_target.c (create_dispatcher_calls): Set the default
25215 clone to be static, not public.
25217 2017-06-28 Richard Biener <rguenther@suse.de>
25219 * tree-vect-loop.c (vectorizable_reduction): Move special
25220 cond reduction IV var creation ...
25221 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25222 parameter. Use STMT_VINFO_VECTYPE.
25223 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25226 2017-06-28 Martin Liska <mliska@suse.cz>
25229 * ipa-visibility.c (non_local_p): Handle visibility.
25231 2017-06-28 Martin Liska <mliska@suse.cz>
25234 * common.opt: Add IntegerRange to various options.
25235 * opt-functions.awk (integer_range_info): New function.
25236 * optc-gen.awk: Add integer_range_info to cl_options struct.
25237 * opts-common.c (decode_cmdline_option): Handle
25238 CL_ERR_INT_RANGE_ARG.
25239 (cmdline_handle_error): Likewise.
25240 * opts.c (print_filtered_help): Show valid interval in
25241 when --help is provided.
25242 * opts.h (struct cl_option): Add range_min and range_max fields.
25243 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25245 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25247 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25248 (x * C EQ/NE y * C): New transformation.
25250 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25252 * genmultilib (combination_space): Accept '+' in option names.
25254 2017-06-28 Martin Liska <mliska@suse.cz>
25257 * asan.c (instrument_derefs): Bail out inner references
25258 that are hard register variables.
25260 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25263 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25264 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25266 2017-06-28 Richard Biener <rguenther@suse.de>
25268 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25269 (vect_get_slp_defs): Adjust.
25270 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25272 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25274 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25275 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25276 (vectorizable_reduction): Adjust.
25277 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25279 (vect_get_slp_defs): Likewise.
25280 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25281 (vectorizable_bswap): Adjust.
25282 (vectorizable_call): Likewise.
25283 (vectorizable_conversion): Likewise.
25284 (vectorizable_assignment): Likewise.
25285 (vectorizable_shift): Likewise.
25286 (vectorizable_operation): Likewise.
25287 (vectorizable_store): Likewise.
25288 (vectorizable_condition): Likewise.
25289 (vectorizable_comparison): Likewise.
25291 2017-06-28 Michael Collison <michael.collison@arm.com>
25294 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25296 (arm_gen_movmemqi): Removed unused variable 'i'.
25297 Convert 'for' loop into 'while' loop.
25298 (arm_expand_prologue): Remove last unnecessary set of insn.
25299 (thumb_pop): Remove unused variable 'pushed_words'.
25300 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25302 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25304 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25305 * config/s390/s390-protos.h: Add prototype of
25306 s390_rel_address_ok_p.
25307 * config/s390/s390.c (s390_got_symbol): New function.
25308 (s390_rel_address_ok_p): New function.
25309 (legitimize_pic_address): Use s390_rel_address_ok_p.
25310 (s390_load_got): Use s390_got_symbol.
25311 (s390_option_override): Issue error if
25312 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25313 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25315 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25317 2017-06-27 Andrew Pinski <apinski@cavium.com>
25319 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25320 (X * copysign (1.0, X)): New pattern.
25321 (X * copysign (1.0, -X)): New pattern.
25322 (copysign (-1.0, CST)): New pattern.
25324 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25326 * genmultilib (combination_space): Remove variable.
25327 Validate reuse rules against regular expression for any sequence
25328 of multilib options in any order.
25330 2017-06-27 Michael Collison <michael.collison@arm.com>
25332 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25333 call aarch64_split_simd_combine.
25334 * (aarch64_combine_internal<mode>): Delete pattern.
25335 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25336 Allow register and subreg operands.
25338 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25340 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25341 specific need, just fallback on defaults.
25342 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25344 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25345 Olivier Hainque <hainque@adacore.com>
25347 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25349 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25350 targets. Pick a default if no particular attempt applied.
25351 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25354 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25356 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25357 (x86_64-wrs-vxworks7): Likewise.
25359 2017-06-27 Marek Polacek <polacek@redhat.com>
25362 * ubsan.c (instrument_null): Check get_base_address's result for null.
25364 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25366 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25368 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25370 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25371 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25372 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25373 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25374 New function types.
25375 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25376 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25377 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25378 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25379 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25380 BUILT_IN_FEUPDATEENV): New builtins.
25381 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25382 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25383 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25384 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25386 (builtin_structptr_types): Adjust size.
25387 * tree.c (builtin_structptr_types): Add four entries.
25389 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25390 Olivier Hainque <hainque@adacore.com>
25392 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25393 (TLS_SYM): New local macro, forcing reference to __tls__ on
25394 link command lines for VxWorks 7 RTPs, triggering initialization
25396 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25397 OS features TLS support, true for RTPs on VxWorks 7.
25398 * config/vxworks.c (vxworks_override_options): Setup emutls
25401 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25403 * predict.c (test_prediction_value_range): Use -1U instead of -1
25404 to avoid narrowing conversion warning.
25405 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25406 to avoid narrowing conversion warning.
25407 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25409 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25411 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25413 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25414 64bit configurations.
25415 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25416 (SIZE_TYPE): Likewise.
25417 * config/vxworks.c (vxworks_emutls_var_fields): Use
25418 long_unsigned_type_node instead of unsigned_type_node as the offset
25419 field type, which is "pointer" mode in emutls.c.
25421 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25424 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25426 PR middle-end/81207
25427 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25428 gimple_vuse copying separately from gimple_vdef copying.
25430 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25432 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25435 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25436 Olivier Hainque <hainque@adacore.com>
25438 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25439 for all vxworks7 targets.
25440 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25441 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25442 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25443 variations for VX6/VX7 and 32/64bits later on in ...
25444 (VXWORKS_LIB_SPEC): Leverage new macros.
25445 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25446 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25448 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25450 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25451 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25453 2017-06-26 Carl Love <cel@us.ibm.com>
25455 * config/rs6000/rs6000-c.c: Add support for built-in functions
25456 vector bool char vec_reve (vector bool char);
25457 vector signed char vec_reve (vector signed char);
25458 vector unsigned char vec_reve (vector unsigned char);
25459 vector bool int vec_reve (vector bool int);
25460 vector signed int vec_reve (vector signed int);
25461 vector unsigned int vec_reve (vector unsigned int);
25462 vector bool long long vec_reve (vector bool long long);
25463 vector signed long long vec_reve (vector signed long long);
25464 vector unsigned long long vec_reve (vector unsigned long long);
25465 vector bool short vec_reve (vector bool short);
25466 vector signed short vec_reve (vector signed short);
25467 vector double vec_reve (vector double);
25468 vector float vec_reve (vector float);
25469 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25470 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25471 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25472 (altivec_vreve): New pattern.
25473 * config/rs6000/altivec.h (vec_reve): New define.
25474 * doc/extend.texi (vec_rev): Update the built-in documentation file
25475 for the new built-in functions.
25477 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25479 PR tree-optimization/71815
25480 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25482 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25484 (slsr_process_phi): Likewise.
25485 (replace_uncond_cands_and_profitable_phis): Don't replace a
25486 multiply candidate with a stride of 1 (copy or cast).
25487 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25489 (lowest_cost_path): Likewise.
25490 (total_savings): Likewise.
25492 2017-06-26 Richard Biener <rguenther@suse.de>
25495 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25496 Use def_builtin_pure for all gather builtins.
25498 2017-06-26 Richard Biener <rguenther@suse.de>
25500 PR tree-optimization/81203
25501 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25502 non-dominating BBs.
25504 2017-06-26 Marek Polacek <polacek@redhat.com>
25507 * doc/invoke.texi: Document -Wmultistatement-macros.
25509 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25511 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25512 arm_neon_ok_no_float_abi effective target.
25514 2017-06-26 Richard Biener <rguenther@suse.de>
25516 PR tree-optimization/80928
25517 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25518 (copy_bbs): Set BB_DUPLICATED flag early.
25519 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25521 (execute_on_shrinking_pred): Likewise.
25522 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25523 BB_DUPLICATED blocks.
25524 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25525 iterate over all PHIs considering removal of *gsi.
25527 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25529 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25532 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25534 * config/rs6000/rs6000-string.c: (expand_block_clear,
25535 do_load_for_compare, select_block_compare_mode,
25536 compute_current_alignment, expand_block_compare,
25537 expand_strncmp_align_check, expand_strn_compare,
25538 expand_block_move, rs6000_output_load_multiple)
25539 Move functions related to string/block move/compare
25540 to a separate file.
25541 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25542 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25543 for this function which is now used in two files.
25544 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25545 * config.gcc: Add rs6000-string.o to extra_objs for
25546 targets powerpc*-*-* and rs6000*-*-*.
25548 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25551 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25552 32-bit, since indexed is not valid for DImode.
25553 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25554 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25555 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25556 (define_peephole2 for Altivec d-form store): Likewise.
25559 * multiple_target.c (create_dispatcher_calls): Only create the
25560 dispatcher call if the function is the default clone of a
25561 versioned function.
25563 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25565 PR middle-end/80902
25566 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25567 a call, force the call to not be a tail call.
25569 2017-06-23 Jeff Law <law@redhat.com>
25571 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25574 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25576 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25577 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25578 handling for early expansion of vector shifts (sl,sr,sra,rl).
25579 (builtin_function_type): Add vector shift right instructions
25580 to the unsigned argument list.
25582 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25584 rtl-optimizatoin/79286
25585 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25586 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25587 trap. PIC register plus a const unspec without offset can never trap.
25589 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25591 * tree.h (builtin_structptr_type): New type.
25592 (builtin_structptr_types): Declare new array.
25593 * tree.c (builtin_structptr_types): New array.
25594 (free_lang_data, build_common_tree_nodes): Use it.
25596 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25599 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25602 2017-06-22 Matt Turner <mattst88@gmail.com>
25604 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25605 Lake models to skylake case. Assume skylake for unknown
25606 models with clflushopt.
25608 2017-06-22 Jeff Law <law@redhat.com>
25610 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25611 frame sizes that do not satisfy aarch64_uimm12_shift.
25613 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25615 * profile-count.h (apply_probability,
25616 apply_scale, probability_in): Fix checks for zero.
25618 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25620 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25621 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25623 2016-06-22 Richard Biener <rguenther@suse.de>
25625 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25626 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25627 REDUC_MAX_EXPR support.
25628 (vectorizable_reduction): Likewise.
25629 (vect_create_epilog_for_reduction): Likewise.
25631 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25633 * match.pd (A / (1 << B) -> A >> B): New.
25634 * generic-match-head.c: Include optabs-tree.h.
25635 * gimple-match-head.c: Likewise.
25636 * optabs-tree.h (target_supports_op_p): New.
25637 * optabs-tree.c (target_supports_op_p): New.
25639 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25641 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25642 $gcc_cv_ld --help output.
25643 (gcc_cv_ld_demangle): Likewise.
25644 (gcc_cv_ld_eh_frame_hdr): Likewise.
25645 (gcc_cv_ld_pie): Likewise.
25646 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25647 (gcc_cv_ld_buildid): Likewise.
25648 (gcc_cv_ld_sysroot): Likewise.
25649 (ld_bndplt_support): Likewise.
25650 (ld_pushpopstate_support): Likewise.
25651 * configure: Regenerate.
25652 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25654 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25657 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25658 operands indexes to avoid gap between operands and match_dups.
25660 2017-06-21 Andrew Pinski <apinski@cavium.com>
25662 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25663 Increment Arith_shift and Arith_shift_reg by 1.
25664 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25666 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25667 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25668 (aarch64_strip_extend): Add new argument and test for it.
25669 (aarch64_cheap_mult_shift_p): New function.
25670 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25671 add a cost if it is true.
25672 Update calls to aarch64_strip_extend.
25673 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25675 2017-06-21 Andrew Pinski <apinski@cavium.com>
25677 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25679 (thunderxt88): Likewise.
25680 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25681 (thunderx_prefetch_tune): New variable.
25682 (thunderx2t99_prefetch_tune): Update for the correct values.
25683 (thunderxt88_tunings): New variable.
25684 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25685 generic_prefetch_tune.
25686 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25688 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25690 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25691 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25692 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25693 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25694 (aarch64_atomic_cas<mode>, GPI): Likewise.
25696 2017-06-21 Martin Liska <mliska@suse.cz>
25698 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25699 statements on cold and hot labels.
25700 * predict.c (tree_estimate_probability_bb): Remove the
25701 prediction from this place.
25703 2017-06-21 Martin Liska <mliska@suse.cz>
25705 PR tree-optimization/79489
25706 * gimplify.c (maybe_add_early_return_predict_stmt): New
25708 (gimplify_return_expr): Call the function.
25709 * predict.c (tree_estimate_probability_bb): Remove handling
25711 * predict.def: Update comment about early return predictor.
25712 * gimple-predict.h (is_gimple_predict): New function.
25713 * predict.def: Change default value of early return to 66.
25714 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25716 * passes.def: Put pass_strip_predict_hints to the beginning of
25719 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25721 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25722 FUNCTION_DECL declarations.
25723 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25725 (dwaf2out_decl): Likewise.
25726 * godump.c (go_early_global_decl): Skip call to the real debug hook
25727 for FUNCTION_DECL declarations.
25728 * passes.c (rest_of_decl_compilation): Skip call to the
25729 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25730 -fdump-go-spec is passed.
25732 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25734 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25735 Reorder for compactness.
25736 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25737 (def_builtin_pure, def_builtin_pure2): New functions.
25738 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25740 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25742 * match.pd (nop_convert): New predicate.
25743 ((A +- CST1) +- CST2): Allow some NOP conversions.
25745 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25748 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25749 with ctors/dtors if GOVD_SHARED is set.
25751 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25753 * config/aarch64/aarch64.md (movti_aarch64):
25754 Emit mov rather than orr.
25755 (movtf_aarch64): Likewise.
25756 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25757 Emit mov rather than orr.
25759 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25761 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25762 Swap alternatives, make integer dup more expensive.
25764 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25766 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25767 Return true for non-tls symbols.
25769 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25771 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25772 (cortex-a75): Likewise.
25773 (cortex-a75.cortex-a55): Likewise.
25774 * config/aarch64/aarch64-tune.md: Regenerate.
25775 * doc/invoke.texi (-mtune): Document new values for -mtune.
25777 2017-06-21 Tom de Vries <tom@codesourcery.com>
25779 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25780 stack_size feature.
25781 (Effective-Target Keywords, Other attributes): Suggest using
25782 dg-add-options stack_size feature to get stack limit in stack_size
25783 effective target documentation.
25785 2017-06-21 Julian Brown <julian@codesourcery.com>
25786 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25788 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25789 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25790 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25792 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25793 attribute type list for neon_multiply.
25794 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25795 attribute type list for neon_multiply.
25796 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25797 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25798 attribute type list for neon_multiply.
25799 * config/arm/types.md (crypto_pmull): Add.
25800 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25801 attribute type list.
25803 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25805 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25808 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25810 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25811 to make sure not to dereference a NULL cost_classes_ptr pointer.
25813 2017-06-20 Carl Love <cel@us.ibm.com>
25815 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25816 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25817 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25818 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25819 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25820 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25821 VMULOSW): New enum "unspec" values.
25822 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25823 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25824 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25825 altivec_vmulosw): New patterns.
25826 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25827 VMULOSW): Add definitions.
25829 2017-06-20 Julia Koval <julia.koval@intel.com>
25831 * config/i386/i386.c: Fix rounding expand for new pattern.
25832 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25834 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25836 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25837 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25839 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25841 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25844 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25846 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25847 architecture, then by implementer ID.
25848 * config/aarch64/aarch64-tune.md: Regenerate.
25850 2017-06-20 Richard Biener <rguenther@suse.de>
25852 PR middle-end/81097
25853 * fold-const.c (split_tree): Fold to type before negating.
25855 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25857 * diagnostic-show-locus.c
25858 (selftest::test_fixit_deletion_affecting_newline): New function.
25859 (selftest::diagnostic_show_locus_c_tests): Call it.
25861 2017-06-20 Andreas Schwab <schwab@suse.de>
25864 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25867 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25869 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25870 __ARM_FEATURE_COPROC according to support.
25872 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25874 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25875 Rewritten to avoid overflow for > 32-bit pointers.
25878 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25879 by removing enum keyword.
25880 (ubsan_type_descriptor): Likewise. Formatting fix.
25883 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25884 splitter): Require TARGET_SSE2 in the condition.
25886 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25889 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25890 for doing vector set of SFmode on ISA 3.0.
25891 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25892 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25894 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25895 SFmode value into a V4SF variable that was extracted from another
25896 V4SF variable without converting the element to double precision
25897 and back to single precision vector format.
25898 (vsx_insert_extract_v4sf_p9_2): Likewise.
25900 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25902 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25903 in UWHI to avoid undefined overflow.
25906 * ubsan.h (enum ubsan_encode_value_phase): New.
25907 (ubsan_encode_value): Change second argument to
25908 enum ubsan_encode_value_phase with default value of
25909 UBSAN_ENCODE_VALUE_GENERIC.
25910 * ubsan.c (ubsan_encode_value): Change second argument to
25911 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25912 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25913 create_tmp_var_raw instead of create_tmp_var and use a
25915 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25916 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25917 ubsan_encode_value callers.
25920 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25921 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25922 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25924 2017-06-19 Richard Biener <rguenther@suse.de>
25926 PR middle-end/81118
25927 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25928 estimates if we changed anything.
25930 2017-06-19 Richard Biener <rguenther@suse.de>
25932 PR tree-optimization/80887
25933 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25934 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25935 simplified lookups, then reset mprts_hook.
25936 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25938 (try_to_simplify): Likewise.
25940 2017-06-19 Martin Liska <mliska@suse.cz>
25943 * gimplify.c (gimplify_switch_expr):
25944 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25946 2017-06-19 Martin Liska <mliska@suse.cz>
25948 * doc/install.texi: Document that PGO runs in 4 stages.
25950 2017-06-19 Martin Liska <mliska@suse.cz>
25953 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25954 to dispatcher function name.
25955 * multiple_target.c (replace_function_decl): New function.
25956 (create_dispatcher_calls): Redirect both edges and references.
25958 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25960 * profile-count.c (profile_count::dump): Dump quality.
25961 (profile_count::differs_from_p): Update for unsigned val.
25962 * profile-count.h (profile_count_quality): New enum.
25963 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25965 2017-06-19 Richard Biener <rguenther@suse.de>
25967 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25968 struct function as arg.
25969 (estimate_numbers_of_iterations): Export overload with loop arg.
25970 (free_numbers_of_iterations_estimates_loop): Use an overload of
25971 free_numbers_of_iterations_estimates instead.
25972 * tree-cfg.c (remove_bb): Adjust.
25973 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25974 * tree-parloops.c (gen_parallel_loop): Likewise.
25975 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25977 (tree_unroll_loops_completely): Likewise.
25978 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25979 Use an overload instead and export.
25980 (estimated_loop_iterations): Adjust.
25981 (max_loop_iterations): Likewise.
25982 (likely_max_loop_iterations): Likewise.
25983 (estimate_numbers_of_iterations): Take struct function as arg
25985 (loop_exits_before_overflow): Adjust.
25986 (free_numbers_of_iterations_estimates_loop): Use an overload.
25987 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25988 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25990 2017-06-19 Richard Biener <rguenther@suse.de>
25993 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25994 RANGE_EXPR conservatively.
25996 2017-06-16 Carl Love <cel@us.ibm.com>
25998 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25999 definitions for vec_float, vec_float2, vec_floato,
26000 vec_floate built-ins.
26001 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26002 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26004 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26005 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26006 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26007 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26008 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26009 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26010 vec_floato): Add builtin defines.
26011 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26012 Update the built-in documentation file for the new built-in
26015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26017 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26018 (mthumb): Mark as the negative of -marm.
26020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26022 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26024 (ARM Options, -mtune): Document that this accepts the same
26025 extension options as -mcpu.
26026 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26030 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26031 permitted extensions.
26033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26035 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26036 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26037 (armv8-m.main): Add option +nodsp.
26038 * config/arm/arm-cpu-cdata.h: Regenerated.
26040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26042 * config/arm/t-fuchsia: New file.
26043 * config.gcc (arm*-*-fuchsia*): Use it.
26045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26047 * config/arm/t-symbian: Rewrite for new option infrastructure.
26049 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26051 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26052 (MULTILIB_REQUIRED): Likewise.
26054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26056 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26057 (MULTILIB_RESUE): Likewise.
26058 (MULTILIB_MATCHES): Likewise.
26059 (MULTLIB_REQUIRED): Likewise.
26061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26063 * config/arm/t-rtems: Rewrite for new option framework.
26065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26067 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26068 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26069 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26070 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26071 * config/arm/t-multilib: ... here.
26072 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26073 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26074 armv7-a and armv8*-a when A-profile libraries have not been built.
26075 * config/arm/t-rmprofile: Rewrite.
26077 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26079 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26080 with a backslash. Remove the backslash after substituting unescaped
26082 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26086 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26087 the driver to pass through the default setting of -mfloat-abi.
26088 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26090 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26091 (all_feat_combs): New rule.
26092 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26094 * config/arm/t-aprofile: Rewrite.
26096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26098 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26099 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26100 fpu is not specified by the user/command-line.
26101 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26102 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26103 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26104 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26105 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26106 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26108 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26110 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26111 * config/arm/t-arm-elf: Rewritten.
26113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26115 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26116 have some floating-point instructions.
26117 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26118 (TARGET_MAYBE_HARD_FLOAT): New macro.
26119 * config/arm/arm-builtins.c (arm_init_builtins): Use
26120 TARGET_MAYBE_HARD_FLOAT.
26121 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26125 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26126 (configargs.h): Include it.
26127 (arm_print_hint_for_fpu_option): New function.
26128 (arm_parse_fpu_option): New function.
26129 (candidate_extension): New class.
26130 (arm_canon_for_multilib): New function.
26131 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26132 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26133 (ARCH_CANONICAL_SPECS): New macro.
26134 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26136 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26138 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26139 are set after handling multilib fragments. Set target_cpu_default2
26142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26144 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26146 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26147 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26149 (check_arch): Likewise.
26150 * config/arm/arm.c (arm_configure_build_target): Handle
26151 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26152 options in the default.
26154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26156 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26157 when an option is an alias of another.
26158 * config/arm/parsecpu.awk (optalias): New parser token.
26159 (gen_comm_data): Mark non-alias options as such. Emit entries
26160 for extension aliases.
26161 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26162 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26163 (armv6kz, armv6zk, armv6t2): Likewise.
26164 (armv7): Make vfpv3-d16 an alias.
26165 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26167 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26168 Sort in canonical order.
26169 (armv8-a): Sort in canonical order.
26170 (armv8.1-a, armv8.2-a): Likewise.
26171 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26173 (cortex-a9): Sort in canonical order.
26174 * config/arm/arm.c (selftests.h): Include it.
26175 (arm_test_cpu_arch_data): New function.
26176 (arm_run_self_tests): New function.
26177 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26178 (targetm): Move declaration to the end of the file.
26179 * arm-cpu-cdata.h: Regenerated.
26181 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26183 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26184 call to target_mode_check describing the type of option passed.
26185 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26186 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26187 arm_parse_cpu_option_name to match parameters against list of
26189 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26190 arm_arch_core_flags data structure.
26191 * config/arm/arm-cpu_cdata.h: Regenerated.
26193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26195 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26197 (arm_print_hint_for_cpu_option): Likewise.
26198 (arm_print_hint_for_arch_option): Likewise.
26199 (arm_parse_cpu_option_name): Likewise.
26200 (arm_parse_arch_option_name): Likewise.
26201 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26202 of entries in the all_fpus list.
26203 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26204 (arm_parse_cpu_option_name): Declare.
26205 (arm_parse_arch_option_name): Declare.
26206 (arm_parse_option_features): Declare.
26207 (arm_intialize_isa): Declare.
26208 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26210 (gen_comm_data): ... here. Make definitions non-static.
26211 * config/arm/arm-cpu-data.h: Regenerated.
26212 * config/arm/arm-cpu-cdata.h: Regenerated.
26214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26216 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26217 (cpu_arch_extension): New structure.
26218 (cpu_arch_option, arch_option, cpu_option): New structures.
26219 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26220 architecture types.
26221 (gen_data): Generate new format data tables.
26222 * config/arm/arm.c (cpu_tune): New structure.
26223 (cpu_option, processors): Delete.
26224 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26225 (arm_print_hint_for_cpu_option): ... this and ...
26226 (arm_print_hint_for_arch_option): ... this.
26227 (arm_parse_arch_cpu_name): Delete. Replace with ...
26228 (arm_parse_cpu_option_name): ... this and ...
26229 (arm_parse_arch_option_name): ... this.
26230 (arm_unrecognized_feature): Change type of target parameter to
26232 (arm_parse_arch_cpu_features): Delete. Replace with ...
26233 (arm_parse_option_features): ... this.
26234 (arm_configure_build_target): Rework to use new configuration data
26236 (arm_print_tune_info): Rework for new configuration data tables.
26237 * config/arm/arm-cpu-data.h: Regenerated.
26238 * config/arm/arm-cpu.h: Regenerated.
26240 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26242 * Makefile.in (OBJS): Move sbitmap.o from here ...
26243 (OBJS-libcommon): ... to here.
26245 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26247 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26248 (ISA_ALL_CRYPTO): New macro.
26249 (ISA_ALL_SIMD): New macro
26250 (ISA_ALL_FP): New macro.
26251 * config/arm/arm.c (fpu_bitlist): Update initializer.
26252 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26254 (arm9e): Add fpu. Add option for nofp
26255 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26256 (arm926ej-s, arm1026ej-s): Likewise.
26257 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26258 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26259 neon-fp16, neon-vfpv4, nofp and nosimd.
26260 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26261 (cortex-a8): Add fpu. Add option for nofp.
26262 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26263 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26264 (cortex-r4f): Add fpu.
26265 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26266 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26268 (cortex-r8): Likewise.
26269 (cortex-m4): Add fpu. Add option for nofp.
26270 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26271 (cortex-a17.cortex-a7): Likewise.
26272 (cortex-a32): Add fpu. Add options for crypto and nofp.
26273 (cortex-a35, cortex-a53): Likewise.
26274 (cortex-a57): Add fpu. Add option for crypto.
26275 (cortex-a72, cortex-a73): Likewise.
26276 (exynos-m1): Likewise.
26277 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26278 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26279 (cortex-m33): Add fpu. Add option for nofp.
26280 * config/arm/arm-cpu-cdata.h: Regenerated
26281 * config/arm/arm-cpu-data.h: Regenerated.
26283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26285 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26286 (armv5te, armv5tej): Likewise.
26287 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26288 (armv7): Add options fp and vfpv3-d16.
26289 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26290 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26292 (armv7ve): Likewise.
26293 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26294 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26295 (armv8-a): Add nocrypto option.
26296 (armv8.1-a, armv8.2-a): Likewise.
26297 (armv8-m.main): add options fp, fp.dp and nofp.
26299 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26301 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26303 (armv8-a+crc): Delete.
26304 (armv8.1-a): Add options simd, crypto and nofp.
26305 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26306 (armv8.2-a+fp16): Delete.
26307 (armv8-m.main): Add option dsp.
26308 (armv8-m.main+dsp): Delete.
26309 (cortex-a8): Add fpu. Add nofp option.
26310 (cortex-a9): Add fpu. Add nofp and nosimd options.
26311 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26312 link to main cpu and architecture data structures.
26313 (gen_comm_data): Only put isa attributes from the main architecture
26315 (option): New statement for architecture and CPU entries.
26316 * arm.c (struct cpu_option): New structure.
26317 (struct processors): Add entry for options.
26318 (arm_unrecognized_feature): New function.
26319 (arm_parse_arch_cpu_name): Ignore any characters after the first
26321 (arm_parse_arch_cpu_feature): New function.
26322 (arm_configure_build_target): Separate out any CPU and architecture
26323 features and parse separately. Don't error out if -mfpu=auto is
26324 used with only an architecture string.
26325 (arm_print_asm_arch_directives): New function.
26326 (arm_file_start): Call it.
26327 * config/arm/arm-cpu-cdata.h: Regenerated.
26328 * config/arm/arm-cpu-data.h: Likewise.
26329 * config/arm/arm-tables.opt: Likewise.
26331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26333 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26334 assembler when it is not -mfpu=auto.
26336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26338 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26339 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26340 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26341 (ASM_CPU_SPEC): Rewrite.
26342 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26343 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26344 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26345 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26346 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26347 copied string is NUL-terminated. Also strip any characters prefixed
26349 (arm_rewrite_selected_arch): New function.
26350 (arm_rewrite_march): New function.
26352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26354 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26355 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26356 (march, mcpu, mtune): Convert to string-based options.
26357 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26358 (arm_parse_arch_cpu_name): New function.
26359 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26360 identify selected architecture or CPU.
26361 (arm_option_save): New function.
26362 (TARGET_OPTION_SAVE): Redefine.
26363 (arm_option_restore): Restore string options.
26364 (arm_option_print): Print string options.
26366 2017-06-16 Martin Sebor <msebor@redhat.com>
26368 PR tree-optimization/80933
26369 PR tree-optimization/80934
26370 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26371 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26372 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26373 (gimple_fold_builtin): Call them.
26375 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26377 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26378 as unlikely; update profile.
26380 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26382 * predict.c (force_edge_cold): Handle declaring edges impossible
26385 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26387 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26389 (try_unroll_loop_completely): Fix reporting.
26391 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26393 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26395 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26398 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26399 if given a non-constant argument for an intrinsic which requires a
26402 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26404 * profile.c (compare_freqs): New function.
26405 (branch_prob): Sort edge list.
26406 (find_spanning_tree): Assume that the list is priority sorted.
26408 2017-06-16 Richard Biener <rguenther@suse.de>
26410 PR tree-optimization/81090
26411 * passes.def (pass_record_bounds): Remove.
26412 * tree-pass.h (make_pass_record_bounds): Likewise.
26413 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26414 make_pass_record_bounds): Likewise.
26415 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26416 not free niter estimates at the beginning but at the end.
26417 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26419 2017-06-16 Richard Biener <rguenther@suse.de>
26421 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26422 initializer to workaround ICE in host GCC 4.8.
26424 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26426 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26428 (clone_inlined_nodes): Update.
26430 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26432 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26433 prefetch settings, and enable prefetching by default at -O3.
26435 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26437 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26438 Set flag_prefetch_loop_arrays according to tuning data.
26440 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26442 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26443 New tune structure.
26444 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26445 [Unrelated to main purpose of the patch] Place the pointer field last
26446 to enable type checking errors when tune structure are wrongly merged.
26447 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26448 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26449 (thunderx2t99_prefetch_tune): New tune constants.
26450 (tune_params *_tunings): Update all tunings (no functional change).
26451 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26452 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26453 from tunings structures.
26455 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26458 * ubsan.c (instrument_null): Add T argument, use it instead
26459 of computing it based on IS_LHS.
26460 (instrument_object_size): Likewise.
26461 (pass_ubsan::execute): Adjust instrument_null and
26462 instrument_object_size callers to pass gimple_get_lhs or
26463 gimple_assign_rhs1 result to it. Use instrument_null instead of
26464 calling get_base_address and instrument_mem_ref. Handle
26465 aggregate call arguments for object-size sanitization.
26467 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26469 PR tree-optimization/81089
26470 * tree-vrp.c (is_masked_range_test): Validate operands of
26473 2017-06-15 Martin Sebor <msebor@redhat.com>
26476 * dumpfile.c (dump_register): Avoid calling memset to initialize
26477 a class with a default ctor.
26478 * gcc.c (struct compiler): Remove const qualification.
26479 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26480 * hash-table.h: Ditto.
26481 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26483 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26484 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26486 * params.h (struct param_info): Make struct members non-const.
26487 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26488 with default initialization.
26489 * vec.h (vec_copy_construct, vec_default_construct): New helper
26491 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26492 with vec_copy_construct.
26493 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26494 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26495 * doc/invoke.texi (-Wclass-memaccess): Document.
26497 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26499 * emit-rtl.h (is_leaf): Update comment about local
26500 register allocator.
26502 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26505 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26506 for a variable to have a section before checking if the section has a
26508 Set section to.persistent if persistent attribute is set.
26509 Warn if .persistent attribute is used on an automatic variable.
26511 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26513 PR rtl-optimization/80474
26514 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26516 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26518 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26521 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26523 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26524 (casesi): Emit bounds checking as RTL.
26525 (casesi_internal_mips16_<mode>): Remove bounds checking.
26527 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26529 * config/xtensa/xtensa.c (xtensa_option_override): Append
26530 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26531 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26532 xtensa_doloop_hooks): Define unconditionally.
26533 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26535 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26536 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26537 for it in xtensa_option_override.
26538 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26539 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26541 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26543 * doc/cppopts.texi: Document '-' special value to -MF.
26545 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26547 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26548 (cortex_a53_fconst): Likewise.
26549 (cortex_a53_fpmul): Likewise.
26550 (cortex_a53_f_load_64): Likewise.
26551 (cortex_a53_f_load_many): Likewise.
26552 (cortex_a53_advsimd_alu): Likewise.
26553 (cortex_a53_advsimd_alu_q): Likewise.
26554 (cortex_a53_advsimd_mul): Likewise.
26555 (cortex_a53_advsimd_mul_q): Likewise.
26556 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26557 Add missing fmul, r2f_cvt and fconst cases.
26559 2017-06-14 Richard Biener <rguenther@suse.de>
26561 PR middle-end/81088
26562 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26564 (fold_binary_loc): When associating do not treat pre-existing
26565 TREE_OVERFLOW on literal constants as a reason to allow
26566 TREE_OVERFLOW on associated literal constants.
26568 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26570 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26571 (MASK_FEATURES): New macro.
26572 * config/sparc/sparc.c (sparc_option_override): Remove the special
26573 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26575 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26577 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26578 a division of 0 if non-call exceptions are enabled.
26580 2017-06-14 Andrew Pinski <apinski@cavium.com>
26581 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26584 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26585 Improve vector initialization code gen for only variable case.
26587 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26589 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26591 2017-06-14 Richard Biener <rguenther@suse.de>
26593 PR tree-optimization/81083
26594 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26597 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26599 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26600 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26601 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26602 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26603 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26604 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26606 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26608 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26609 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26611 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26613 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26615 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26617 * config/rs6000/t-rtems: Don't handle SPE.
26619 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26621 * config/rs6000/t-linux: Don't handle SPE.
26623 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26625 * config/rs6000/eabispe.h: Delete file.
26627 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26629 * config/rs6000/t-spe: Delete file.
26631 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26633 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26634 (rs6000_legitimate_offset_address_p): Return false for anything in
26635 V2SImode or V2SFmode.
26637 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26639 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26640 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26641 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26643 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26644 (rs6000_legitimate_offset_address_p): Ditto.
26645 (rs6000_emit_move): Ditto.
26646 (rs6000_init_builtins): Remove V4HI_type_node.
26648 2017-06-13 Martin Liska <mliska@suse.cz>
26651 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26652 (gate_asan): Likewise.
26653 * asan.h (asan_no_sanitize_address_p): Remove the function.
26654 (sanitize_flags_p): New function.
26655 * builtins.def: Fix coding style.
26656 * common.opt: Use renamed enum value.
26657 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26658 * doc/extend.texi: Document no_sanitize attribute.
26659 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26660 to SANITIZE_UNDEFINED_NONDEFAULT.
26661 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26662 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26663 Use sanitize_flags_p.
26664 * gimplify.c (gimplify_function_tree): Likewise.
26665 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26666 * opts.c (parse_no_sanitize_attribute): New function.
26667 (common_handle_option): Use renamed enum value.
26668 * opts.h (parse_no_sanitize_attribute): Declare.
26669 * tree.c (sanitize_flags_p): New function.
26670 * tree.h: Declared here.
26671 * tsan.c: Use sanitize_flags_p.
26672 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26673 (instrument_mem_ref): Likewise.
26674 (instrument_bool_enum_load): Likewise.
26675 (do_ubsan_in_current_function): Remove the function.
26676 (pass_ubsan::execute): Use sanitize_flags_p.
26677 * ubsan.h: Remove do_ubsan_in_current_function
26678 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26679 (dump_function_to_file): Use it here.
26681 2017-06-13 Martin Jambor <mjambor@suse.cz>
26683 PR tree-optimization/80803
26684 PR tree-optimization/81063
26685 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26686 (propagate_subaccesses_across_link): Enqueue subtree whenever
26687 necessary instead of relying on the caller.
26689 2017-06-13 Martin Jambor <mjambor@suse.cz>
26691 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26692 that have a first_link.
26693 (sort_and_splice_var_accesses): Do not check first_link before
26695 (subtree_mark_written_and_enqueue): Likewise.
26696 (propagate_all_subaccesses): Likewise and do not stop at first
26697 parent with a first_link.
26699 2017-06-13 Martin Jambor <mjambor@suse.cz>
26701 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26704 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26706 * match.pd: New pattern.
26708 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26710 * tree-vrp.c (is_masked_range_test): New function.
26711 (register_edge_assert_for): Determine ranges for
26714 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26716 PR tree-optimization/67328
26717 * fold-const.c (maskable_range_p): New function.
26718 (build_range_check): Generate bittests if possible.
26720 2017-06-13 Martin Liska <mliska@suse.cz>
26722 * gimple-pretty-print.c (dump_probability): Add new argument.
26723 (dump_edge_probability): Dump both probability and count.
26724 (dump_gimple_label): Likewise.
26725 (dump_gimple_bb_header): Likewise.
26727 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26730 * config/avr/avr-devices.c: Fix indentation.
26731 * config/avr/gen-avr-mmcu-specs.c: Dito.
26733 2017-06-13 Richard Biener <rguenther@suse.de>
26735 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26736 instead get vector type from stmt_info.
26737 (vectorizable_reduction): Adjust. Remove dead code.
26739 2017-06-13 Richard Biener <rguenther@suse.de>
26741 PR middle-end/81065
26742 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26743 case of C * (x * C2 + C3).
26744 (fold_addr_of_array_ref_difference): Properly fold index difference.
26746 2017-06-12 David S. Miller <davem@davemloft.net>
26749 * config/sparc/sparc.md (return expander): Emit frame blockage if
26750 function uses alloca.
26752 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26754 * combine.c (make_field_assignment): Check len rather than the mode
26755 precision when calling force_to_mode.
26757 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26759 Support multilibs and devices that see flash in RAM address range.
26762 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26763 (avr_mcu_t) <flash_pm_offset>: New field.
26764 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26765 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26766 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26767 (AVR_TINY_PM_OFFSET): Remove macro.
26768 * config/avr/avr.opt (-mshort-calls): New option.
26769 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26770 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26771 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26772 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26773 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26774 instead of avr_arch->have_jmp_call.
26775 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26776 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26777 avr_arch->flash_pm_offset to define.
26778 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26779 new field flash_pm_offset. Add entry for avrxmega3.
26780 (avr_texinfo): Add entry for avrxmega3.
26781 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26782 attiny212, attiny214,
26783 attiny412, attiny414, attiny416, attiny417,
26784 attiny814, attiny816, attiny817,
26785 attiny1614, attiny1616, attiny1617,
26786 attiny3214, attiny3216, attiny3217.
26787 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26788 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26789 (avr_print_operand_address) [AVR_TINY]: Same.
26790 (avr_asm_init_sections) <readonly_data_section>: Only patch
26791 callback if avr_arch->flash_pm_offset = 0.
26792 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26793 for rodata if avr_arch->flash_pm_offset != 0.
26794 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26795 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26796 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26797 (m_options): Append opt_rcall.
26798 (m_dirnames): Append dir_rcall.
26799 * config/avr/t-multilib: Regenerate.
26801 * configure.ac [target=avr]: Check whether avrxmega3 default
26802 linker description file works as needed.
26803 * configure: Regenerate.
26804 * doc/avr-mmcu.texi: Regenerate.
26805 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26806 <__AVR_ARCH__>: Document avrxmega3 and 103.
26807 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26808 <__AVR_SHORT_CALLS__>: Document it.
26809 <__AVR_PM_BASE_ADDRESS__>: Document it.
26810 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26811 (AVR Variable Attributes) <progmem>: Document this is
26812 not needed for avrxmega3.
26813 (AVR Named Address Spaces) <__flash>: Dito.
26815 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26817 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26819 2017-06-12 Doug Rupp <rupp@adacore.com>
26821 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26822 Append vxworks-stdint.h to the tm_file list.
26823 * config/vxworks-stdint.h: New file.
26825 2017-06-12 Martin Liska <mliska@suse.cz>
26827 PR tree-optimization/81041
26828 * tree-profile.c (gimple_gen_ic_func_profiler):
26829 Create an extra BB in profile-generate
26830 (gimple_gen_time_profiler): Likewise.
26832 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26834 PR tree-optimization/81003
26835 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26836 (update_range_test): Use it instead of force_gimple_operand_gsi.
26838 2017-06-12 Richard Biener <rguenther@suse.de>
26840 PR tree-optimization/81053
26841 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26842 with backedge value not defined in loop. Simplify def stmt
26845 2017-06-11 Tom de Vries <tom@codesourcery.com>
26848 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26850 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26851 nvptx_cannot_force_const_mem.
26853 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26855 * opts.c (finish_options): Move test for flag_split_stack after
26856 it has been initialized.
26858 2017-06-11 Jason Merrill <jason@redhat.com>
26860 * tree.h (id_equal): New.
26861 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26862 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26863 instead of strcmp of IDENTIFIER_POINTER.
26865 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26867 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26868 (mark_all_inlined_calls_cdtor): Fix formating.
26869 (inline_transform): Rescale profile before inlining.
26871 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26873 * cgraph.h (cgraph_edge::clone): Update prototype.
26874 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26875 (cgraph_node::create_clone): Update.
26876 (cgraph_node::create_version_clone): Update.
26877 * tree-inline.c (copy_bb): Update.
26878 (expand_call_inline): Update.
26880 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26882 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26883 factored out from ...
26884 (rs6000_emit_prologue): ... here.
26886 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26888 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26889 factored out from ...
26890 (rs6000_emit_prologue): ... here.
26892 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26894 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26896 (handle_missing_profiles): Fix computation of tp_first_run.
26897 (counts_to_freqs): Do not touch freqs when count is 0.
26899 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26901 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26904 2017-06-10 Tom de Vries <tom@codesourcery.com>
26906 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26907 attributes): Document signal effective target.
26909 2017-06-10 Tom de Vries <tom@codesourcery.com>
26911 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26912 Document effective target stack_size.
26914 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26916 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26917 to the edit_context if they can be auto-applied.
26919 2017-06-9 Ian Lance Taylor <iant@golang.org>
26921 * opts.c (finish_options): If -fsplit-stack, disable implicit
26922 -forder-blocks-and-partition.
26923 * doc/invoke.texi (Optimize Options): Document that when using
26924 -fsplit-stack -forder-blocks-and-partition is not implicitly
26927 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26929 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26930 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26931 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26932 * builtins.def (abort, trap, unreachable): Declare cold.
26933 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26934 * tree-core.h (ECF_COLD): New.
26935 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26936 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26938 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26940 * predict.c (unlikely_executed_stmt_p): Cleanup.
26942 2017-06-09 Richard Biener <rguenther@suse.de>
26944 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26945 model if the ref is always written to.
26947 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26949 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26951 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26953 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26956 2017-06-09 Tom de Vries <tom@codesourcery.com>
26959 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26960 "target cannot support label values" when encountering LABEL_REF.
26962 2017-06-09 Martin Liska <mliska@suse.cz>
26964 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26965 (gimple_gen_ic_func_profiler): Emit direct comparison
26966 of __gcov_indirect_call_callee with NULL.
26967 (gimple_gen_time_profiler): Change probability from
26968 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26970 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26972 * profile.c (edge_gcov_counts): Turn to pointer.
26973 (compute_branch_probabilities, compute_branch_probabilities): Update.
26974 (branch_prob): Do not clear edge_gcov_count.
26975 * profile.h (edge_gcov_counts): Turn to pointer.
26976 (edge_gcov_count): Update.
26978 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26980 * gimple.h (gimple_check_failed): Mark cold.
26982 2017-06-09 Richard Biener <rguenther@suse.de>
26984 PR tree-optimization/66623
26985 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26986 refactor check_reduction into two parts, properly computing
26987 whether we have to check reduction validity for outer loop
26990 2017-06-09 Richard Biener <rguenther@suse.de>
26992 PR tree-optimization/79483
26993 * graphite-scop-detection.c (order): New global.
26994 (get_order): Compute bb to order mapping that satisfies code
26995 generation constraints.
26996 (cmp_pbbs): New helper.
26997 (build_scops): Start domwalk at entry block, sort generated
27000 2017-06-09 Richard Biener <rguenther@suse.de>
27002 PR middle-end/81007
27003 * ipa-polymorphic-call.c
27004 (ipa_polymorphic_call_context::restrict_to_inner_class):
27005 Skip FIELD_DECLs with error_mark_node type.
27006 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27009 2017-06-09 Martin Liska <mliska@suse.cz>
27011 * predict.c (struct branch_predictor): New struct.
27012 (test_prediction_value_range): New test.
27013 (predict_c_tests): New function.
27014 * selftest-run-tests.c (selftest::run_tests): Run the function.
27015 * selftest.h: Declare new tests.
27017 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27020 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27021 gen_add3_insn did not fail.
27022 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27023 r0, construct that number in a temporary reg and add that reg to r0.
27024 If asked to put the result in r0 as well, fail.
27026 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27028 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27029 for early expansion of vec_eqv.
27031 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27033 PR middle-end/81005
27034 * ubsan.c (instrument_null): Avoid pointless code temporary.
27035 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27037 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27041 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27044 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27047 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27049 * predict.c (unlikely_executed_edge_p): Move ahead.
27050 (probably_never_executed_edge_p): Use it.
27052 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27054 PR middle-end/79988
27055 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27056 gimple_call_builtin_p call.
27058 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27060 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27061 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27062 rtl_check_failed_type2, rtl_check_failed_code1,
27063 rtl_check_failed_code2, rtl_check_failed_code_mode,
27064 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27065 rtvec_check_failed_bounds, rtl_check_failed_flag,
27066 _fatal_insn_not_found, _fatal_insn): Likewise.
27067 * tree.h (tree_contains_struct_check_failed,
27068 tree_check_failed, tree_not_check_failed,
27069 tree_class_check_failed, tree_range_check_failed,
27070 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27071 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27072 tree_operand_check_failed, omp_clause_check_failed,
27073 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27076 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27078 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27079 flag_branch_probabilities.
27080 * ipa-inline.c (edge_badness): Likewise.
27081 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27082 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27083 * predict.c (maybe_hot_frequency_p): Likewise.
27084 (probably_never_executed): Likewise.
27085 * sched-ebb.c (schedule_ebbs): Likewise.
27086 * sched-rgn.c (find_single_block_region): Likewise.
27087 * tracer.c (tail_duplicate): Likewise.
27089 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27091 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27092 longer requires x_flag_profile_use.
27094 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27096 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27097 instead of flag_reorder_blocks_and_partition.
27098 * dbxout.c (dbxout_function_end): Likewise.
27099 * dwarf2out.c (gen_subprogram_die): Likewise.
27100 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27101 * hw-doloop.c (reorg_loops): Likewise.
27102 * varasm.c (assemble_start_function,
27103 assemble_end_function): Likewise.
27104 (decide_function_section): Do not check for
27105 flag_reorder_blocks_and_partition.
27107 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27109 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27111 (chkp_get_hard_register_fake_addr_expr): Ditto.
27112 (chkp_build_addr_expr): Add check for hard reg case.
27113 (chkp_parse_array_and_component_ref): Ditto.
27114 (chkp_find_bounds_1): Ditto.
27115 (chkp_process_stmt): Don't generate bounds store for
27118 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27120 * predict.c (maybe_hot_bb_p): Do not check profile status.
27121 (maybe_hot_edge_p): Likewise.
27122 (probably_never_executed): Check for zero counts even if profile
27124 (unlikely_executed_edge_p): New function.
27125 (unlikely_executed_stmt_p): New function.
27126 (unlikely_executed_bb_p): New function.
27127 (set_even_probabilities): Use unlikely predicates.
27128 (combine_predictions_for_bb): Likewise.
27129 (predict_paths_for_bb): Likewise.
27130 (predict_paths_leading_to_edge): Likewise.
27131 (determine_unlikely_bbs): New function.
27132 (estimate_bb_frequencies): Use it.
27133 (compute_function_frequency): Use zero counts even if profile is
27135 * profile-count.h: Fix typo.
27137 2017-08-08 Julia Koval <julia.koval@intel.com>
27139 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27140 _mm512_mask_cvtsepi16_storeu_epi8,
27141 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27142 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27143 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27144 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27145 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27146 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27147 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27148 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27149 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27150 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27151 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27152 __builtin_ia32_pmovuswb256mem_mask,
27153 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27154 __builtin_ia32_pmovwb512mem_mask): New builtins.
27156 2017-08-08 Julia Koval <julia.koval@intel.com>
27158 PR target/73350,80862
27159 * config/i386/subst.md (round): Fix round pattern.
27160 * config/i386/i386.c (ix86_erase_embedded_rounding):
27161 Fix erasing rounding for the fixed pattern.
27163 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27165 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27167 2017-06-08 Martin Liska <mliska@suse.cz>
27169 PR gcov-profile/80911
27170 * gcov.c (block_info::block_info): New constructor.
27172 2017-06-07 Carl Love <cel@us.ibm.com>
27174 * config/rs6000/rs6000-c: The return type of the following
27175 built-in functions was implemented as int not long long. Fix sign
27176 of return value for the unsigned version of vec_mulo and vec_mule.
27177 vector unsigned long long vec_bperm (vector unsigned long long,
27178 vector unsigned char)
27179 vector signed long long vec_mule (vector signed int,
27181 vector unsigned long long vec_mule (vector unsigned int,
27182 vector unsigned int)
27183 vector signed long long vec_mulo (vector signed int,
27185 vector unsigned long long vec_mulo (vector unsigned int,
27186 vector unsigned int)
27187 * doc/extend.texi: Fix the documentation for the built-in
27190 2017-06-07 Carl Love <cel@us.ibm.com>
27193 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27196 2017-06-07 Carl Love <cel@us.ibm.com>
27198 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27199 support, Generate doublehv for signed int/float for BE case only.
27201 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27203 * doc/invoke.texi (mcx16): Rewrite.
27205 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27207 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27208 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27209 *mov<mode>_softfloat, and an anonymous splitter): Use
27210 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27212 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27214 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27216 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27217 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27218 (rs6000_debug_reg_global): Adjust.
27219 (rs6000_init_hard_regno_mode_ok): Adjust.
27220 (rs6000_dbx_register_number): Adjust.
27221 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27222 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27223 Remove SPE_ACC and SPEFSCR.
27224 (REG_ALLOC_ORDER): Ditto.
27225 (FRAME_POINTER_REGNUM): Change to 111.
27226 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27227 (REG_CLASS_NAMES): Ditto.
27228 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27229 (REGISTER_NAMES): Ditto.
27230 (ADDITIONAL_REG_NAMES): Ditto.
27231 (rs6000_reg_names): Ditto.
27232 * config/rs6000/rs6000.md: Renumber some register number
27235 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27237 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27239 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27240 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27242 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27243 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27244 Delete the SPE high registers.
27245 (REG_ALLOC_ORDER): Ditto.
27246 (enum reg_class): Remove SPE_HIGH_REGS.
27247 (REG_CLASS_NAMES): Ditto.
27248 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27249 (REGISTER_NAMES): Ditto.
27250 (rs6000_reg_names): Ditto.
27251 * doc/tm.texi.in: Remove SPE as example.
27252 * doc/tm.texi: Regenerate.
27254 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27256 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27257 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27258 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27259 * config/rs6000/rs6000.md (type): Remove "brinc".
27261 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27263 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27264 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27265 * config/rs6000/linuxspe.h: Delete file.
27266 * config/rs6000/rs6000.md: Don't include spe.md.
27267 * config/rs6000/spe.h: Delete file.
27268 * config/rs6000/spe.md: Delete file.
27269 * config/rs6000/t-rs6000: Remove spe.md.
27271 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27273 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27274 (reg_or_none500mem_operand): Delete.
27275 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27276 instead of reg_or_none500mem_operand.
27278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27280 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27281 handling of SPE flags.
27282 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27286 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27288 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27289 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27290 paired_divv2sf3): Similar.
27291 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27292 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27293 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27295 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27296 Rename the paired_* instruction patterns.
27297 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27299 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27300 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27301 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27302 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27303 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27304 PAIRED_VECTOR_MODE.
27305 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27306 (spe_func_has_64bit_regs_p): Delete.
27307 (spe_expand_predicate_builtin): Delete.
27308 (spe_expand_evsel_builtin): Delete.
27309 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27310 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27311 (invalid_e500_subreg): Delete.
27312 (rs6000_legitimize_address): Always force_reg op2 as well, for
27313 paired single memory accesses.
27314 (rs6000_member_type_forces_blk): Delete.
27315 (rs6000_spe_function_arg): Delete.
27316 (rs6000_expand_unop_builtin): Delete SPE handling.
27317 (rs6000_expand_binop_builtin): Ditto.
27318 (spe_expand_stv_builtin): Delete.
27319 (bdesc_2arg_spe): Delete.
27320 (spe_expand_builtin): Delete.
27321 (spe_expand_predicate_builtin): Delete.
27322 (spe_expand_evsel_builtin): Delete.
27323 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27324 (spe_init_builtins): Delete.
27325 (spe_func_has_64bit_regs_p): Delete.
27326 (savres_routine_name): Delete "info" parameter. Adjust callers.
27327 (rs6000_emit_stack_reset): Ditto.
27328 (rs6000_dwarf_register_span): Delete.
27329 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27330 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27331 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27333 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27335 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27336 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27337 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27338 mulv2sf3, divv2sf3): Delete expanders.
27340 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27342 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27344 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27346 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27347 * config/rs6000/rs6000.c: Ditto.
27349 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27351 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27352 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27353 comparison_operator.
27355 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27357 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27358 * config/rs6000/rs6000.opt: Ditto.
27359 * config/rs6000/t-rtems: Ditto.
27361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27363 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27364 TARGET_E500_SINGLE by 0, simplify.
27365 * config/rs6000/rs6000.c: Ditto.
27366 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27367 (spe_build_register_parallel): Delete.
27368 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27369 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27370 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27371 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27372 (E500_CONVERT): Delete.
27373 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27375 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27377 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27378 * config/rs6000/dfp.md: Ditto.
27379 (negdd2, *negdd2_fpr): Merge.
27380 (absdd2, *absdd2_fpr): Merge.
27381 (negtd2, *negtd2_fpr): Merge.
27382 (abstd2, *abstd2_fpr): Merge.
27383 * config/rs6000/e500.h: Delete file.
27384 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27385 TARGET_FPRS by 1 and simplify.
27386 * config/rs6000/rs6000-c.c: Ditto.
27387 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27388 TARGET_DF_SPE by 0.
27389 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27391 * config/rs6000/rs6000.md: Ditto.
27392 (floatdidf2, *floatdidf2_fpr): Merge.
27393 (move_from_CR_gt_bit): Delete.
27394 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27395 (E500_CR_IOR_COMPARE): Delete.
27396 (All patterns that require !TARGET_FPRS): Delete.
27397 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27399 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27401 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27403 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27405 * graphds.c (add_edge): Intitialize edge's attached data.
27406 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27407 pointer parameter. Call pointed function on each edge during
27408 graph traversing. Skip traversing the edge when the function
27410 (graphds_dfs, graphds_scc): Ditto.
27411 (for_each_edge): New parameter. Pass the new parameter to callback
27413 * graphds.h (skip_edge_callback): New function pointer type.
27414 (graphds_dfs, graphds_scc): New function pointer parameter.
27415 (graphds_edge_callback, for_each_edge): New parameter.
27417 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27419 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27420 out code checking if runtime alias check is possible to below ...
27421 Call the new function.
27422 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27423 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27425 2017-06-07 Marek Polacek <polacek@redhat.com>
27428 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27429 TYPE_OVERFLOW_WRAPS check.
27431 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27433 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27434 if versioning is required.
27435 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27436 peeling with the check for versioning.
27438 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27440 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27441 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27442 Set true to new parameter if new ssa variable is defined.
27443 (vect_gen_vector_loop_niters): Refactor. Set range information
27444 for the new vector loop bound variable.
27445 (vect_do_peeling): Ditto.
27447 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27449 * tree-affine.c (ssa.h): Include header file.
27450 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27451 has wrapping overflow behavior.
27453 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27455 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27457 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27459 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27460 (tree_to_aff_combination): ... here.
27462 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27464 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27465 reg_pressure model function.
27466 (ivopts_global_cost_for_size): Delete.
27467 (determine_set_costs, iv_ca_recount_cost): Call new model function
27468 ivopts_estimate_reg_pressure.
27470 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27472 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27473 expensive than udiv. Remove floating point cases from mod.
27475 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27477 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27478 Increase idiv cost.
27480 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27482 * config/aarch64/aarch64.md
27483 (copysignsf3): Fix mask generation.
27485 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27487 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27489 (class dump_manager): Add register_dumps method.
27490 * dumpfile.c: Include langhooks.h.
27491 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27492 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27493 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27494 (dump_manager::dump_register): Start with 512 entries instead of 32.
27495 (dump_manager::register_dumps): New method.
27496 * toplev.c (general_init): Instead of invoking register_dumps
27497 langhook, invoke register_dumps method on the dump manager.
27498 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27501 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27503 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27504 Say that # requires an associated define_split to exist, and that
27505 the define_split must be suitable for use after register allocation.
27507 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27509 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27510 (compute_outgoing_frequencies): Also initialize zero counts.
27511 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27512 around loops; preserve more of profile when nothing changes.
27514 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27516 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27518 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27519 * config/arm/arm-cpu-cdata.h: Regenerate.
27520 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27521 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27522 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27523 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27524 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27526 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27528 * config/arm/t-rmprofile: Likewise.
27529 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27531 2017-06-06 David S. Miller <davem@davemloft.net>
27534 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27535 blockage if function uses alloca.
27537 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27539 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27540 New "uid" fields to hold pretty-print IDs of group and ref.
27541 Memory references are now identified as <group_id>:<ref_id>
27542 instead of using [random] addresses.
27543 (dump_mem_details): Simplify, no functional change.
27544 (dump_mem_ref): Simplify and make output more concise.
27545 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27546 (find_or_create_group): Initialize group uid.
27547 (record_ref): Initialize ref uid. Improve debug output.
27548 (prune_group_by_reuse, should_issue_prefetch_p,)
27549 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27550 (mark_nontemporal_store, determine_loop_nest_reuse):
27551 Improve debug output.
27553 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27555 * dbgcnt.def (prefetch): New debug counter.
27556 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27557 (schedule_prefetches): Stop issueing prefetches if debug counter
27560 2017-06-06 Tom de Vries <tom@codesourcery.com>
27562 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27563 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27565 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27567 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27568 Use aarch64_reg_or_zero predicate for operand 4.
27569 (aarch64_compare_and_swap<mode> define_insn_and_split):
27570 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27571 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27573 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27575 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27576 (arm_compute_save_core_reg_mask): This.
27577 (thumb1_compute_save_reg_mask): Rename into ...
27578 (thumb1_compute_save_core_reg_mask): This.
27579 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27580 (arm_compute_frame_layout): Likewise.
27582 2017-06-06 Richard Biener <rguenther@suse.de>
27584 PR tree-optimization/80974
27585 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27586 keep or clear leaders SSA info.
27588 2017-06-06 Tom de Vries <tom@codesourcery.com>
27590 * config/nvptx/nvptx.c (split_mode_p): New function.
27591 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27593 2017-06-06 Tom de Vries <tom@codesourcery.com>
27595 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27597 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27600 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27603 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27605 * shrink-wrap.c (handle_simple_exit): Update profile.
27606 (try_shrink_wrapping): Upate profile.
27608 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27610 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27611 (tree_guess_outgoing_edge_probabilities): New.
27612 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27613 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27615 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27617 * ipa-split.c (split_function): Initialize return bb profile.
27619 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27621 * profile.c (compute_branch_probabilities): Also initialize
27622 EXIT_BLOCK profile.
27624 2017-06-06 Richard Biener <rguenther@suse.de>
27626 PR tree-optimization/80928
27627 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27628 (vect_analyze_loop_operations): Properly guard analysis for
27630 (vect_transform_loop): Likewise.
27631 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27632 (vect_model_induction_cost): Do not cost for pure SLP.
27633 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27634 of induction in inner loop vectorization.
27635 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27636 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27637 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27639 (vect_analyze_slp_cost_1): Cost induction.
27640 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27641 (vect_get_slp_vect_defs): Likewise.
27642 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27643 (vect_transform_stmt): Handle SLP reductions.
27644 * tree-vectorizer.h (vectorizable_induction): Adjust.
27646 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27648 * config/rs6000/rs6000.c (make_resolver_func): Update
27649 init_lowered_empty_function call.
27651 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27653 * doc/invoke.texi: Document the -fprofile-abs-path option.
27654 * common.opt (fprofile-abs-path): New option.
27655 * gcov-io.h (gcov_write_filename): Declare.
27656 * gcov-io.c (gcov_write_filename): New function.
27657 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27658 * profile.c (output_location): Likewise.
27660 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27662 * shring-wrap.c: Revert accidental commit.
27664 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27666 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27668 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27670 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27672 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27673 profile in callgraph edge.
27674 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27675 (apply_scale): Likewise.
27676 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27677 Also scale profile when inlining function with zero profile.
27678 (initialize_cfun): Update exit block profile even when it is zero.
27679 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27680 when profile is read.
27682 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27684 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27685 (CLONE_*): New constants to define the processors we can generate
27686 code for with the target_clone attribute.
27687 (rs6000_clone_map): New array to identify which clone processors
27688 the current program is running on.
27689 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27690 target_clone attribute.
27691 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27692 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27693 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27694 (cpu_expand_builtin): Add support for target_clone attribute.
27695 (rs6000_valid_attribute_p): Allow "default" attribute.
27696 (get_decl_name): New debug function to simplify printing the
27697 current function name in debugging statements.
27698 (rs6000_clone_priority): New functions to support the target_clone
27699 attribute, and be able to generate code to switch between ISA 2.05
27700 through ISA 3.0 (power6 through power9).
27701 (rs6000_compare_version_priority): Likewise.
27702 (rs6000_get_function_versions_dispatcher): Likewise.
27703 (make_resolver_func): Likewise.
27704 (add_condition_to_bb): Likewise.
27705 (dispatch_function_versions): Likewise.
27706 (rs6000_generate_version_dispatcher_body): Likewise.
27707 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27708 (fusion_gpr_load_p): Fix a spacing issue.
27709 * doc/extend.texi (Common Function Attributes): Document that the
27710 PowerPC supports the target_clone attribute.
27712 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27714 * config/arm/arm.h: explain F symbol found in description of ARM
27715 register allocation in its legend.
27717 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27719 * config/mips/frame-header-opt.c: Include profile-count.h.
27720 * config/riscv/riscv.c: Include profile-count.h
27722 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27724 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27726 (sm_set_flag_if_changed): Add bbs field.
27727 (execute_sm_if_changed_flag_set): Pass BBS.
27728 (execute_sm): Update.
27730 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27732 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27735 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27737 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27738 (peephole2): New peephole2 to emit the above.
27739 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27741 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27743 * config/aarch64/aarch64.c (define_peephole2 above
27744 *sub_<shift>_<mode>): New peephole.
27746 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27748 * config/i386/i386.c (make_resolver_func): Update.
27749 * Makefile.in: Add profile-count.h and profile-count.o
27750 * auto-profile.c (afdo_indirect_call): Update to new API.
27751 (afdo_set_bb_count): Update.
27752 (afdo_propagate_edge): Update.
27753 (afdo_propagate_circuit): Update.
27754 (afdo_calculate_branch_prob): Update.
27755 (afdo_annotate_cfg): Update.
27756 * basic-block.h: Include profile-count.h
27757 (struct edge_def): Turn count to profile_count.
27758 (struct basic_block_def): Likewie.
27759 (REG_BR_PROB_BASE): Move to profile-count.h
27760 (RDIV): Move to profile-count.h
27761 * bb-reorder.c (max_entry_count): Turn to profile_count.
27762 (find_traces): Update.
27763 (rotate_loop):Update.
27764 (connect_traces):Update.
27765 (sanitize_hot_paths):Update.
27766 * bt-load.c (migrate_btr_defs): Update.
27767 * cfg.c (RDIV): Remove.
27768 (init_flow): Use alloc_block.
27769 (alloc_block): Uninitialize count.
27770 (unchecked_make_edge): Uninitialize count.
27771 (check_bb_profile): Update.
27772 (dump_edge_info): Update.
27773 (dump_bb_info): Update.
27774 (update_bb_profile_for_threading): Update.
27775 (scale_bbs_frequencies_int): Update.
27776 (scale_bbs_frequencies_gcov_type): Update.
27777 (scale_bbs_frequencies_profile_count): New.
27778 * cfg.h (update_bb_profile_for_threading): Update.
27779 (scale_bbs_frequencies_profile_count): Declare.
27780 * cfgbuild.c (compute_outgoing_frequencies): Update.
27781 (find_many_sub_basic_blocks): Update.
27782 * cfgcleanup.c (try_forward_edges): Update.
27783 (try_crossjump_to_edge): Update.
27784 * cfgexpand.c (expand_gimple_tailcall): Update.
27785 (construct_exit_block): Update.
27786 * cfghooks.c (verify_flow_info): Update.
27787 (dump_bb_for_graph): Update.
27788 (split_edge): Update.
27789 (make_forwarder_block): Update.
27790 (duplicate_block): Update.
27791 (account_profile_record): Update.
27792 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27793 (get_estimated_loop_iterations): Update.
27794 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27795 (single_likely_exit): Update.
27796 * cfgloopmanip.c (scale_loop_profile): Update.
27798 (set_zero_probability): Update.
27799 (lv_adjust_loop_entry_edge): Update.
27800 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27801 (purge_dead_edges): Update.
27802 (rtl_account_profile_record): Update.
27803 * cgraph.c (cgraph_node::create): Uninitialize count.
27804 (symbol_table::create_edge): Uninitialize count.
27805 (cgraph_update_edges_for_call_stmt_node): Update.
27806 (cgraph_edge::dump_edge_flags): Update.
27807 (cgraph_node::dump): Update.
27808 (cgraph_edge::maybe_hot_p): Update.
27809 * cgraph.h: Include profile-count.h
27810 (create_clone), create_edge, create_indirect_edge): Update.
27811 (cgraph_node): Turn count to profile_count.
27812 (cgraph_edge0: Likewise.
27813 (make_speculative, clone): Update.
27814 (create_edge): Update.
27815 (init_lowered_empty_function): Update.
27816 * cgraphclones.c (cgraph_edge::clone): Update.
27817 (duplicate_thunk_for_node): Update.
27818 (cgraph_node::create_clone): Update.
27819 * cgraphunit.c (cgraph_node::analyze): Update.
27820 (cgraph_node::expand_thunk): Update.
27821 * final.c (dump_basic_block_info): Update.
27822 * gimple-streamer-in.c (input_bb): Update.
27823 * gimple-streamer-out.c (output_bb): Update.
27824 * graphite.c (print_global_statistics): Update.
27825 (print_graphite_scop_statistics): Update.
27826 * hsa-brig.c: Include basic-block.h.
27827 * hsa-dump.c: Include basic-block.h.
27828 * hsa-gen.c (T sum_slice): Update.
27829 (convert_switch_statements):Update.
27830 * hsa-regalloc.c: Include basic-block.h.
27831 * ipa-chkp.c (chkp_produce_thunks): Update.
27832 * ipa-cp.c (struct caller_statistics): Update.
27833 (init_caller_stats): Update.
27834 (gather_caller_stats): Update.
27835 (ipcp_cloning_candidate_p): Update.
27836 (good_cloning_opportunity_p): Update.
27837 (get_info_about_necessary_edges): Update.
27838 (dump_profile_updates): Update.
27839 (update_profiling_info): Update.
27840 (update_specialized_profile): Update.
27841 (perhaps_add_new_callers): Update.
27842 (decide_about_value): Update.
27843 (ipa_cp_c_finalize): Update.
27844 * ipa-devirt.c (struct odr_type_warn_count): Update.
27845 (struct decl_warn_count): Update.
27846 (struct final_warning_record): Update.
27847 (possible_polymorphic_call_targets): Update.
27848 (ipa_devirt): Update.
27849 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27850 * ipa-icf.c (sem_function::merge): Update.
27851 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27852 * ipa-inline.c (compute_uninlined_call_time): Update.
27853 (compute_inlined_call_time): Update.
27854 (want_inline_small_function_p): Update.
27855 (want_inline_self_recursive_call_p): Update.
27856 (edge_badness): Update.
27857 (lookup_recursive_calls): Update.
27858 (recursive_inlining): Update.
27859 (inline_small_functions): Update.
27860 (dump_overall_stats): Update.
27861 (dump_inline_stats): Update.
27862 * ipa-profile.c (ipa_profile_generate_summary): Update.
27863 (ipa_propagate_frequency): Update.
27864 (ipa_profile): Update.
27865 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27866 * ipa-utils.c (ipa_merge_profiles): Update.
27867 * loop-doloop.c (doloop_modify): Update.
27868 * loop-unroll.c (report_unroll): Update.
27869 (unroll_loop_runtime_iterations): Update.
27870 * lto-cgraph.c (lto_output_edge): Update.
27871 (lto_output_node): Update.
27872 (input_node): Update.
27873 (input_edge): Update.
27874 (merge_profile_summaries): Update.
27875 * lto-streamer-in.c (input_cfg): Update.
27876 * lto-streamer-out.c (output_cfg): Update.
27877 * mcf.c (create_fixup_graph): Update.
27878 (adjust_cfg_counts): Update.
27879 (sum_edge_counts): Update.
27880 * modulo-sched.c (sms_schedule): Update.
27881 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27882 * predict.c (maybe_hot_count_p): Update.
27883 (probably_never_executed): Update.
27884 (dump_prediction): Update.
27885 (combine_predictions_for_bb): Update.
27886 (propagate_freq): Update.
27887 (handle_missing_profiles): Update.
27888 (counts_to_freqs): Update.
27889 (rebuild_frequencies): Update.
27890 (force_edge_cold): Update.
27891 * predict.h: Include profile-count.h
27892 (maybe_hot_count_p, counts_to_freqs): UPdate.
27893 * print-rtl-function.c: Do not include cfg.h
27894 * print-rtl.c: Include basic-block.h
27895 * profile-count.c: New file.
27896 * profile-count.h: New file.
27897 * profile.c (is_edge_inconsistent): Update.
27898 (correct_negative_edge_counts): Update.
27899 (is_inconsistent): Update.
27900 (set_bb_counts): Update.
27901 (read_profile_edge_counts): Update.
27902 (compute_frequency_overlap): Update.
27903 (compute_branch_probabilities): Update; Initialize and deinitialize
27905 (branch_prob): Update.
27906 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27907 (edge_gcov_count): New.
27908 (bb_gcov_count): New.
27909 * shrink-wrap.c (try_shrink_wrapping): Update.
27910 * tracer.c (better_p): Update.
27911 * trans-mem.c (expand_transaction): Update.
27912 (ipa_tm_insert_irr_call): Update.
27913 (ipa_tm_insert_gettmclone_call): Update.
27914 * tree-call-cdce.c: Update.
27915 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27916 (gimple_duplicate_sese_tail): Update.
27917 (gimple_account_profile_record): Update.
27918 (execute_fixup_cfg): Update.
27919 * tree-inline.c (copy_bb): Update.
27920 (copy_edges_for_bb): Update.
27921 (initialize_cfun): Update.
27922 (freqs_to_counts): Update.
27923 (copy_cfg_body): Update.
27924 (expand_call_inline): Update.
27925 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27926 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27927 (try_unroll_loop_completely): Update.
27928 (try_peel_loop): Update.
27929 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27930 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27931 * tree-ssa-loop-split.c (connect_loops): Update.
27932 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27933 * tree-ssa-reassoc.c (branch_fixup): Update.
27934 * tree-ssa-tail-merge.c (replace_block_by): Update.
27935 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27936 (compute_path_counts): Update.
27937 (update_profile): Update.
27938 (recompute_probabilities): Update.
27939 (update_joiner_offpath_counts): Update.
27940 (estimated_freqs_path): Update.
27941 (freqs_to_counts_path): Update.
27942 (clear_counts_path): Update.
27943 (ssa_fix_duplicate_block_edges): Update.
27944 (duplicate_thread_path): Update.
27945 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27946 (struct switch_conv_info): Update.
27947 * tree-tailcall.c (decrease_profile): Update.
27948 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27949 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27950 * value-prof.c (check_counter): Update.
27951 (gimple_divmod_fixed_value): Update.
27952 (gimple_mod_pow2): Update.
27953 (gimple_mod_subtract): Update.
27954 (gimple_ic_transform): Update.
27955 (gimple_stringop_fixed_value): Update.
27956 * value-prof.h (gimple_ic): Update.
27958 2017-06-02 Carl Love <cel@us.ibm.com>
27960 * config/rs6000/rs6000-c: Add support for built-in functions
27961 vector double vec_doublee (vector signed int);
27962 vector double vec_doublee (vector unsigned int);
27963 vector double vec_doublee (vector float);
27964 vector double vec_doubleh (vector signed int);
27965 vector double vec_doubleh (vector unsigned int);
27966 vector double vec_doubleh (vector float);
27967 vector double vec_doublel (vector signed int);
27968 vector double vec_doublel (vector unsigned int);
27969 vector double vec_doublel (vector float);
27970 vector double vec_doubleo (vector signed int);
27971 vector double vec_doubleo (vector unsigned int);
27972 vector double vec_doubleo (vector float);.
27973 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27974 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27976 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27977 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27978 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27980 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27981 vec_doublel, vec_doubleh.
27982 * doc/extend.texi: Update the built-in documentation file for the
27983 new built-in functions.
27985 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27988 * ipa-inline-analysis.c (free_growth_caches): Set
27989 edge_removal_hook_holder to NULL after removing it.
27991 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27993 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27994 comparision with zero.
27996 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27997 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27998 for early expansion of vec_min and vec_max builtins.
27999 (builtin_function_type): Add min/max unsigned variants to those
28000 identified as having unsigned arguments.
28002 2017-06-02 Olivier Hainque <hainque@adacore.com>
28004 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28006 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28008 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28009 Use VALL_F16 iterator rather than VALL.
28011 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28013 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28014 Emit CBNZ inside loop when doing a strong exchange and comparing
28015 against zero. Generate the CC flags after the loop.
28017 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28019 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28020 (dl_section_ref): New.
28021 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28022 On AIX, append an expression to subtract the size of the
28023 section length to dl_section_ref.
28025 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28027 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28028 for early expansion of vector absolute builtins.
28030 2017-06-02 Richard Biener <rguenther@suse.de>
28032 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28033 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28035 2017-06-02 Richard Biener <rguenther@suse.de>
28037 PR tree-optimization/80948
28038 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28039 stmt order as well.
28041 2017-06-02 Richard Biener <rguenther@suse.de>
28043 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28045 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28046 for inductions as relevant.
28048 2017-06-02 Richard Biener <rguenther@suse.de>
28050 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28051 (vectorizable_induction): ... this. Remove dead code.
28053 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28055 * builtins. (expand_builtin_alloca): Remove second parameter and
28056 infer its value from the first parameter instead.
28057 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28059 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28061 PR rtl-optimization/80903
28062 * loop-doloop.c (add_test): Unshare sequence.
28064 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28066 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28068 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28070 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28072 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28073 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28074 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28075 (xlogue_layout::get_stub_rtx): Make static.
28076 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28077 (xlogue_layout::compute_stub_managed_regs): Rename to...
28078 (xlogue_layout::count_stub_managed_regs): ...this.
28079 (xlogue_layout::is_stub_managed_reg): New function.
28080 (xlogue_layout::m_stub_names): Rename to...
28081 (xlogue_layout::s_stub_names): ...this, make static.
28082 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28083 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28084 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28085 xlogue_layout::s_stub_names): Instantiate statics.
28086 (stub_managed_regs): Remove.
28087 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28088 (disable_call_ms2sysv_xlogues): Rename to...
28089 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28090 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28092 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28093 change after reload_completed.
28094 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28096 (ix86_expand_prologue): Likewise.
28097 (ix86_expand_epilogue): Likewise.
28098 (ix86_expand_split_stack_prologue): Likewise.
28099 (ix86_compute_frame_layout): Remove frame parameter ...
28100 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28101 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28103 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28104 (ix86_frame): Move from here ...
28105 * config/i386/i386.h (ix86_frame): ... to here.
28106 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28107 complete ix86_frame data structure instead. Remove some_ld_name.
28109 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28111 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28112 symbols that hold a DECL_VALUE_EXPR.
28114 2017-06-01 Martin Jambor <mjambor@suse.cz>
28116 PR tree-optimization/80898
28117 * tree-sra.c (process_subtree_disqualification): Removed.
28118 (disqualify_candidate): Do not acll
28119 process_subtree_disqualification.
28120 (subtree_mark_written_and_enqueue): New function.
28121 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28122 RHS has been disqualified and re-queue LHS if necessary. Apart
28123 from that, ignore disqualified RHS.
28125 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28127 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28128 address fetch for z10 or later.
28130 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28132 * config/arc/arc.md (tst_movb): Add guard when splitting.
28134 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28136 * config/arc/arc.c (arc_can_eliminate): Test against
28137 arc_frame_pointer_needed.
28139 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28141 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28142 to prevent store reordering.
28143 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28144 (type): Add block type.
28145 (stack_tie): Define special instruction to be used in
28148 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28150 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28151 constraint. It is not valid for the pattern.
28152 (noncommutative_binary_comparison): Likewise.
28154 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28156 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28159 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28161 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28162 be used by the reg-alloc.
28164 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28166 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28167 reg-alloc when having mul64 or mul32x16 instructions.
28168 (mulsidi3): Likewise.
28169 (umulsidi3): Likewise.
28170 (mulsi32x16): New pattern.
28171 (mulsi64): Likewise.
28172 (mulsidi64): Likewise.
28173 (umulsidi64): Likewise.
28174 (MUL32x16_REG): Define.
28175 (mul64_600): Use MUL32x16_REG.
28176 (mac64_600): Likewise.
28177 (umul64_600): Likewise.
28178 (umac64_600): Likewise.
28180 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28182 * config/arc/arc.md (mulsi3_700): Make it commutative.
28184 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28186 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28188 (*sign_extendsidi2_insn): Likewise for movstosw.
28190 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28192 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28193 the type of the input discriminant value. Convert the
28194 discriminant value of signedness vary.
28196 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28198 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28199 Add to -Wall section.
28201 2017-06-01 Richard Biener <rguenther@suse.de>
28203 PR middle-end/66313
28204 * fold-const.c (fold_plusminus_mult_expr): If the factored
28205 factor may be zero use a wrapping type for the inner operation.
28206 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28207 and handle moved defs.
28208 (process_assignment): Properly guard the unary op case. Return a
28209 tri-state indicating that moving the stmt before the call may allow
28210 to continue. Pass through to_move.
28211 (find_tail_calls): Handle moving unrelated defs before
28214 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28217 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28218 splitter result in the canonical way.
28220 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28222 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28223 also for 32bit target. Update insn attributes.
28224 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28226 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28228 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28229 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28230 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28231 (_mm_maskz_min_ss): New intrinsics.
28233 2017-05-31 Martin Liska <mliska@suse.cz>
28235 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28236 Change comment style to one we normally use.
28237 (vectorizable_reduction): Likewise.
28238 (vectorizable_induction): Likewise.
28239 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28240 (vectorizable_call): Likewise.
28241 (vectorizable_simd_clone_call): Likewise.
28242 (vectorizable_conversion): Likewise.
28243 (vectorizable_assignment): Likewise.
28244 (vectorizable_shift): Likewise.
28245 (vectorizable_operation): Likewise.
28246 (vectorizable_store): Likewise.
28247 (vectorizable_load): Likewise.
28248 * tree-vectorizer.h: Likewise.
28250 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28252 * passes.c (emergency_dump_function): New.
28253 * tree-pass.h (emergency_dump_function): Declare.
28254 * plugin.c (plugins_internal_error_function): Remove.
28255 * plugin.h (plugins_internal_error_function): Remove declaration.
28256 * toplev.c (internal_error_function): New static function. Use it...
28257 (general_init): ...here.
28259 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28261 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28262 (arc_rtx_costs): Add costs for new patterns.
28263 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28264 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28266 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28268 * tree-ssa-strlen.c (get_next_strinfo): New function.
28269 (get_stridx_plus_constant): Use it.
28270 (zero_length_string): Likewise.
28271 (adjust_related_strinfos): Likewise.
28272 (adjust_last_stmt): Likewise.
28274 2017-05-31 Richard Biener <rguenther@suse.de>
28277 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28278 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28280 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28282 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28283 loop_vinfo argument and use of dependence distance vectors.
28284 Check instead whether the two references differ only in their
28285 initial value and assume that they have the same alignment if the
28286 difference is a multiple of the vector alignment.
28287 (vect_analyze_data_refs_alignment): Update call accordingly.
28289 2017-05-31 Martin Liska <mliska@suse.cz>
28292 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28294 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28296 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28297 (create_intersect_range_checks): Move from ...
28298 * tree-data-ref.c (create_intersect_range_checks_index)
28299 (create_intersect_range_checks): ... to here.
28300 (create_runtime_alias_checks): New function factored from ...
28301 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28302 here. Call above function.
28303 * tree-data-ref.h (create_runtime_alias_checks): New function.
28305 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28307 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28308 segment length for dr_b and compute it in wide_int.
28310 2017-05-31 Richard Biener <rguenther@suse.de>
28312 PR tree-optimization/80906
28313 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28314 and pass through iv_map.
28315 (copy_bb_and_scalar_dependences): Adjust.
28316 (translate_pending_phi_nodes): Likewise.
28317 (copy_loop_close_phi_args): Handle code-generating IVs instead
28320 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28322 * diagnostic-color.c (color_dict): Add "type-diff".
28323 (parse_gcc_colors): Update comment.
28324 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28325 -fdiagnostics-show-template-tree and -fno-elide-type.
28326 (GCC_COLORS): Add type-diff to example.
28328 (-fdiagnostics-show-template-tree): New.
28329 (-fno-elide-type): New.
28330 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28331 the pp_format_decoder callback. Call any m_format_postprocessor's
28333 (pretty_printer::pretty_printer): Initialize
28334 m_format_postprocessor.
28335 (pretty_printer::~pretty_printer): Delete any
28336 m_format_postprocessor.
28337 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28338 (class format_postprocessor): New class.
28339 (struct pretty_printer::format_decoder): Document the new parameters.
28340 (struct pretty_printer::m_format_postprocessor): New field.
28341 * tree-diagnostic.c (default_tree_printer): Update for new
28342 bool and const char ** params.
28343 * tree-diagnostic.h (default_tree_printer): Likewise.
28345 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28347 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28348 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28349 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28350 rs6000_gen_cell_microcode code.
28351 (rs6000_final_prescan_insn): Delete.
28352 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28353 "warn-cell-microcode" entries.
28354 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28355 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28356 throughout. Change cc_reg_not_micro_cr0_operand to
28357 cc_reg_not_cr0_operand throughout.
28358 (*extendhi<mode>2_noload): Delete.
28359 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28360 (mwarn-cell-microcode): Delete.
28361 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28362 -mgen-cell-microcode and -mwarn-cell-microcode.
28364 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28367 * config/i386/constraints.md (Yd): New constraint.
28369 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28370 and (?Yd, r) alternatives. Update insn attributes.
28371 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28372 and (?*Yd, r) alternatives. Update insn attributes.
28373 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28375 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28377 * gimplify.c (gimplify_modify_expr): Don't create a
28378 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28381 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28383 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28385 2017-05-30 Richard Biener <rguenther@suse.de>
28387 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28388 and reduc_def fields.
28389 (STMT_VINFO_REDUC_TYPE): New define.
28390 (STMT_VINFO_REDUC_DEF): Likewise.
28391 (vect_force_simple_reduction): Adjust prototype.
28392 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28393 (vect_is_simple_reduction): Remove check_reduction argument.
28394 (vect_force_simple_reduction): Adjust and set
28395 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28396 (vectorizable_reduction): Do not re-do reduction analysis
28397 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28398 * tree-parloops.c (gather_scalar_reductions): Adjust.
28400 2017-05-30 Richard Biener <rguenther@suse.de>
28402 PR middle-end/80901
28403 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28406 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28408 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28409 Introduce unknown_misalignment parameter and remove vf.
28410 (vect_peeling_hash_get_lowest_cost):
28411 Pass unknown_misalignment parameter.
28412 (vect_enhance_data_refs_alignment):
28413 Fix unsupportable data ref treatment.
28415 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28417 * tree-vect-data-refs.c (vect_get_data_access_cost):
28418 Workaround for SLP handling.
28419 (vect_enhance_data_refs_alignment):
28420 Compute costs for doing no peeling at all, compare to the best
28421 peeling costs so far and avoid peeling if cheaper.
28423 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28425 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28426 Return peeling info and set costs to zero for unlimited cost
28428 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28429 with unknown misalignment. Compute and costs for unknown
28430 misalignment, compare them to the costs for known misalignment
28431 and choose the cheapest for peeling.
28433 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28435 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28436 (vect_get_peeling_costs_all_drs): Create function.
28437 (vect_peeling_hash_get_lowest_cost):
28438 Use vect_get_peeling_costs_all_drs.
28439 (vect_peeling_supportable): Create function.
28440 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28442 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28444 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28445 DR_HAS_NEGATIVE_STEP.
28446 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28447 (vect_enhance_data_refs_alignment): Use.
28448 (vect_duplicate_ssa_name_ptr_info): Use.
28449 * tree-vectorizer.h (dr_misalignment): Use.
28450 (known_alignment_for_access_p): Use.
28452 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28455 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28457 (has_section_name): New function.
28459 2017-05-30 Martin Liska <mliska@suse.cz>
28462 * auto-profile.c (get_function_decl_from_block): Fix
28465 2017-05-30 Richard Biener <rguenther@suse.de>
28467 PR middle-end/80876
28468 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28470 2017-05-30 Martin Liska <mliska@suse.cz>
28472 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28473 * dumpfile.h (struct dump_file_info): Remove ctors.
28475 2017-05-30 Martin Liska <mliska@suse.cz>
28477 * predict.def: Fix GNU coding style.
28479 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28481 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28482 Mark 'to' argument with ATTRIBUTE_UNUSED.
28484 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28486 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28487 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28488 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28491 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28493 * doc/install.texi (Options specification): Restore entry of
28494 --enable-sjlj-exceptions.
28496 2017-05-27 Michael Eager <eager@eagercon.com>
28499 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28501 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28503 * config/microblaze/microblaze.h
28504 (FIXED_REGISTERS): Update in macro.
28505 (CALL_USED_REGISTERS): Update in macro.
28507 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28509 * doc/install.texi: Add links to macOS binary distributions.
28511 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28515 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28517 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28519 2017-05-26 Martin Liska <mliska@suse.cz>
28521 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28523 2017-05-26 Martin Liska <mliska@suse.cz>
28525 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28526 always leading ';; '.
28527 (dump_bb_info): Likewise.
28528 (brief_dump_cfg): Likewise.
28529 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28530 * dumpfile.c: Remove usage of TDF_VERBOSE.
28531 * dumpfile.h (enum dump_kind): Likewise.
28532 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28533 * print-tree.c (debug_verbose): Remove.
28534 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28535 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28536 * tree-diagnostic.c (default_tree_printer): Replace
28537 TDF_DIAGNOSTIC with TDF_SLIM.
28539 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28541 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28542 in parameter loop, rather than loop_vinfo.
28543 (create_intersect_range_checks): Ditto.
28544 (vect_create_cond_for_alias_checks): Update call to above functions.
28546 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28548 PR tree-optimization/80815
28549 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28550 for merging runtime alias checks. Handle negative DR_STEPs.
28552 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28554 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28556 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28557 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28558 out code pruning runtime alias checks.
28559 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28560 factored out from above.
28561 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28563 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28565 (prune_runtime_alias_test_list): New decalaration.
28567 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28569 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28570 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28571 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28572 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28573 (operator==, comp_dr_with_seg_len_pair): Ditto.
28574 (vect_prune_runtime_alias_test_list): Ditto.
28576 2017-05-26 Martin Liska <mliska@suse.cz>
28579 * params.def: Bound partial-inlining-entry-probability param.
28581 2017-05-26 Marek Polacek <polacek@redhat.com>
28584 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28587 2017-05-26 Richard Biener <rguenther@suse.de>
28589 PR tree-optimization/80842
28590 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28593 2017-05-26 Richard Biener <rguenther@suse.de>
28595 PR tree-optimization/80844
28596 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28598 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28600 * doc/md.texi (Machine Constraints): Update x86 family
28601 machine constraints section to match 'config/i386/constraints.md'.
28603 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28605 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28607 2017-05-25 Nathan Sidwell <nathan@acm.org>
28609 * doc/invoke.texi (--enable-languages): Update documentation.
28611 2017-05-25 Martin Liska <mliska@suse.cz>
28613 * dumpfile.c: Add TDF_FOLDING.
28614 * dumpfile.h (enum dump_kind): Likewise.
28615 * genmatch.c (dt_simplify::gen_1): Use it.
28617 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28619 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28621 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28623 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28624 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28626 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28628 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28629 * match.pd (X == C): Rewrite it here.
28630 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28631 with_certain_nonzero_bits2): New predicates.
28632 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28634 2017-05-24 Nathan Sidwell <nathan@acm.org>
28636 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28639 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28642 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28644 * config/powerpcspe: New port. Files are copied from the rs6000
28645 port, with "rs6000" in filenames replaced by "powerpcspe".
28647 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28649 PR rtl-optimization/80754
28650 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28652 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28654 * config/sparc/sparc.md (length): Return the correct value for -mflat
28655 sibcalls to match output_sibcall.
28657 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28661 * config/rs6000/rs6000.c (struct machine_function): Add new field
28663 (rs6000_get_separate_components): Init that field, use it.
28664 (rs6000_components_for_bb): Use the field.
28666 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28668 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28670 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28672 PR middle-end/80823
28673 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28675 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28678 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28679 address against address_operand predicate.
28680 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28682 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28684 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28685 parameters passed indirectly.
28687 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28689 * config/i386/i386.md (*movdi_internal): Remove SSE4
28690 alternative 18 (?r, *v). Update insn attributes.
28691 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28692 Update insn attributes.
28693 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28694 Update insn attributes.
28695 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28696 alternative 1 (r, v). Remove isa attribute.
28697 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28698 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28699 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28701 2017-05-23 Tom de Vries <tom@codesourcery.com>
28703 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28706 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28708 * cgraphunit.c (symbol_table::process_new_functions): Update.
28709 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28710 (inline_generate_summary): Rename to ...
28711 (ipa_fn_summary_generate): ... this one.
28712 (inline_read_summary): Rename to ...
28713 (ipa_fn_summary_read): ... this one.
28714 (inline_write_summary): Rename to ...
28715 (ipa_fn_summary_write): ... this one.
28716 (inline_free_summary): Rename to ...
28717 (ipa_free_fn_summary): ... this one.
28718 (pass_data_local_fn_summary, pass_local_fn_summary,
28719 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28720 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28721 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28722 make_pass_ipa_fn_summary): New.
28723 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28724 inline_write_summary, inline_free_summary): Remove.
28725 (ipa_free_fn_summary) : New.
28726 * ipa-inline.c (ipa_inline): Update.
28727 (pass_ipa_inline): Do not generate summaries.
28728 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28730 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28731 and add pass_ipa_fn_summary.
28732 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28734 (make_pass_inline_parameters): Remove.
28736 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28738 * omp-low.c (struct omp_context): Remove "default_kind" member.
28741 * omp-offload.c (execute_oacc_device_lower): Remove the
28742 parallelism dimensions function attributes for unparallelized
28743 OpenACC kernels constructs.
28745 2017-05-23 Martin Liska <mliska@suse.cz>
28747 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28749 (cgraph_edge::make_speculative): Likewise.
28750 (cgraph_edge::resolve_speculation): Likewise.
28751 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28752 (cgraph_node::dump): Likewise.
28753 * cgraph.h: Likewise.
28754 * cgraphunit.c (analyze_functions): Likewise.
28755 (symbol_table::compile): Likewise.
28756 * ipa-cp.c (print_all_lattices): Likewise.
28757 (determine_versionability): Likewise.
28758 (initialize_node_lattices): Likewise.
28759 (ipcp_verify_propagated_values): Likewise.
28760 (estimate_local_effects): Likewise.
28761 (update_profiling_info): Likewise.
28762 (create_specialized_node): Likewise.
28763 (perhaps_add_new_callers): Likewise.
28764 (decide_about_value): Likewise.
28765 (decide_whether_version_node): Likewise.
28766 (identify_dead_nodes): Likewise.
28767 (ipcp_store_bits_results): Likewise.
28768 * ipa-devirt.c (dump_targets): Likewise.
28769 (ipa_devirt): Likewise.
28770 * ipa-icf.c (sem_item::dump): Likewise.
28771 (sem_function::equals): Likewise.
28772 (sem_variable::equals): Likewise.
28773 (sem_item_optimizer::read_section): Likewise.
28774 (sem_item_optimizer::execute): Likewise.
28775 (congruence_class::dump): Likewise.
28776 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28777 (dump_inline_summary): Likewise.
28778 (estimate_node_size_and_time): Likewise.
28779 (inline_analyze_function): Likewise.
28780 * ipa-inline-transform.c (inline_call): Likewise.
28781 * ipa-inline.c (report_inline_failed_reason): Likewise.
28782 (want_early_inline_function_p): Likewise.
28783 (edge_badness): Likewise.
28784 (update_edge_key): Likewise.
28785 (inline_small_functions): Likewise.
28786 * ipa-profile.c (ipa_profile): Likewise.
28787 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28788 (ipa_make_edge_direct_to_target): Likewise.
28789 (remove_described_reference): Likewise.
28790 (ipa_impossible_devirt_target): Likewise.
28791 (propagate_controlled_uses): Likewise.
28792 (ipa_print_node_params): Likewise.
28793 (ipcp_transform_function): Likewise.
28794 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28795 (propagate_pure_const): Likewise.
28796 * ipa-reference.c (generate_summary): Likewise.
28797 (read_write_all_from_decl): Likewise.
28798 (propagate): Likewise.
28799 (ipa_reference_read_optimization_summary): Likewise.
28800 * ipa-utils.c (ipa_merge_profiles): Likewise.
28801 * ipa.c (walk_polymorphic_call_targets): Likewise.
28802 (symbol_table::remove_unreachable_nodes): Likewise.
28803 (ipa_single_use): Likewise.
28804 * passes.c (execute_todo): Likewise.
28805 * predict.c (drop_profile): Likewise.
28806 * symtab.c (symtab_node::get_dump_name): New function.
28807 (symtab_node::dump_name): Likewise.
28808 (symtab_node::dump_asm_name): Likewise.
28809 (symtab_node::dump_references): Likewise.
28810 (symtab_node::dump_referring): Likewise.
28811 (symtab_node::dump_base): Likewise.
28812 (symtab_node::debug_symtab): Likewise.
28813 * tree-sra.c (convert_callers_for_node): Likewise.
28814 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28815 * value-prof.c (init_node_map): Likewise.
28817 2017-05-23 Martin Liska <mliska@suse.cz>
28819 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28820 and symtab_node::debug_symtab to symbol_table::debug.
28821 * cgraphunit.c (analyze_functions): Use the renamed function.
28822 (symbol_table::compile): Likewise.
28823 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28824 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28825 * passes.c (execute_todo): Likewise.
28826 * symtab.c (symbol_table::dump): New function.
28827 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28829 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28831 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28832 that nonconst implies exec.
28834 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28836 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28837 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28838 (inline_edge_summary_vec): Turn into ...
28839 (ipa_call_summaries): ... this one.
28840 (redirect_to_unreachable, edge_set_predicate,
28841 evaluate_properties_for_edge, inline_summary_alloc,
28842 reset_ipa_call_summary, reset_inline_summary,
28843 inline_summary_t::duplicate): Update.
28844 (inline_edge_duplication_hook): Turn to ...
28845 (ipa_call_summary_t::duplicate): ... this one.
28846 (inline_edge_removal_hook): Turn to ...
28847 (ipa_call_summary_t::remove): ... this one.
28848 (dump_inline_edge_summary): Turn to ...
28849 (dump_ipa_call_summary): ... this one.
28850 (estimate_function_body_sizes): Update.
28851 (inline_update_callee_summaries): Update.
28852 (remap_edge_change_prob): Update.
28853 (remap_edge_summaries): Update.
28854 (inline_merge_summary): Update.
28855 (do_estimate_edge_time): Update.
28856 (inline_generate_summary): Update.
28857 (inline_read_section): Update.
28858 (inline_read_summary): Update.
28859 (inline_free_summary): Update.
28860 * ipa-inline.c (can_inline_edge_p): Update.
28861 (compute_inlined_call_time): Update.
28862 (want_inline_small_function_p): Update.
28863 (edge_badness): Update.
28864 (early_inliner): Update.
28865 * ipa-inline.h (inline_edge_summary): Turn to ...
28866 (ipa_call_summary): ... this one.
28867 (ipa_call_summary_t): New class.
28868 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28869 (ipa_call_summaries): New.
28870 (inline_edge_summary): Remove.
28871 (estimate_edge_growth): Update.
28872 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28873 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28874 * ipa-split.c (execute_split_functions): Update.
28875 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28877 2017-05-23 Tom de Vries <tom@codesourcery.com>
28879 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28880 attributes): Document rdrand effective target.
28882 2017-05-23 Tom de Vries <tom@codesourcery.com>
28884 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28885 attributes): Sort alphabetically.
28887 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28889 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28891 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28894 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28895 V2DF/V2DI splat into two separate patterns, one that handles
28896 registers, and the other that only handles memory. Drop support
28897 for splatting from a GPR on ISA 2.07 and then splitting the
28898 splat into direct move and splat.
28899 (vsx_splat_<mode>_reg): Likewise.
28900 (vsx_splat_<mode>_mem): Likewise.
28902 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28904 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28906 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28908 PR middle-end/80809
28909 * omp-low.c (finish_taskreg_remap): New function.
28910 (finish_taskreg_scan): If unit size of ctx->record_type
28911 is non-constant, unshare the size expression and replace
28912 decls in it with possible outer var refs.
28914 PR middle-end/80809
28915 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28916 GOVD_SHARED rather than GOVD_PRIVATE with it.
28917 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28918 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28920 PR middle-end/80853
28921 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28922 as last argument to build_outer_var_ref for pointer bases of array
28923 section reductions.
28925 2017-05-19 Martin Sebor <msebor@redhat.com>
28927 * print-tree.c (print_node): Print DECL_READ_P flag.
28929 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28931 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28932 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28933 * cgraph.c: Likewise.
28934 * cgraphunit.c: Likewise.
28935 * gengtype.c: Likewise.
28936 * ipa-cp.c: Likewise.
28937 * ipa-devirt.c: Likewise.
28938 * ipa-icf.c: Likewise.
28939 * ipa-predicate.c: Likewise.
28940 * ipa-profile.c: Likewise.
28941 * ipa-prop.c: Likewise.
28942 * ipa-split.c: Likewise.
28944 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28945 edge_predicate_pool, dump_inline_hints,
28946 inline_summary::account_size_time, redirect_to_unreachable,
28947 edge_set_predicate, set_hint_predicate,
28948 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28949 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28950 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28951 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28952 ipa_call_summary_t::remove, initialize_growth_caches,
28953 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28954 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28955 mark_modified, unmodified_parm_1, unmodified_parm,
28956 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28957 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28958 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28959 will_be_nonconstant_predicate, record_modified_bb_info,
28960 get_minimal_bb, record_modified, param_change_prob,
28961 phi_result_unknown_predicate, predicate_for_phi_result,
28962 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28963 estimate_function_body_sizes, compute_inline_parameters,
28964 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28965 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28966 inline_update_callee_summaries, remap_edge_change_prob,
28967 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28968 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28969 inline_analyze_function, inline_summary_t::insert,
28970 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28971 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28972 inline_free_summary): Move to ipa-fnsummary.h
28973 (predicate_t): Remove.
28974 * ipa-fnsummary.c: New file.
28975 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28976 (enum inline_hints_vals, inline_hints, agg_position_info,
28977 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28978 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28979 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28980 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28981 inline_read_summary, inline_write_summary, inline_free_summary,
28982 inline_analyze_function, initialize_inline_failed,
28983 inline_merge_summary, inline_update_overall_summary,
28984 compute_inline_parameters): Move to ipa-fnsummary.h
28985 * ipa-fnsummary.h: New file.
28986 * ipa-inline-transform.h: Include ipa-inline.h.
28987 * ipa-inline.c: LIkewise.
28989 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28991 * ipa-inline.c (edge_badness): Use inlined_time instead of
28992 inline_summaries->get.
28994 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28996 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28998 2017-05-22 Nathan Sidwell <nathan@acm.org>
29000 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29001 (fdump-lang): Document 'raw' option.
29002 * dumpfile.h (TDI_tu): Delete.
29003 * dumpfile.c (dump_files): Remove translation-unit.
29004 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29006 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29008 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29009 command option from $(AWK) call.
29010 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29011 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29012 [FORMAT]: Remove handling of variable.
29013 * config/avr/t-multilib: Regenerate.
29015 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29017 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29019 (dump_inline_summary): Do not print self_time.
29020 (estimate_function_body_sizes): Do not set self_time.
29021 (compute_inline_parameters): Likewise.
29022 (inline_read_section, inline_write_summary): Do not stream self_time.
29023 * ipa-inline.h (inline_summary): Drop self_time.
29025 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29027 * ipa-inline-analysis.c (account_size_time): Rename to ...
29028 (inline_summary::account_size_time): ... this one.
29029 (reset_ipa_call_summary): Turn to ...
29030 (ipa_call_summary::reset): ... this one.
29031 (reset_inline_summary): Turn to ...
29032 (inline_summary::reset): ... this one.
29033 (inline_summary_t::remove): Update.
29034 (inline_summary_t::duplicate): Update.
29035 (ipa_call_summary_t::remove): Update.
29036 (dump_inline_summary): Update.
29037 (estimate_function_body_sizes): Update.
29038 (compute_inline_parameters): Update.
29039 (estimate_node_size_and_time): Update.
29040 (inline_merge_summary): Update.
29041 (inline_update_overall_summary): Update.
29042 (inline_read_section): Update.
29043 (inline_write_summary): Update.
29044 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29045 add account_size_time and reset member functions.
29046 (ipa_call_summary): Add reset function.
29047 * ipa-predicate.h (predicate::operator &): Constify.
29049 2017-05-22 Richard Biener <rguenther@suse.de>
29051 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29053 2017-05-19 Jason Merrill <jason@redhat.com>
29055 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29057 2017-05-19 Marek Polacek <polacek@redhat.com>
29060 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29061 TYPE_OVERFLOW_WRAPS checks.
29063 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29065 * tree-core.h (enum omp_clause_default_kind): Add
29066 "OMP_CLAUSE_DEFAULT_PRESENT".
29067 * tree-pretty-print.c (dump_omp_clause): Handle it.
29068 * gimplify.c (enum gimplify_omp_var_data): Add
29069 "GOVD_MAP_FORCE_PRESENT".
29070 (gimplify_adjust_omp_clauses_1): Map it to
29071 "GOMP_MAP_FORCE_PRESENT".
29072 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29074 * gimplify.c (oacc_default_clause): Clarify.
29076 2017-05-19 Nathan Sidwell <nathan@acm.org>
29078 LANG_HOOK_REGISTER_DUMPS
29079 * toplev.c (general_init): Call register dump lang hook.
29080 * doc/invoke.texi: Document -fdump-lang option family.
29081 * dumpfile.c (dump_files): Remove class dump here.
29082 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29083 * dumpfile.h (tree_dump_index): Remove TDI_class.
29084 * langhooks-def.h (lhd_register_dumps): Declare.
29085 (LANG_HOOKS_REGISTER_DUMPS): Define.
29086 (LANG_HOOKS_INITIALIZER): Add it.
29087 * langhooks.c (lhd_register_dumps): Define.
29088 * langhooks.h (struct lang_hooks): Add register_dumps.
29090 2017-05-19 Nathan Sidwell <nathan@acm.org>
29092 * context.h (context::set_passes): New.
29093 * context.c (context::context): Do not create pass manager.
29094 * toplev.c (general_init): Create pass manager here.
29096 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29098 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29099 use this splitter if two add or or instructions would also work for
29100 the constant we want to generate.
29102 2017-05-19 Richard Biener <rguenther@suse.de>
29105 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29106 predicate evaluation.
29108 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29110 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29112 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29115 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29117 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29118 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29119 (inline_edge_summary_vec): Turn into ...
29120 (ipa_call_summaries): ... this one.
29121 (redirect_to_unreachable, edge_set_predicate,
29122 evaluate_properties_for_edge, inline_summary_alloc,
29123 reset_ipa_call_summary, reset_inline_summary,
29124 inline_summary_t::duplicate): Update.
29125 (inline_edge_duplication_hook): Turn to ...
29126 (ipa_call_summary_t::duplicate): ... this one.
29127 (inline_edge_removal_hook): Turn to ...
29128 (ipa_call_summary_t::remove): ... this one.
29129 (dump_inline_edge_summary): Turn to ...
29130 (dump_ipa_call_summary): ... this one.
29131 (estimate_function_body_sizes): Update.
29132 (inline_update_callee_summaries): Update.
29133 (remap_edge_change_prob): Update.
29134 (remap_edge_summaries): Update.
29135 (inline_merge_summary): Update.
29136 (do_estimate_edge_time): Update.
29137 (inline_generate_summary): Update.
29138 (inline_read_section): Update.
29139 (inline_read_summary): Update.
29140 (inline_free_summary): Update.
29141 * ipa-inline.c (can_inline_edge_p): Update.
29142 (compute_inlined_call_time): Update.
29143 (want_inline_small_function_p): Update.
29144 (edge_badness): Update.
29145 (early_inliner): Update.
29146 * ipa-inline.h (inline_edge_summary): Turn to ...
29147 (ipa_call_summary): ... this one.
29148 (ipa_call_summary_t): New class.
29149 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29150 (ipa_call_summaries): New.
29151 (inline_edge_summary): Remove.
29152 (estimate_edge_growth): Update.
29153 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29154 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29155 * ipa-split.c (execute_split_functions): Update.
29156 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29158 2017-05-19 Richard Biener <rguenther@suse.de>
29160 PR middle-end/80764
29161 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29163 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29165 * config/rs6000/rs6000.c (struct machine_function): Add field
29166 fpr_is_wrapped_separately.
29167 (rs6000_get_separate_components): Use 64 components. Handle the
29168 new FPR components.
29169 (rs6000_components_for_bb): Handle the FPR components.
29170 (rs6000_emit_prologue_components): Handle the FPR components.
29171 (rs6000_emit_epilogue_components): Handle the FPR components.
29172 (rs6000_set_handled_components): Handle the FPR components.
29173 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29174 that are already separately shrink-wrapped.
29175 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29176 that are already separately shrink-wrapped.
29178 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29181 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29184 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29185 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29186 cases where the register allocator uses a move and an offsettable
29187 memory operation to/from a FPR register on ISA 2.06/2.07.
29188 (define_peephole2 for Altivec d-form store): Likewise.
29190 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29193 * config/i386/mmx.md (*mov<mode>_internal): Enable
29194 alternatives 11, 12, 13 and 14 also for 32bit targets.
29195 Remove alternatives 15, 16, 17 and 18.
29196 * config/i386/sse.md (vec_concatv2di): Change
29197 alternative (!x, *y) to (x, ?!*Yn).
29199 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29201 * dumpfile.h (enum dump_kind): Remove stray comma.
29203 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29205 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29206 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29207 predicate::num_conditions
29208 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29209 (CHANGED): turn into predicate::changed.
29210 (agg_position_info): Move to ipa-predicate.h
29211 (add_condition, predicate::add_clause, predicate::operator &=,
29212 predicate::or_with, predicate::evaluate, predicate::probability,
29213 dump_condition, dump_clause, predicate::dump,
29214 predicate::remap_after_duplication, predicate::remap_after_inlining,
29215 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29216 (evaluate_conditions_for_known_args): Update.
29217 (set_cond_stmt_execution_predicate): Update.
29218 * ipa-inline.h: Include ipa-predicate.h
29219 (condition, inline_param_summary, conditions, agg_position_info,
29220 predicate): Move to ipa-predicate.h
29221 * ipa-predicate.c: New file.
29222 * ipa-predicate.h: New file.
29224 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29226 * final.c (leaf_function_p): Check we are not in a sequence.
29228 2017-05-18 Martin Liska <mliska@suse.cz>
29230 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29231 * dumpfile.c (dump_register): Use new enum dump_kind.
29232 (get_dump_file_name): Likewise.
29233 (dump_enable_all): Likewise.
29234 (dump_switch_p_1): Likewise.
29235 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29236 * dumpfile.h (enum dump_kind): New enum type.
29237 (struct dump_file_info): Create constructor and
29238 format fields and comments.
29239 * passes.c (pass_manager::register_one_dump_file):
29241 * statistics.c (statistics_early_init): Likewise.
29242 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29243 TDF_TREE with TDF_SLIM.
29244 (gather_memory_references_ref): Likewise.
29246 2017-05-18 Martin Liska <mliska@suse.cz>
29248 * vec.h (struct vnull): Use it.
29250 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29252 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29253 (true_predicate, false_predicate, true_predicate_p,
29254 false_predicate_p): Remove.
29255 (single_cond_predicate, not_inlined_predicate): Turn to member function
29257 (add_condition): Update.
29258 (add_clause): Turn to...
29259 (predicate::add_clause): ... this one; update; allow passing NULL
29261 (and_predicates): Turn to ...
29262 (predicate::operator &=): ... this one.
29263 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29264 (or_predicates): Turn to ...
29265 (predicate::or_with): ... this one.
29266 (evaluate_predicate): Turn to ...
29267 (predicate::evaluate): ... this one.
29268 (predicate_probability): Turn to ...
29269 (predicate::probability): ... this one.
29270 (dump_condition): Update.
29271 (dump_predicate): Turn to ...
29272 (predicate::dump): ... this one.
29273 (account_size_time): Update.
29274 (edge_set_predicate): Update.
29275 (set_hint_predicate): UPdate.
29276 (evaluate_conditions_for_known_args): Update.
29277 (evaluate_properties_for_edge): Update.
29278 (remap_predicate_after_duplication): Turn to...
29279 (predicate::remap_after_duplication): ... this one.
29280 (remap_hint_predicate_after_duplication): Update.
29281 (inline_summary_t::duplicate): UPdate.
29282 (dump_inline_edge_summary): Update.
29283 (dump_inline_summary): Update.
29284 (set_cond_stmt_execution_predicate): Update.
29285 (set_switch_stmt_execution_predicate): Update.
29286 (compute_bb_predicates): Update.
29287 (will_be_nonconstant_expr_predicate): Update.
29288 (will_be_nonconstant_predicate): Update.
29289 (phi_result_unknown_predicate): Update.
29290 (predicate_for_phi_result): Update.
29291 (array_index_predicate): Update.
29292 (estimate_function_body_sizes): Update.
29293 (estimate_node_size_and_time): Update.
29294 (estimate_ipcp_clone_size_and_time): Update.
29295 (remap_predicate): Rename to ...
29296 (predicate::remap_after_inlining): ... this one.
29297 (remap_hint_predicate): Update.
29298 (inline_merge_summary): Update.
29299 (inline_update_overall_summary): Update.
29300 (estimate_size_after_inlining): Update.
29301 (read_predicate): Rename to ...
29302 (predicate::stream_in): ... this one.
29303 (read_inline_edge_summary): Update.
29304 (write_predicate): Rename to ...
29305 (predicate::stream_out): ... this one.
29306 (write_inline_edge_summary): Update.
29307 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29308 (clause_t): Turn to uint32_t
29309 (predicate): Turn to class; implement constructor and operators
29311 (size_time_entry): Update.
29312 (inline_summary): Update.
29313 (inline_edge_summary): Update.
29315 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29317 * fold-const.c (fold_binary_loc): Move transformation...
29318 * match.pd (C - X CMP X): ... here.
29320 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29322 * config/sparc/sparc.c (sparc_option_override): Set function
29323 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29324 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29326 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29328 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29330 2017-05-18 Marek Polacek <polacek@redhat.com>
29333 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29334 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29335 gimple_assign_load_p.
29337 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29339 PR middle-end/80692
29340 * real.c (do_compare): Give decimal_do_compare preference over
29341 comparing just the signs.
29343 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29345 * doc/md.texi (Canonicalization of Instructions): Describe the
29346 canonical form of instructions that inherently set a condition
29349 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29351 PR middle-end/80775
29352 * tree-cfg.c: Move deletion of unreachable case statements to after
29353 the merging of consecutive case labels.
29355 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29357 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29358 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29359 restoring of callee-saved registers.
29361 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29363 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29364 * config/visium/visium.c (single_set_and_flags): Likewise.
29365 * config/visium/visium.md (Substitutions): Likewise.
29367 2017-05-17 Martin Liska <mliska@suse.cz>
29369 * cfg.c: Introduce dump_flags_t type and
29370 use it instead of int type.
29372 * cfghooks.c: Likewise.
29373 * cfghooks.h (struct cfg_hooks): Likewise.
29374 * cfgrtl.c: Likewise.
29375 * cfgrtl.h: Likewise.
29376 * cgraph.c (cgraph_node::get_body): Likewise.
29377 * coretypes.h: Likewise.
29378 * domwalk.c: Likewise.
29379 * domwalk.h: Likewise.
29380 * dumpfile.c (struct dump_option_value_info): Likewise.
29381 (dump_enable_all): Likewise.
29382 (dump_switch_p_1): Likewise.
29383 (opt_info_switch_p): Likewise.
29384 * dumpfile.h (enum tree_dump_index): Likewise.
29385 (struct dump_file_info): Likewise.
29386 * genemit.c: Likewise.
29387 * generic-match-head.c: Likewise.
29388 * gengtype.c (open_base_files): Likewise.
29389 * gimple-pretty-print.c: Likewise.
29390 * gimple-pretty-print.h: Likewise.
29391 * graph.c (print_graph_cfg): Likewise.
29392 * graphite-scop-detection.c (dot_all_sese): Likewise.
29393 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29394 * loop-unroll.c (report_unroll): Likewise.
29395 * passes.c (pass_manager::register_one_dump_file): Likewise.
29396 * print-tree.c: Likewise.
29397 * statistics.c: Likewise.
29398 * tree-cfg.c: Likewise.
29399 * tree-cfg.h: Likewise.
29400 * tree-dfa.c: Likewise.
29401 * tree-dfa.h: Likewise.
29402 * tree-dump.c (dump_function): Likewise.
29403 * tree-dump.h (struct dump_info): Likewise.
29404 * tree-pretty-print.c: Likewise.
29405 * tree-pretty-print.h: Likewise.
29406 * tree-ssa-live.c: Likewise.
29407 * tree-ssa-live.h: Likewise.
29408 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29409 * tree-vect-loop.c: Likewise.
29410 * tree-vect-slp.c: Likewise.
29412 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29413 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29415 PR tree-optimization/80457
29416 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29417 of all arguments to a statement as scalar_to_vec operations.
29418 (vectorizable_call): Adjust call to vect_model_simple_cost for
29420 (vectorizable_conversion): Likewise.
29421 (vectorizable_assignment): Likewise.
29422 (vectorizable_shift): Likewise.
29423 (vectorizable_operation): Likewise.
29424 (vectorizable_comparison): Likewise.
29425 (vect_is_simple_cond): Record the def types for operands.
29426 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29427 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29428 for statement argument count.
29430 2017-05-16 Carl Love <cel@us.ibm.com>
29432 * config/rs6000/rs6000-c: Add support for built-in functions
29433 vector unsigned long long vec_bperm (vector unsigned long long,
29434 vector unsigned char)
29435 vector signed long long vec_mule (vector signed int,
29437 vector unsigned long long vec_mule (vector unsigned int,
29438 vector unsigned int)
29439 vector signed long long vec_mulo (vector signed int,
29441 vector unsigned long long vec_mulo (vector unsigned int,
29442 vector unsigned int)
29443 vector signed char vec_sldw (vector signed char,
29444 vector signed char,
29446 vector unsigned char vec_sldw (vector unsigned char,
29447 vector unsigned char,
29449 vector signed short vec_sldw (vector signed short,
29450 vector signed short,
29452 vector unsigned short vec_sldw (vector unsigned short,
29453 vector unsigned short,
29455 vector signed int vec_sldw (vector signed int,
29458 vector unsigned int vec_sldw (vector unsigned int,
29459 vector unsigned int,
29461 vector signed long long vec_sldw (vector signed long long,
29462 vector signed long long,
29464 vector unsigned long long vec_sldw (vector unsigned long long,
29465 vector unsigned long long,
29467 * config/rs6000/rs6000-c: Add support for built-in functions
29468 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29469 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29470 * doc/extend.texi: Update the built-in documentation for the
29471 new built-in functions.
29473 2017-05-16 Marek Polacek <polacek@redhat.com>
29477 * tree.c (save_expr): Don't fold the expression.
29479 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29481 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29482 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29483 and (?*y,m). Update insn attributes.
29485 2017-05-16 Martin Liska <mliska@suse.cz>
29487 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29488 flags argument of print_gimple_stmt, print_gimple_expr,
29489 print_generic_stmt and print_generic_expr.
29490 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29491 * coretypes.h: Likewise.
29492 * except.c (dump_eh_tree): Likewise.
29493 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29494 * gimple-pretty-print.h: Likewise.
29495 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29496 (backprop::push_to_worklist): Likewise.
29497 (backprop::pop_from_worklist): Likewise.
29498 (backprop::process_use): Likewise.
29499 (backprop::intersect_uses): Likewise.
29500 (note_replacement): Likewise.
29501 * gimple-ssa-store-merging.c
29502 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29503 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29504 (pass_store_merging::execute): Likewise.
29505 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29506 (ssa_base_cand_dump_callback): Likewise.
29507 (dump_incr_vec): Likewise.
29508 (replace_refs): Likewise.
29509 (replace_mult_candidate): Likewise.
29510 (create_add_on_incoming_edge): Likewise.
29511 (create_phi_basis): Likewise.
29512 (insert_initializers): Likewise.
29513 (all_phi_incrs_profitable): Likewise.
29514 (introduce_cast_before_cand): Likewise.
29515 (replace_one_candidate): Likewise.
29516 * gimplify.c (gimplify_expr): Likewise.
29517 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29518 (set_rename): Likewise.
29519 (rename_uses): Likewise.
29520 (copy_loop_phi_nodes): Likewise.
29521 (add_close_phis_to_merge_points): Likewise.
29522 (copy_loop_close_phi_args): Likewise.
29523 (copy_cond_phi_args): Likewise.
29524 (graphite_copy_stmts_from_block): Likewise.
29525 (translate_pending_phi_nodes): Likewise.
29526 * graphite-poly.c (print_pdr): Likewise.
29527 (dump_gbb_cases): Likewise.
29528 (dump_gbb_conditions): Likewise.
29529 (print_scop_params): Likewise.
29530 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29531 (build_cross_bb_scalars_use): Likewise.
29532 (gather_bbs::before_dom_children): Likewise.
29533 * hsa-dump.c (dump_hsa_immed): Likewise.
29534 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29535 (get_replacement_map): Likewise.
29536 * ipa-inline-analysis.c (dump_condition): Likewise.
29537 (estimate_function_body_sizes): Likewise.
29538 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29539 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29540 * ipa-prop.c (ipa_dump_param): Likewise.
29541 (ipa_print_node_jump_functions_for_edge): Likewise.
29542 (ipa_modify_call_arguments): Likewise.
29543 (ipa_modify_expr): Likewise.
29544 (ipa_dump_param_adjustments): Likewise.
29545 (ipa_dump_agg_replacement_values): Likewise.
29546 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29547 * ipa-pure-const.c (check_stmt): Likewise.
29548 (pass_nothrow::execute): Likewise.
29549 * ipa-split.c (execute_split_functions): Likewise.
29550 * omp-offload.c (dump_oacc_loop_part): Likewise.
29551 (dump_oacc_loop): Likewise.
29552 * trans-mem.c (tm_log_emit): Likewise.
29553 (tm_memopt_accumulate_memops): Likewise.
29554 (dump_tm_memopt_set): Likewise.
29555 (dump_tm_memopt_transform): Likewise.
29556 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29557 (print_loop): Likewise.
29558 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29559 (chkp_gather_checks_info): Likewise.
29560 (chkp_get_check_result): Likewise.
29561 (chkp_remove_check_if_pass): Likewise.
29562 (chkp_use_outer_bounds_if_possible): Likewise.
29563 (chkp_reduce_bounds_lifetime): Likewise.
29564 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29565 (chkp_mark_completed_bounds): Likewise.
29566 (chkp_register_incomplete_bounds): Likewise.
29567 (chkp_mark_invalid_bounds): Likewise.
29568 (chkp_maybe_copy_and_register_bounds): Likewise.
29569 (chkp_build_returned_bound): Likewise.
29570 (chkp_get_bound_for_parm): Likewise.
29571 (chkp_build_bndldx): Likewise.
29572 (chkp_get_bounds_by_definition): Likewise.
29573 (chkp_generate_extern_var_bounds): Likewise.
29574 (chkp_get_bounds_for_decl_addr): Likewise.
29575 * tree-chrec.c (chrec_apply): Likewise.
29576 * tree-data-ref.c (dump_data_reference): Likewise.
29577 (dump_subscript): Likewise.
29578 (dump_data_dependence_relation): Likewise.
29579 (analyze_overlapping_iterations): Likewise.
29580 * tree-inline.c (expand_call_inline): Likewise.
29581 (tree_function_versioning): Likewise.
29582 * tree-into-ssa.c (dump_defs_stack): Likewise.
29583 (dump_currdefs): Likewise.
29584 (dump_names_replaced_by): Likewise.
29585 (dump_update_ssa): Likewise.
29586 (update_ssa): Likewise.
29587 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29588 * tree-parloops.c (build_new_reduction): Likewise.
29589 (try_create_reduction_list): Likewise.
29590 (ref_conflicts_with_region): Likewise.
29591 (oacc_entry_exit_ok_1): Likewise.
29592 (oacc_entry_exit_single_gang): Likewise.
29593 * tree-pretty-print.h: Likewise.
29594 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29595 (get_scalar_evolution): Likewise.
29596 (add_to_evolution): Likewise.
29597 (get_loop_exit_condition): Likewise.
29598 (analyze_evolution_in_loop): Likewise.
29599 (analyze_initial_condition): Likewise.
29600 (analyze_scalar_evolution): Likewise.
29601 (instantiate_scev): Likewise.
29602 (number_of_latch_executions): Likewise.
29603 (gather_chrec_stats): Likewise.
29604 (final_value_replacement_loop): Likewise.
29605 (scev_const_prop): Likewise.
29606 * tree-sra.c (dump_access): Likewise.
29607 (disqualify_candidate): Likewise.
29608 (create_access): Likewise.
29609 (reject): Likewise.
29610 (maybe_add_sra_candidate): Likewise.
29611 (create_access_replacement): Likewise.
29612 (analyze_access_subtree): Likewise.
29613 (analyze_all_variable_accesses): Likewise.
29614 (sra_modify_assign): Likewise.
29615 (initialize_constant_pool_replacements): Likewise.
29616 (find_param_candidates): Likewise.
29617 (decide_one_param_reduction): Likewise.
29618 (replace_removed_params_ssa_names): Likewise.
29619 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29620 * tree-ssa-copy.c (dump_copy_of): Likewise.
29621 (copy_prop_visit_cond_stmt): Likewise.
29622 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29623 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29624 (record_equivalences_from_stmt): Likewise.
29625 * tree-ssa-dse.c (compute_trims): Likewise.
29626 (delete_dead_call): Likewise.
29627 (delete_dead_assignment): Likewise.
29628 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29629 (forward_propagate_into_cond): Likewise.
29630 (pass_forwprop::execute): Likewise.
29631 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29632 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29634 (move_computations_worker): Likewise.
29635 (execute_sm): Likewise.
29636 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29637 (remove_exits_and_undefined_stmts): Likewise.
29638 (remove_redundant_iv_tests): Likewise.
29639 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29640 (adjust_iv_update_pos): Likewise.
29641 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29642 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29643 (value_replacement): Likewise.
29644 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29645 * tree-ssa-pre.c (print_pre_expr): Likewise.
29646 (get_representative_for): Likewise.
29647 (create_expression_by_pieces): Likewise.
29648 (insert_into_preds_of_block): Likewise.
29649 (eliminate_insert): Likewise.
29650 (eliminate_dom_walker::before_dom_children): Likewise.
29651 (eliminate): Likewise.
29652 (remove_dead_inserted_code): Likewise.
29653 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29654 * tree-ssa-reassoc.c (get_rank): Likewise.
29655 (eliminate_duplicate_pair): Likewise.
29656 (eliminate_plus_minus_pair): Likewise.
29657 (eliminate_not_pairs): Likewise.
29658 (undistribute_ops_list): Likewise.
29659 (eliminate_redundant_comparison): Likewise.
29660 (update_range_test): Likewise.
29661 (optimize_range_tests_var_bound): Likewise.
29662 (optimize_vec_cond_expr): Likewise.
29663 (rewrite_expr_tree): Likewise.
29664 (rewrite_expr_tree_parallel): Likewise.
29665 (linearize_expr): Likewise.
29666 (break_up_subtract): Likewise.
29667 (linearize_expr_tree): Likewise.
29668 (attempt_builtin_powi): Likewise.
29669 (attempt_builtin_copysign): Likewise.
29670 (transform_stmt_to_copy): Likewise.
29671 (transform_stmt_to_multiply): Likewise.
29672 (dump_ops_vector): Likewise.
29673 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29674 (print_scc): Likewise.
29675 (set_ssa_val_to): Likewise.
29676 (visit_reference_op_store): Likewise.
29677 (visit_use): Likewise.
29678 (sccvn_dom_walker::before_dom_children): Likewise.
29679 (run_scc_vn): Likewise.
29680 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29682 (expr_hash_elt::print): Likewise.
29683 (const_and_copies::pop_to_marker): Likewise.
29684 (const_and_copies::record_const_or_copy_raw): Likewise.
29685 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29686 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29687 (dump_predicates): Likewise.
29688 (find_uninit_use): Likewise.
29689 (warn_uninitialized_phi): Likewise.
29690 (pass_late_warn_uninitialized::execute): Likewise.
29691 * tree-ssa.c (verify_vssa): Likewise.
29692 (verify_ssa): Likewise.
29693 (maybe_optimize_var): Likewise.
29694 * tree-vrp.c (dump_value_range): Likewise.
29695 (dump_all_value_ranges): Likewise.
29696 (dump_asserts_for): Likewise.
29697 (register_edge_assert_for_2): Likewise.
29698 (vrp_visit_cond_stmt): Likewise.
29699 (vrp_visit_switch_stmt): Likewise.
29700 (vrp_visit_stmt): Likewise.
29701 (vrp_visit_phi_node): Likewise.
29702 (simplify_cond_using_ranges_1): Likewise.
29703 (fold_predicate_in): Likewise.
29704 (evrp_dom_walker::before_dom_children): Likewise.
29705 (evrp_dom_walker::push_value_range): Likewise.
29706 (evrp_dom_walker::pop_value_range): Likewise.
29707 (execute_early_vrp): Likewise.
29709 2017-05-16 Richard Biener <rguenther@suse.de>
29711 * dwarf2out.c (loc_list_from_tree_1): Do not create
29712 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29714 2017-05-16 Richard Biener <rguenther@suse.de>
29716 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29718 (note_variable_value_in_expr): If we resolved the decl ref
29719 do not push to the stack.
29721 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29723 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29724 operations in fast-math mode.
29725 (vaddq_f16): Likewise.
29726 (vmul_f16): Likewise.
29727 (vmulq_f16): Likewise.
29728 (vsub_f16): Likewise.
29729 (vsubq_f16): Likewise.
29730 * config/arm/neon.md (add<mode>3): New.
29732 (fma:<VH:mode>3): New. Also remove outdated comment.
29735 2017-05-16 Martin Liska <mliska@suse.cz>
29739 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29740 (odr_types_equivalent_p): Likewise.
29742 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29744 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29746 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29749 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29750 non-interunit SSE move alternatives with '?'.
29751 (zero-extendsidi peephole2): New peephole to skip intermediate
29752 general register in SSE zero-extend sequence.
29754 2017-05-15 Jeff Law <law@redhat.com>
29756 * reorg.c (relax_delay_slots): Create a new variable to hold
29757 the temporary target rather than clobbering TARGET_LABEL.
29759 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29760 missing argument to extract_bit_field call.
29761 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29763 2017-05-15 Martin Liska <mliska@suse.cz>
29766 * gcc.c (process_command): Do not allow empty argument of -o option.
29768 2017-05-15 Renlin Li <renlin.li@arm.com>
29770 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29771 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29772 * config/aarch64/constraints.md (Usf): Add long call check.
29773 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29774 (call_value): Likewise.
29775 (sibcall): Likewise.
29776 (sibcall_value): Likewise.
29778 (call_value_insn): New.
29779 (sibcall_insn): Update rtx pattern.
29780 (sibcall_value_insn): Likewise.
29781 (call_internal): Remove.
29782 (call_value_internal): Likewise.
29783 (sibcall_internal): Likewise.
29784 (sibcall_value_internal): Likewise.
29785 (call_reg): Likewise.
29786 (call_symbol): Likewise.
29787 (call_value_reg): Likewise.
29788 (call_value_symbol): Likewise.
29790 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29793 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29795 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29797 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29798 compatible with CCGOCmode and with CCZmode.
29800 2017-05-14 Martin Sebor <msebor@redhat.com>
29802 PR middle-end/77671
29803 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29804 (gimple_fold_builtin_snprintf): Same.
29805 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29806 (gimple_fold_builtin_snprintf): Same.
29807 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29808 of character types.
29809 (is_call_safe): New function.
29810 (try_substitute_return_value): Call it.
29811 (try_simplify_call): New function.
29812 (pass_sprintf_length::handle_gimple_call): Call it.
29814 2017-05-14 Martin Sebor <msebor@redhat.com>
29816 PR middle-end/80669
29817 * builtins.c (expand_builtin_stpncpy): Simplify.
29819 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29821 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29822 * config/i386/i386.h
29823 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29824 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29825 (struct machine_function): Add new members call_ms2sysv,
29826 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29827 (struct machine_frame_state): New fields sp_realigned and
29828 sp_realigned_offset.
29829 * config/i386/i386.c
29830 (enum xlogue_stub): New enum.
29831 (enum xlogue_stub_sets): New enum.
29832 (class xlogue_layout): New class.
29833 (struct ix86_frame): New fields stack_realign_allocate_offset,
29834 stack_realign_offset and outlined_save_offset. Modify comments to
29835 detail stack layout when using out-of-line stubs.
29836 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29837 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29838 -mcall-ms2sysv-xlogues.
29839 (stub_managed_regs): New static variable.
29840 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29841 registers managed by out-of-line stub.
29842 (disable_call_ms2sysv_xlogues): New function.
29843 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29844 m->call_ms2sysv when appropriate and compute frame layout for
29846 (sp_valid_at, fp_valid_at): New inline functions.
29847 (choose_basereg): New function.
29848 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29850 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29851 Use align parameter of choose_baseaddr to generated aligned SSE movs
29853 (pro_epilogue_adjust_stack): Modify to track
29854 machine_frame_state::sp_realigned.
29855 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29856 (ix86_nsaved_sseregs): Likewise.
29857 (ix86_emit_save_regs): Likewise.
29858 (ix86_emit_save_regs_using_mov): Likewise.
29859 (ix86_emit_save_sse_regs_using_mov): Likewise.
29860 (get_scratch_register_on_entry): Likewise.
29861 (gen_frame_set): New function.
29862 (gen_frame_load): Likewise.
29863 (gen_frame_store): Likewise.
29864 (emit_outlined_ms2sysv_save): Likewise.
29865 (emit_outlined_ms2sysv_restore): Likewise.
29866 (ix86_expand_prologue): Modify stack re-alignment code and call
29867 emit_outlined_ms2sysv_save when appropriate.
29868 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29869 parameter rtx_insn *insn, which allows the function to be used to only
29870 generate the notes.
29871 (ix86_expand_epilogue): Modify validity checks of frame and stack
29872 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29873 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29874 * config/i386/predicates.md
29875 (save_multiple): New predicate.
29876 (restore_multiple): Likewise.
29877 * config/i386/sse.md
29878 (save_multiple<mode>): New pattern.
29879 (save_multiple_realign<mode>): Likewise.
29880 (restore_multiple<mode>): Likewise.
29881 (restore_multiple_and_return<mode>): Likewise.
29882 (restore_multiple_leave_return<mode>): Likewise.
29883 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29885 2017-05-14 Julia Koval <julia.koval@intel.com>
29887 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29888 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29889 (__builtin_ia32_xsetbv): New builtins.
29890 * config/i386/i386.c (ix86_expand_special_args_builtin):
29892 (ix86_expand_builtin): Special expand for new intrinsics.
29893 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29894 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29895 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29899 * cfganal.c (inverted_post_order_compute): Change argument type
29901 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29902 * df-core.c (rest_of_handle_df_initialize): Adjust.
29903 (rest_of_handle_df_finish): Likewise.
29904 (df_analyze_1): Likewise.
29905 (df_analyze): Likewise.
29906 (loop_inverted_post_order_compute): Change argument to be a vec *.
29907 (df_analyze_loop): Adjust.
29908 (df_get_n_blocks): Likewise.
29909 (df_get_postorder): Likewise.
29910 * df.h (struct df_d): Change field to be a vec.
29911 * lcm.c (compute_laterin): Adjust.
29912 (compute_available): Likewise.
29913 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29914 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29915 * tree-ssa-pre.c (compute_antic): Likewise.
29917 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29919 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29920 (depth_first_search::depth_first_search): Change structure init
29921 function to this constructor.
29922 (depth_first_search::add_bb): Rename function to this member.
29923 (depth_first_search::execute): Likewise.
29924 (flow_dfs_compute_reverse_finish): Adjust.
29926 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29928 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29929 (longest_simple_path): Likewise.
29930 * shrink-wrap.c (spread_components): Likewise.
29931 (disqualify_problematic_components): Likewise.
29932 (emit_common_heads_for_components): Likewise.
29933 (emit_common_tails_for_components): Likewise.
29934 (insert_prologue_epilogue_for_components): Likewise.
29936 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29938 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29941 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29943 * df-core.c (df_set_blocks): Start using auto_bitmap.
29944 (df_compact_blocks): Likewise.
29945 * df-problems.c (df_rd_confluence_n): Likewise.
29946 * df-scan.c (df_insn_rescan_all): Likewise.
29947 (df_process_deferred_rescans): Likewise.
29948 (df_update_entry_block_defs): Likewise.
29949 (df_update_exit_block_uses): Likewise.
29950 (df_entry_block_bitmap_verify): Likewise.
29951 (df_exit_block_bitmap_verify): Likewise.
29952 (df_scan_verify): Likewise.
29953 * lra-constraints.c (lra_constraints): Likewise.
29954 (undo_optional_reloads): Likewise.
29955 (lra_undo_inheritance): Likewise.
29956 * lra-remat.c (calculate_gen_cands): Likewise.
29957 (do_remat): Likewise.
29958 * lra-spills.c (assign_spill_hard_regs): Likewise.
29959 (spill_pseudos): Likewise.
29960 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29961 (bitmap_set_subtract_values): Likewise.
29963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29965 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29966 management with auto_bitmap.
29967 (fix_inter_tick): Likewise.
29968 (fix_recovery_deps): Likewise.
29969 * ira.c (add_store_equivs): Likewise.
29970 (find_moveable_pseudos): Likewise.
29971 (split_live_ranges_for_shrink_wrap): Likewise.
29972 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29973 (rtx_reuse_manager::seen_def_p): Likewise.
29974 (rtx_reuse_manager::set_seen_def): Likewise.
29975 * print-rtl.h (class rtx_reuse_manager): Likewise.
29977 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29979 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29981 (migrate_btr_def): Likewise.
29982 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29983 * df-core.c (loop_post_order_compute): Likewise.
29984 (loop_inverted_post_order_compute): Likewise.
29985 * hsa-common.h: Likewise.
29986 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29987 * init-regs.c (initialize_uninitialized_regs): Likewise.
29988 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29989 (inline_small_functions): Likewise.
29990 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29991 * ira.c (combine_and_move_insns): Likewise.
29992 (build_insn_chain): Likewise.
29993 * loop-invariant.c (find_invariants): Likewise.
29994 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29995 * predict.c (tree_predict_by_opcode): Likewise.
29996 (predict_paths_leading_to): Likewise.
29997 (predict_paths_leading_to_edge): Likewise.
29998 (estimate_loops_at_level): Likewise.
29999 (estimate_loops): Likewise.
30000 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30001 (spread_components): Likewise.
30002 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30003 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30004 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30005 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30006 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30007 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30008 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30009 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30010 (mark_threaded_blocks): Likewise.
30011 (thread_through_all_blocks): Likewise.
30012 * tree-ssa.c (verify_ssa): Likewise.
30013 (execute_update_addresses_taken): Likewise.
30014 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30016 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30018 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30020 (post_order_compute): Likewise.
30021 (inverted_post_order_compute): Likewise.
30022 (pre_and_rev_post_order_compute_fn): Likewise.
30024 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30026 * genrecog.c (int_set::int_set): Explicitly construct our
30027 auto_vec base class.
30028 * vec.h (auto_vec::auto_vec): New constructor.
30030 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30032 * bitmap.h (class auto_bitmap): New constructor taking
30033 bitmap_obstack * argument.
30035 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30037 * bitmap.h (class auto_bitmap): Change type of m_bits to
30038 bitmap_head, and adjust ctor / dtor and member operators.
30040 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30042 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30043 when returned register mode doesn't match original mode.
30045 2017-05-12 Jeff Law <law@redhat.com>
30046 Jakub Jelinek <jakub@redhat.com>
30048 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30049 we look for cc setter after the compare-elim changes.
30050 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30051 within the vector to match what compare-elim now expects.
30052 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30053 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30055 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30056 after the compare-elim changes.
30057 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30058 the vector to match what compare-elim now expects.
30059 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30060 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30061 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30062 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30063 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30065 * config/visium/visium.c (single_set_and_flags): Fix where
30066 we look for cc setter after the compare-elim changes.
30067 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30068 with the vector to match what compare-elim now expects.
30069 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30070 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30071 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30072 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30073 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30074 (neg<mode>2_insn_set_overflow): Likewise.
30076 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30078 PR middle-end/79794
30079 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30080 maybe_expand_insn call, set ops[0].target. If still set after call,
30081 set alt_rtl. Add extra arg to recursive calls.
30082 (extract_bit_field): Add alt_rtl argument. Pass to
30084 * expmed.h (extract_bit_field): Fix prototype.
30085 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30086 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30087 to extract_bit_field_calls.
30088 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30089 Pass alt_rtl to extract_bit_field calls.
30090 * calls.c (store_unaligned_arguments_into_psuedos)
30091 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30092 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30094 * optabs.h (struct expand_operand): Add target bitfield.
30096 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30098 * compare-elim.c (try_eliminate_compare): Canonicalize
30099 operation with embedded compare to
30100 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30101 (set (reg) (operation)].
30103 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30105 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30108 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30109 cost of adding a carry flag for ADC instruction.
30110 [case MINUS]: Ignore the cost of subtracting a carry flag
30111 for SBB instruction.
30113 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30115 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30117 * config/rs6000/bmiintrin.h: New file.
30118 * config/rs6000/bmi2intrin.h: New file.
30119 * config/rs6000/x86intrin.h: New file.
30121 2017-05-12 Jeff Law <law@redhat.com>
30123 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30126 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30128 PR middle-end/80707
30129 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30131 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30133 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30134 early expansion of vector divide builtins.
30135 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30136 builtins identified as having unsigned arguments.
30138 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30140 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30141 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30142 expansion of vector logical operations (and, andc, or, xor,
30145 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30147 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30148 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30150 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30152 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30153 early GIMPLE expansion of vector multiplies.
30155 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30157 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30158 TARGET_HAVE_MOVT conditional.
30159 (movt splitter): Likewise.
30161 2017-05-12 Richard Biener <rguenther@suse.de>
30163 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30164 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30165 Fold all stmts not inplace.
30167 2017-05-12 Richard Biener <rguenther@suse.de>
30169 PR tree-optimization/80713
30170 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30171 inserted_exprs bit for not removed stmts.
30173 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30175 PR middle-end/69921
30176 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30177 parallelized" attribute for parallelized OpenACC kernels.
30178 * omp-offload.c (execute_oacc_device_lower): Use it.
30180 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30181 Set "oacc kernels" attribute.
30182 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30183 parameter. Adjust all users.
30184 (oacc_fn_attrib_kernels_p): Remove function.
30185 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30186 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30187 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30188 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30189 assert "oacc kernels" attribute is set.
30191 2017-05-11 Carl Love <cel@us.ibm.com>
30193 * config/rs6000/rs6000-c: Add support for built-in functions
30194 vector unsigned char vec_popcnt (vector signed char)
30195 vector unsigned char vec_popcnt (vector unsigned char)
30196 vector unsigned short vec_popcnt (vector signed short)
30197 vector unsigned short vec_popcnt (vector unsigned short)
30198 vector unsigned int vec_popcnt (vector signed int)
30199 vector unsigned int vec_popcnt (vector unsigned int)
30200 vector unsigned long long vec_popcnt (vector signed long long)
30201 vector unsigned long long vec_popcnt (vector unsigned long long)
30202 vector signed long long vec_slo (vector signed long long,
30203 vector signed char)
30204 vector signed long long vec_slo (vector signed long long,
30205 vector unsigned char)
30206 vector unsigned long long vec_slo (vector unsigned long long,
30207 vector signed char)
30208 vector unsigned long long vec_slo (vector unsigned long long,
30209 vector unsigned char)
30210 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30211 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30212 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30213 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30214 * doc/extend.texi: Update the built-in documentation file for the
30215 new built-in functions.
30217 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30219 * attribs.h (sorted_attr_string): Move machine independent
30220 functions for target clone support from the i386 port to common
30221 code. Rename ix86_function_versions to common_function_versions.
30222 Rename make_name to make_unique_name.
30223 (common_function_versions): Likewise.
30224 (make_unique_name): Likewise.
30225 (make_dispatcher_decl): Likewise.
30226 (is_function_default_version): Likewise.
30227 * attribs.c (attr_strcmp): Likewise.
30228 (sorted_attr_string): Likewise.
30229 (common_function_versions): Likewise.
30230 (make_unique_name): Likewise.
30231 (make_dispatcher_decl): Likewise.
30232 (is_function_default_version): Likewise.
30233 * config/i386/i386.c (attr_strcmp): Likewise.
30234 (sorted_attr_string): Likewise.
30235 (ix86_function_versions): Likewise.
30236 (make_name): Likewise.
30237 (make_dispatcher_decl): Likewise.
30238 (is_function_default_version): Likewise.
30239 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30241 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30244 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30245 Account for direct move costs for vec_construct of integer
30248 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30251 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30252 (UNSPEC_STX_ATOMIC): Ditto.
30253 (loaddi_via_sse): New insn.
30254 (storedi_via_sse): Ditto.
30255 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30256 Update corresponding peephole2 patterns.
30257 (atomic_storedi_fpu): Ditto.
30259 2017-05-11 Julia Koval <julia.koval@intel.com>
30261 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30262 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30264 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30265 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30266 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30268 2017-05-11 Nathan Sidwell <nathan@acm.org>
30270 * graphite-poly.c: Include dumpfile.h.
30272 * dumpfle.h (dump_function): Declare here ...
30273 * tree-dump.h (dump_function): ... not here.
30274 * dumpfile.c: #include tree-cfg.h.
30275 (dump_function): Move here from ...
30276 * tree-dump.c (dump_function): ... here.
30277 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30278 * graphite-poly.c: Don't include tree-dump.h.
30279 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30280 * print-tree.c: Likewise.
30281 * stor-layout.c: Likewise.
30282 * tree-nested.c: Likewise.
30284 * dumpfile.c (dump_start): Use TDF_FLAGS.
30285 (dump_enable_all): Fix TDF_KIND check thinko.
30287 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30289 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30290 array entries to represent two legal parameterizations of the
30291 overloaded __builtin_cmpb function, as represented by the
30292 P6_OV_BUILTIN_CMPB constant.
30293 (altivec_resolve_overloaded_builtin): Add special case handling
30294 for the __builtin_cmpb function, as represented by the
30295 P6_OV_BUILTIN_CMPB constant.
30296 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30297 (BU_P6_64BIT_2): New macro.
30298 (BU_P6_OVERLOAD_2): New macro
30299 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30300 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30301 (CMPB): Add overload support to represent both 32-bit and 64-bit
30302 compare-bytes function.
30303 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30304 support for TARGET_CMPB.
30305 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30306 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30307 documentation of the __builtin_cmpb overloaded built-in function.
30309 2017-05-11 Richard Biener <rguenther@suse.de>
30311 PR tree-optimization/80705
30312 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30313 bases are not vectorizable.
30315 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30317 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30318 when counting register pressure.
30320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30322 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30323 (iv_ca_more_deps): Renamed to ...
30324 (iv_ca_compare_deps): ... this.
30325 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30327 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30329 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30331 (determine_group_iv_costs): ... here.
30332 (find_inv_vars_cb): Record inv var if it's not recorded before.
30334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30336 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30337 (get_shiftadd_cost): Ditto.
30339 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30341 * tree-ssa-address.c: Include header file.
30342 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30344 (add_to_parts): Refactor.
30345 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30346 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30349 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30351 PR tree-optimization/53090
30352 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30354 (extract_cond_operands): Detect condition with IV on both sides
30355 and return COMP_IV_EXPR_2.
30356 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30357 (rewrite_use_compare): Simplify by removing call to function
30358 extract_cond_operands.
30360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30362 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30363 (extract_cond_operands): Detect condition comparing against non-
30364 invariant bound and return appropriate enum value.
30365 (find_interesting_uses_cond): Update use of extract_cond_operands.
30366 Handle its return value accordingly.
30367 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30369 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30371 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30372 nonlinear iv_use computation in loop invariant sensitive way.
30374 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30376 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30377 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30381 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30382 (dump_cand): Support iv_cand.inv_exprs.
30383 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30385 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30388 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30390 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30392 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30393 as local function. Include necessary header files.
30394 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30396 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30398 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30402 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30403 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30404 RSHIFT_EXPR and BIT_NOT_EXPR.
30406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30408 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30409 (adjust_setup_cost): New parameter supporting round up adjustment.
30410 (struct address_cost_data): Delete.
30411 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30412 (split_address_cost, ptr_difference_cost): Delete.
30413 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30414 (struct ainc_cost_data): New struct.
30415 (get_address_cost_ainc): New function.
30416 (get_address_cost, get_computation_cost): Reimplement.
30417 (determine_group_iv_cost_address): Record inv_expr for all uses of
30419 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30420 (iv_ca_has_deps): Reimplemented to ...
30421 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30423 (iv_ca_extend): Call iv_ca_more_deps.
30425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30427 * tree-ssa-address.c (struct mem_address): Move to header file.
30428 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30429 * tree-ssa-address.h (struct mem_address): Move from C file.
30430 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30434 * tree-affine.h (aff_combination_type): New interface.
30435 (aff_combination_zero_p): Remove static.
30436 (aff_combination_const_p): New interface.
30437 (aff_combination_singleton_var_p): New interfaces.
30439 2017-05-11 Richard Biener <rguenther@suse.de>
30441 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30442 Skip unreachable blocks and destinations.
30443 (eliminate): Move stmt removal and fixup ...
30444 (fini_eliminate): ... here. Skip inserted exprs.
30445 (pass_pre::execute): Move fini_pre after fini_eliminate.
30446 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30447 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30448 PRE to get rid of dead code that has invalid SSA form and
30449 split critical edges again.
30451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30453 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30455 2017-05-11 Richard Biener <rguenther@suse.de>
30457 * passes.c (execute_function_todo): Verify loops if they are
30458 said to be up-to-date.
30459 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30460 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30462 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30465 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30466 handle calling assemble_external ourself.
30469 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30470 modes with zero size. Enhance comment.
30472 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30474 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30475 built-ins for vec_xl and vec_xst with short and char pointer
30478 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30480 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30481 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30482 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30483 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30484 (_mm_maskz_min_round_ss): New intrinsics.
30485 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30486 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30487 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30488 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30489 (__builtin_ia32_minss_mask_round): New builtins.
30490 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30491 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30492 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30494 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30495 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30497 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30500 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30502 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30503 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30504 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30505 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30506 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30507 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30508 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30509 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30510 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30511 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30512 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30513 (__builtin_ia32_mulss_mask_round): New builtins.
30514 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30515 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30516 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30518 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30519 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30521 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30524 2017-05-10 Julia Koval <julia.koval@intel.com>
30526 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30527 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30528 (_mm256_setr_m128i): New intrinsics.
30530 2017-05-10 Julia Koval <julia.koval@intel.com>
30532 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30533 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30534 (_mm_maskz_rcp14_ss): New intrinsics.
30535 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30536 (__builtin_ia32_rcp14ss_mask): New builtins.
30537 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30539 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30541 PR tree-optimization/51513
30542 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30543 (assert_unreachable_fallthru_edge_p): Use it.
30544 (group_case_labels_stmt): Likewise.
30545 * tree-cfg.h: Prototype it.
30546 * stmt.c: Include cfghooks.h and tree-cfg.h.
30547 (emit_case_dispatch_table) <gap_label>: New local variable.
30548 Use it to fill dispatch table gaps.
30549 Test for default_label before updating probabilities.
30550 (expand_case) <default_label>: Remove unneeded initialization.
30551 Test for unreachable default case statement and remove its edge.
30552 Set default_label accordingly.
30553 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30555 2017-05-10 Carl Love <cel@us.ibm.com>
30557 * config/rs6000/rs6000-c: Add support for built-in functions
30558 vector signed char vec_neg (vector signed char)
30559 vector signed short int vec_neg (vector short int)
30560 vector signed int vec_neg (vector signed int)
30561 vector signed long long vec_neg (vector signed long long)
30562 vector float vec_neg (vector float)
30563 vector double vec_neg (vector double)
30564 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30566 * config/rs6000/altivec.h: Add define for vec_neg
30567 * doc/extend.texi: Update the built-in documentation for the
30568 new built-in functions.
30570 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30572 PR tree-optimization/77644
30573 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30575 2017-05-10 Nathan Sidwell <nathan@acm.org>
30577 * dumpfile.h (TDI_lang_all): New.
30578 (TDF_KIND): New. Renumber others
30579 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30581 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30583 (get_dump_file_name): Adjust suffix generation.
30584 (dump_enable_all): Use TDF_KIND.
30585 * doc/invoke.texi (-fdump-lang-all): Document.
30587 * dumpfile.h: Tabify.
30589 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30592 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30593 Move member access before delete.
30595 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30597 * tree-inline.c (expand_call_inline): Split block at stmt
30600 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30603 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30604 are now unused after splitting mov{sf,sd}_hardfloat.
30605 (f32_lr2): Likewise.
30606 (f32_lm): Likewise.
30607 (f32_lm2): Likewise.
30608 (f32_li): Likewise.
30609 (f32_li2): Likewise.
30610 (f32_lv): Likewise.
30611 (f32_sr): Likewise.
30612 (f32_sr2): Likewise.
30613 (f32_sm): Likewise.
30614 (f32_sm2): Likewise.
30615 (f32_si): Likewise.
30616 (f32_si2): Likewise.
30617 (f32_sv): Likewise.
30618 (f32_dm): Likewise.
30619 (f32_vsx): Likewise.
30620 (f32_av): Likewise.
30621 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30622 For movsf, order stores so the VSX stores occur before the GPR
30623 store which encourages the register allocator to use a traditional
30624 FPR instead of a GPR. For movsd, order the stores so that the GPR
30625 store comes before the VSX stores to allow the power6 to work.
30626 This is due to the power6 not having a 32-bit integer store
30627 instruction from a FPR.
30628 (movsf_hardfloat): Likewise.
30629 (movsd_hardfloat): Likewise.
30631 2017-05-09 Martin Sebor <msebor@redhat.com>
30633 PR translation/80280
30634 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30637 PR translation/80280
30638 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30639 data member added in r247778.
30640 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30642 2017-05-09 Nathan Sidwell <nathan@acm.org>
30644 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30646 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30649 2017-05-09 Marek Polacek <polacek@redhat.com>
30651 * doc/invoke.texi: Fix typo.
30653 2017-05-09 Richard Biener <rguenther@suse.de>
30655 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30656 (vrp_val_is_min): Likewise.
30657 (set_value_range_to_value): Likewise.
30658 (set_value_range_to_nonnegative): Likewise.
30659 (gimple_assign_nonzero_p): Likewise.
30660 (gimple_stmt_nonzero_p): Likewise.
30661 (vrp_int_const_binop): Likewise. Remove unreachable case.
30662 (adjust_range_with_scev): Adjust comments.
30663 (compare_range_with_value): Likewise.
30664 (extract_range_from_phi_node): Likewise.
30665 (test_for_singularity): Likewise.
30667 2017-05-09 Richard Biener <rguenther@suse.de>
30669 * tree-vrp.c (get_single_symbol): Add assert that we don't
30670 get overflowed constants as invariant part.
30671 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30672 checks. Use wi::cmp instead of recursing for integer constants.
30673 (compare_values): Just ignore whether we assumed undefined
30674 overflow instead of failing the compare.
30675 (extract_range_for_var_from_comparison_expr): Add comment before the
30676 TREE_NO_WARNING sets.
30677 (test_for_singularity): Likewise.
30678 (extract_range_from_comparison): Do not disable optimization
30679 when we assumed undefined overflow.
30680 (extract_range_basic): Remove init of unused var.
30682 2017-05-09 Richard Biener <rguenther@suse.de>
30684 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30685 (extract_range_from_multiplicative_op_1): Adjust.
30686 (extract_range_from_binary_expr_1): Use int_const_binop.
30688 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30691 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30692 rs6000_store_data_bypass_p in seven define_bypass directives and
30693 in several comments.
30694 * config/rs6000/rs6000-protos.h: Add prototype for
30695 rs6000_store_data_bypass_p function.
30696 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30697 function implements slightly different (rs6000-specific) semantics
30698 than store_data_bypass_p, returning false rather than aborting
30699 with assertion error when arguments do not satisfy the
30700 requirements of store data bypass.
30701 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30702 rs6000_store_data_bypass_p.
30704 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30706 * config/xtensa/xtensa-protos.h
30707 (xtensa_initial_elimination_offset): New declaration.
30708 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30709 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30710 macro definition, add case for FRAME_POINTER_REGNUM when
30711 FRAME_GROWS_DOWNWARD.
30712 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30713 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30714 xtensa_initial_elimination_offset.
30716 2017-05-08 Nathan Sidwell <nathan@acm.org>
30718 * doc/invoke.texi: Alphabetize -fdump options.
30720 2017-05-08 Martin Sebor <msebor@redhat.com>
30722 PR translation/80280
30723 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30725 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30727 * target.def (compute_frame_layout): New optional target hook.
30728 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30729 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30730 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30732 * reload1.c (verify_initial_elim_offsets): Likewise.
30733 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30734 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30735 (arm_get_frame_offsets): Split up into this ...
30736 (arm_compute_frame_layout): ... and this function.
30738 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30740 * config/aarch64/constraints.md (Usa): New constraint.
30741 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30743 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30745 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30746 with_multilib_list after it has been checked.
30748 2017-05-08 Richard Biener <rguenther@suse.de>
30750 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30751 (bitmap_set_subtract_values): Likewise.
30753 2017-05-08 Richard Biener <rguenther@suse.de>
30755 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30756 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30758 (gimple_stmt_nonzero_warnv_p): Rename to ...
30759 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30761 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30762 (extract_range_basic): Adjust, do not disable propagation on
30763 strict overflow sensitive simplification.
30764 (vrp_visit_cond_stmt): Likewise.
30766 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30768 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30769 body size unconditionally.
30771 2017-05-07 Jeff Law <law@redhat.com>
30774 2017-05-06 Jeff Law <law@redhat.com>
30775 PR tree-optimization/78496
30776 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30779 PR tree-optimization/78496
30780 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30781 (simplify_stmt_using_ranges): Call it.
30782 (vrp_dom_walker::before_dom_children): Extract equivalences
30783 from an ASSERT_EXPR with an equality comparison against a
30786 2017-05-06 Jeff Law <law@redhat.com>
30788 PR tree-optimization/78496
30789 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30792 PR tree-optimization/78496
30793 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30794 (simplify_stmt_using_ranges): Call it.
30795 (vrp_dom_walker::before_dom_children): Extract equivalences
30796 from an ASSERT_EXPR with an equality comparison against a
30799 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30801 * lra-constraints.c (lra_copy_reg_equiv): New function.
30802 (split_reg): Use it to copy equivalence information from the
30803 original register to the spill register.
30805 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30807 PR rtl-optimization/75964
30808 * simplify-rtx.c (simplify_const_relational_operation): Remove
30809 invalid handling of comparisons of integer ABS.
30811 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30813 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30814 initialize to zero.
30815 (init_regs): Remove declaration.
30816 (function_arg_advance_32): Initialize error_p as boolean variable.
30818 2017-05-05 Nathan Sidwell <nathan@acm.org>
30820 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30821 lines. Use for (;;).
30823 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30825 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30826 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30827 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30828 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30829 VF=2 that require versioning.
30831 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30833 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30836 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30838 * diagnostic.h (diagnostic_override_option_index): Convert from
30839 macro to inline function.
30841 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30843 * diagnostic.c (last_module_changed_p): New function.
30844 (set_last_module): New function.
30845 (diagnostic_report_current_module): Convert macro usage to
30846 the above functions.
30847 * diagnostic.h (diagnostic_context::last_module): Strengthen
30848 from const line_map * to const line_map_ordinary *.
30849 (diagnostic_last_module_changed): Delete macro.
30850 (diagnostic_set_last_module): Delete macro.
30852 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30854 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30855 with diagnostic_report_diagnostic.
30856 (diagnostic_n_impl_richloc): Likewise.
30857 * diagnostic.h (report_diagnostic): Delete macro.
30858 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30859 with diagnostic_report_diagnostic.
30860 * substring-locations.c (format_warning_va): Likewise.
30862 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30864 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30865 save/restor of format_spec. Move option-printing code to...
30866 (print_option_information): ...this new function, and
30867 reimplement by simply printing to the pretty_printer,
30868 rather than appending to the format string.
30870 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30872 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30873 handling logic into...
30874 (update_effective_level_from_pragmas): ...this new function.
30876 2017-05-04 Andrew Waterman <andrew@sifive.com>
30878 * config/riscv/riscv.opt (mstrict-align): New option.
30879 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30880 (SLOW_UNALIGNED_ACCESS): Define.
30881 (riscv_slow_unaligned_access): Declare.
30882 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30884 (riscv_slow_unaligned_access): New variable.
30885 (rocket_tune_info): Set slow_unaligned_access to true.
30886 (optimize_size_tune_info): Set slow_unaligned_access to false.
30887 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30888 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30889 (riscv_option_override): Set riscv_slow_unaligned_access.
30890 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30892 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30894 * config/riscv/riscv.md: Unify indentation.
30896 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30901 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30903 (extendsi<mode>2): Add support for doing sign extension via
30904 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30905 don't have ISA 3.0 instructions.
30906 (extendsi<mode>2 splitter): Likewise.
30907 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30908 generate the normal insns since SImode can now go in vector
30909 registers. Disallow the special UNSPECs needed for previous
30910 machines to hide SImode being used. Add new insns
30911 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30912 (fix_trunc<mode>si2_stfiwx): Likewise.
30913 (fix_trunc<mode>si2_internal): Likewise.
30914 (fixuns_trunc<mode>si2): Likewise.
30915 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30916 (fctiw<u>z_<mode>_smallint): Likewise.
30917 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30918 of floating point to 32-bit integer from doing a direct move to
30919 the GPR registers to do a store.
30920 (fctiwz_<mode>): Break long line.
30922 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30924 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30925 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30926 (addr_list, addr_offset_valid_p): New.
30927 (split_address_groups): Check offset validity with above function.
30928 (gt-tree-ssa-loop-ivopts.h): Include header file.
30930 2017-05-05 Nathan Sidwell <nathan@acm.org>
30932 * config.gcc (arm*-*-*): Add missing 'fi'.
30934 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30936 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30937 included in -fopt-info does not matter.
30938 * doc/optinfo.texi (-fopt-info): Fix description of default
30939 behavour. Explicitly say order of options included in -fopt-info
30942 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30944 * config.gcc: Allow combinations of aprofile and rmprofile values for
30945 --with-multilib-list.
30946 * config/arm/t-multilib: New file.
30947 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30948 variables. Remove setting of ISA and floating-point ABI in
30949 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30950 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30951 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30952 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30954 * config/arm/t-rmprofile: Likewise except for the matches changes.
30955 * doc/install.texi (--with-multilib-list): Document the combination of
30956 aprofile and rmprofile values and warn about pitfalls in doing that.
30958 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30960 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30961 (movdi_aarch64): Likewise.
30963 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30965 PR tree-optimization/80632
30966 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30968 (build_arrays): Initialize it for virtual phis.
30969 (fix_phi_nodes): Use it for virtual phis.
30971 PR tree-optimization/80558
30972 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30973 [x, y] op z into [x op, y op z] for op & or | if conditions
30976 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30977 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30980 * config/arm/arm.md (use_literal_pool): Remove.
30981 (64-bit immediate split): No longer takes cost into consideration
30982 if arm_disable_literal_pool is enabled.
30983 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30984 used when arm_disable_literal_pool is enabled.
30985 (arm_max_const_double_inline_cost): Remove use of
30986 arm_disable_literal_pool.
30987 (push_minipool_fix): Add assert.
30988 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30989 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30990 (no_literal_pool_sf_immediate): New.
30992 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30994 PR tree-optimization/80613
30995 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30996 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30998 2017-05-05 Richard Biener <rguenther@suse.de>
31000 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31002 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31004 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31005 of this flag from insn conditions due to removal from r247495.
31007 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31009 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31011 (arm_early_store_addr_dep_ptr): Likewise.
31012 * config/arm/aarch-common-protos.h
31013 (arm_early_load_addr_dep_ptr): Add prototype.
31014 (arm_early_store_addr_dep_ptr): Likewise.
31015 * config/arm/cortex-a53.md: Add new bypasses.
31017 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31019 * tree.c (next_type_uid): Change type to unsigned.
31020 (type_hash_canon): Decrement back next_type_uid if
31021 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31022 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31025 2017-05-04 Martin Sebor <msebor@redhat.com>
31027 * builtins.c: Fix a trivial typo in a comment.
31029 PR middle-end/79234
31030 * builtins.c (check_sizes): Adjust to handle reading past the end.
31031 Avoid printing excessive upper bound of ranges. Use %E to print
31032 tree nodes instead of converting them to %wu.
31033 (expand_builtin_memchr): New function.
31034 (compute_dest_size): Rename...
31035 (compute_objsize): ...to this.
31036 (expand_builtin_memcpy): Adjust.
31037 (expand_builtin_mempcpy): Adjust.
31038 (expand_builtin_strcat): Adjust.
31039 (expand_builtin_strcpy): Adjust.
31040 (check_strncat_sizes): Adjust.
31041 (expand_builtin_strncat): Adjust.
31042 (expand_builtin_strncpy): Adjust and simplify.
31043 (expand_builtin_memset): Adjust.
31044 (expand_builtin_bzero): Adjust.
31045 (expand_builtin_memcmp): Adjust.
31046 (expand_builtin): Handle memcmp.
31047 (maybe_emit_chk_warning): Check strncat just once.
31049 2017-05-04 Martin Sebor <msebor@redhat.com>
31051 PR preprocessor/79214
31052 PR middle-end/79222
31053 PR middle-end/79223
31054 * builtins.c (check_sizes): Add inlining context and issue
31055 warnings even when -Wno-system-headers is set.
31056 (check_strncat_sizes): Same.
31057 (expand_builtin_strncat): Same.
31058 (expand_builtin_memmove): New function.
31059 (expand_builtin_stpncpy): Same.
31060 (expand_builtin): Handle memmove and stpncpy.
31062 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31064 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31065 which is not used any more.
31067 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31069 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31071 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31073 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31074 (cortexa53_tunings): Likewise.
31075 (cortexa57_tunings): Likewise.
31076 (cortexa72_tunings): Likewise.
31077 (cortexa73_tunings): Likewise.
31079 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31081 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31082 Set loop alignment to 8.
31084 2017-05-04 Martin Sebor <msebor@redhat.com>
31086 PR translation/80280
31087 * builtins.c (expand_builtin_object_size): Add missing quoting to
31088 %D and like directives.
31089 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31090 (hsa_type_for_tree_type): Same.
31091 (verify_function_arguments): Same.
31092 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31093 * varasm.c (get_section): Same.
31096 2017-05-04 Martin Sebor <msebor@redhat.com>
31098 PR translation/80280
31099 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31101 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31103 * config/aarch64/aarch64.c (generic_addrcost_table):
31104 Change HI/TI mode setting.
31106 2017-05-04 Martin Jambor <mjambor@suse.cz>
31108 PR tree-optimization/80622
31109 * tree-sra.c (comes_initialized_p): New function.
31110 (build_accesses_from_assign): Only set write lazily when
31111 comes_initialized_p is false.
31112 (analyze_access_subtree): Use comes_initialized_p.
31113 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31114 instead of testing for PARM_DECL.
31116 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31118 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31119 constraint on operand 0 to allow more general addressing modes.
31120 Adjust output template.
31121 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31123 * config/aarch64/aarch64-protos.h
31124 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31125 * config/aarch64/constraints.md (Dp): New address constraint.
31126 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31129 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31131 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31132 update use of estimate_ipcp_clone_size_and_time.
31133 (estimate_local_effects): Update use of
31134 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31135 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31136 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31137 Return nonspecialized time.
31139 2017-05-04 Richard Biener <rguenther@suse.de>
31141 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31142 for the last VUSE which def dominates the PHI. Directly call
31144 (get_continuation_for_phi_1): Remove.
31146 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31148 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31149 to explain the use of truncating division. Cap the number of
31150 iterations to the maximum given by nb_iterations_upper_bound,
31153 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31155 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31156 * configure: Regenerate.
31157 * config.in: Regenerate.
31158 * config/i386/driver-mingw32.c: new file.
31159 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31160 * config.host: Link driver-mingw32.o on MinGW host.
31161 * doc/install.texi: Document new --enable-mingw-wildcard configure
31164 2017-05-04 Marek Polacek <polacek@redhat.com>
31166 PR tree-optimization/80612
31167 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31169 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31170 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31172 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31173 (movt splitter): Likewise.
31174 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31175 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31176 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31177 block for Thumb-1 with MOVT.
31178 (thumb2_legitimate_address_p): Move code block ...
31179 (can_avoid_literal_pool_for_label_p): ... into this new function.
31180 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31182 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31183 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31184 "M-profile targets with the MOVT instruction".
31186 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31188 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31189 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31190 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31192 2017-05-04 Martin Liska <mliska@suse.cz>
31194 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31195 variable cond_code.
31197 2017-05-04 Richard Biener <rguenther@suse.de>
31199 * tree.c (array_at_struct_end_p): Handle arrays at struct
31200 end with flexarrays more conservatively. Refactor and treat
31201 arrays of arrays or aggregates more strict. Fix
31202 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31203 * tree.c (array_at_struct_end_p): Adjust prototype.
31204 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31205 * gimple-fold.c (get_range_strlen): Likewise.
31206 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31208 2017-05-04 Richard Biener <rguenther@suse.de>
31210 PR tree-optimization/31130
31211 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31213 (supports_overflow_infinity): Likewise.
31214 (is_negative_overflow_infinity): Likewise.
31215 (is_positive_overflow_infinity): Likewise.
31216 (is_overflow_infinity): Likewise.
31217 (stmt_overflow_infinity): Likewise.
31218 (overflow_infinity_range_p): Likewise.
31219 (usable_range_p): Remove as always returning true.
31220 (make_overflow_infinity): Remove.
31221 (negative_overflow_infinity): Likewise.
31222 (positive_overflow_infinity): Likewise.
31223 (avoid_overflow_infinity): Likewise.
31224 (set_value_range): Adjust accordingly.
31225 (set_value_range_to_nonnegative): Likewise, remove now unused
31226 overflow_infinity arg.
31227 (vrp_operand_equal_p): Adjust.
31228 (update_value_range): Likewise.
31229 (range_int_cst_singleton_p): Likewise.
31230 (operand_less_p): Likewise.
31231 (compare_values_warnv): Likewise.
31232 (extract_range_for_var_from_comparison_expr): Likewise.
31233 (vrp_int_const_binop): Likewise.
31234 (zero_nonzero_bits_from_vr): Likewise.
31235 (extract_range_from_multiplicative_op_1): Likewise.
31236 (extract_range_from_binary_expr_1): Likewise.
31237 (extract_range_from_unary_expr): Likewise.
31238 (extract_range_from_comparison): Likewise.
31239 (extract_range_basic): Likewise.
31240 (adjust_range_with_scev): Likewise.
31241 (compare_ranges): Likewise.
31242 (compare_range_with_value): Likewise.
31243 (dump_value_range): Likewise.
31244 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31246 (simplify_cond_using_ranges): Adjust.
31248 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31250 * brig-builtins.def: Added a builtin for class_f64.
31251 * builtin-types.def: Added a builtin type needed by class_f64.
31253 2017-05-03 Jason Merrill <jason@redhat.com>
31255 * timevar.def: Add TV_CONSTEXPR.
31257 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31259 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31261 2017-05-03 Martin Jambor <mjambor@suse.cz>
31263 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31264 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31265 * ipa-cp.c (ipcp_propagate_stage): Do not call
31266 ipa_update_after_lto_read.
31267 * ipa-inline.c (ipa_inline): Likewise.
31269 2017-05-03 Martin Jambor <mjambor@suse.cz>
31271 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31272 tag. Added a default constructor and a destructor.
31273 (ipa_edge_args_sum_t): New class;
31274 (ipa_edge_args_sum): Declare.
31275 (ipa_edge_args_vector): Remove declaration.
31276 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31277 (ipa_free_edge_args_substructures): Remove declaration.
31278 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31279 (ipa_edge_args_info_available_for_edge_p): Likewise.
31280 * ipa-prop.c (ipa_edge_args_vector): Removed.
31281 (edge_removal_hook_holder): Likewise.
31282 (edge_duplication_hook_holder): Likewise.
31283 (ipa_edge_args_sum): New variable.
31284 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31285 ipa_edge_args_vector.
31286 (ipa_free_edge_args_substructures): Likewise.
31287 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31288 ipa_edge_args_vector.
31289 (ipa_edge_removal_hook): Turned into method
31290 ipa_edge_args_sum_t::remove.
31291 (ipa_edge_duplication_hook): Turned into method
31292 ipa_edge_args_sum_t::duplicate.
31293 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31294 registering edge hooks.
31295 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31296 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31297 ipa_edge_args_sum instead of ipa_edge_args_vector.
31298 * ipa-profile.c (ipa_profile): Likewise.
31300 2017-05-03 Martin Jambor <mjambor@suse.cz>
31302 * symbol-summary.h (function_summary): New method exists.
31303 (function_summary::symtab_removal): Deallocate through release.
31304 (call_summary): New class.
31305 (gt_ggc_mx): New overload.
31306 (gt_pch_nx): Likewise.
31307 (gt_pch_nx): Likewise.
31309 2017-05-03 Jeff Law <law@redhat.com>
31311 PR tree-optimization/78496
31312 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31313 from simplify_cond_using_ranges. Split off code to walk
31314 backwards through casts into ...
31315 (simplify_cond_using_ranges_2): New function.
31316 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31317 (execute_vrp): After identifying jump threads, call
31318 simplify_cond_using_ranges_2.
31320 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31323 * ipa-inline.h (inline_summary): Add ctor.
31324 (create_ggc): Do not use ggc_cleared_alloc.
31326 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31327 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31329 * gcc.c (handle_braces): Support escaping in switch matching
31331 * doc/invoke.texi (Spec Files): Document it.
31332 Remove superfluous @code markup in items.
31334 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31336 * diagnostic-show-locus.c (struct column_range): New struct.
31337 (get_affected_columns): New function.
31338 (get_printed_columns): New function.
31339 (struct correction): New struct.
31340 (correction::ensure_capacity): New function.
31341 (correction::ensure_terminated): New function.
31342 (struct line_corrections): New struct.
31343 (line_corrections::~line_corrections): New dtor.
31344 (line_corrections::add_hint): New function.
31345 (layout::print_trailing_fixits): Reimplement in terms of the new
31347 (selftest::test_overlapped_fixit_printing): New function.
31348 (selftest::diagnostic_show_locus_c_tests): Call it.
31350 2017-05-03 Nathan Sidwell <nathan@acm.org>
31352 Canonicalize canonical type hashing
31353 * tree.h (type_hash_canon_hash): Declare.
31354 * tree.c (type_hash_list, attribute_hash_list): Move into
31355 type_hash_canon_hash.
31356 (build_type_attribute_qual_variant): Break out hash code calc into
31357 type_hash_canon_hash.
31358 (type_hash_canon_hash): New. Generic type hash computation.
31359 (build_range_type_1, build_array_type_1, build_function_type,
31360 build_method_type_directly, build_offset_type, build_complex_type,
31361 make_vector_type): Call it.
31363 2017-05-03 Richard Biener <rguenther@suse.de>
31365 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31366 When all DRs have unknown misaligned do not always peel
31367 when there is a store but apply the same costing model as if
31368 there were only loads.
31370 2017-05-03 Richard Biener <rguenther@suse.de>
31373 PR tree-optimization/80492
31374 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31375 compare_base_decls returning dont-know properly.
31377 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31379 * config/arm/iterators.md (CCSI): New mode iterator.
31380 (arch): New mode attribute.
31381 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31382 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31383 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31384 code iterator for success result mode.
31385 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31386 the corresponding new insn generators.
31388 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31391 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31392 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31394 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31396 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31397 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31398 (DDR_A): Wrap DDR argument in brackets.
31399 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31400 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31401 (DDR_REVERSED_P): Likewise.
31403 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31405 PR tree-optimization/79472
31406 * tree-switch-conversion.c (struct switch_conv_info): Add
31407 contiguous_range and default_case_nonstandard fields.
31408 (collect_switch_conv_info): Compute contiguous_range and
31409 default_case_nonstandard fields, don't clear final_bb if
31410 contiguous_range and only the default case doesn't have the required
31412 (check_all_empty_except_final): Set default_case_nonstandard instead
31413 of failing if contiguous_range and the default case doesn't have empty
31415 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31416 and only the default case doesn't have the required constants. Skip
31418 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31419 if default_case_nonstandard.
31420 (build_constructors): Build constant 1 just once. Assert that default
31421 values aren't inserted in between cases if contiguous_range. Skip
31423 (build_arrays): Skip virtual phis.
31424 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31425 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31426 Handle virtual phis.
31427 (gen_inbound_check): Handle default_case_nonstandard case.
31428 (process_switch): Adjust check_final_bb caller. Call
31429 gather_default_values with the first non-default case instead of
31430 default case if default_case_nonstandard.
31432 2017-05-02 Nathan Sidwell <nathan@acm.org>
31434 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31435 check. Fix formatting.
31437 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31439 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31440 errors when comparing specialized and unspecialized times.
31442 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31444 * diagnostic-show-locus.c
31445 (layout::should_print_annotation_line_p): Make private.
31446 (layout::print_annotation_line): Make private.
31447 (layout::annotation_line_showed_range_p): Make private.
31448 (layout::show_ruler): Make private.
31449 (layout::print_source_line): Make private. Pass in line and
31450 line_width, rather than calling location_get_source_line. Drop
31452 (layout::print_leading_fixits): New method.
31453 (layout::print_any_fixits): Rename to...
31454 (layout::print_trailing_fixits): ...this, and make private.
31455 Don't print newline fixits.
31456 (diagnostic_show_locus): Move logic for printing one row into...
31457 (layout::print_line): ...this new function. Move the
31458 location_get_source_line call and error-handling from
31459 print_source_line to here. Call print_leading_fixits, and rename
31460 print_any_fixits to print_trailing_fixits.
31461 (selftest::test_fixit_insert_containing_newline): Update now that
31462 newlines are partially supported.
31463 (selftest::test_fixit_insert_containing_newline_2): New test.
31464 (selftest::test_fixit_replace_containing_newline): Update comments.
31465 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31466 * edit-context.c (class added_line): New class.
31467 (class edited_line): Describe newline handling in comment.
31468 (edited_line::actually_edited_p): New method.
31469 (edited_line::print_content): Delete redundant decl.
31470 (edited_line::m_predecessors): New field.
31471 (edited_file::print_content): Call edited_line::print_content.
31472 (edited_file::print_diff): Update to support newlines.
31473 (edited_file::print_diff_hunk): Likewise.
31474 (edited_file::print_run_of_changed_lines): New function.
31475 (edited_file::print_diff_line): Convert to...
31476 (print_diff_line): ...this.
31477 (edited_file::get_effective_line_count): New function.
31478 (edited_line::edited_line): Initialize new field m_predecessors.
31479 (edited_line::~edited_line): Clean up m_predecessors.
31480 (edited_line::apply_fixit): Handle newlines.
31481 (edited_line::get_effective_line_count): New function.
31482 (edited_line::print_content): New function.
31483 (edited_line::print_diff_lines): New function.
31484 (selftest::test_applying_fixits_insert_containing_newline): New
31486 (selftest::test_applying_fixits_replace_containing_newline): New
31488 (selftest::insert_line): New function.
31489 (selftest::test_applying_fixits_multiple_lines): Add example of
31491 (selftest::edit_context_c_tests): Call the new tests.
31493 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31495 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31496 parameter cand. Update dump information.
31497 (get_computation_cost): Update uses.
31499 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31501 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31502 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31503 (get_computation_at, rewrite_use_address): Update use of
31504 get_computation_aff.
31506 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31508 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31509 (get_computation): Delete.
31510 (get_computation_cost): Implement like get_computation_cost_at.
31511 Use get_computation_at.
31512 (get_computation_cost_at): Delete.
31513 (rewrite_use_nonlinear_expr): Use get_computation_at.
31514 (rewrite_use_compare, remove_unused_ivs): Ditto.
31516 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31518 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31520 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31522 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31523 (ivopts_global_cost_for_size): Rename parameter and update uses.
31524 (iv_ca_recount_cost): Update uses.
31525 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31526 candidates seperately in n_invs and n_cands.
31527 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31529 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31531 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31532 (find_inv_vars_cb): New.
31533 (find_depends): Renamed to ...
31534 (find_inv_vars): ... this.
31535 (add_candidate_1, force_var_cost): Call find_inv_vars.
31536 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31538 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31540 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31541 inv_vars. Add inv_exprs.
31542 (struct iv_cand): Rename depends_on to inv_vars.
31543 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31544 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31545 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31546 (dump_cand): Dump inv_vars.
31547 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31548 (record_invariant, find_depends, add_candidate_1): Ditto.
31549 (set_group_iv_cost, force_var_cost): Ditto.
31550 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31551 (get_computation_cost_at, get_computation_cost): Ditto.
31552 (determine_group_iv_cost_generic): Ditto.
31553 (determine_group_iv_cost_address): Ditto.
31554 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31555 (determine_group_iv_costs): Ditto.
31556 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31557 (iv_ca_set_remove_invariants): Renamed to ...
31558 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31559 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31560 (iv_ca_set_add_invariants): Renamed to ...
31561 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31562 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31563 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31564 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31565 (create_new_ivs): Remove useless dump.
31567 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31569 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31571 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31572 (iv_ca_set_no_cp, create_new_iv): Ditto.
31574 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31576 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31578 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31580 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31581 function tree_check2.
31583 2017-05-02 Martin Liska <mliska@suse.cz>
31585 * doc/gcov.texi: Add missing preposition.
31586 * gcov.c (function_info::function_info): Properly fill up
31587 all member variables.
31589 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31591 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31593 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31595 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31597 2017-05-02 Martin Liska <mliska@suse.cz>
31600 * lto-streamer-in.c (lto_read_tree_1): Remove
31601 LTO_STREAMER_DEBUG.
31602 * lto-streamer.c (struct tree_hash_entry): Likewise.
31603 (struct tree_entry_hasher): Likewise.
31604 (tree_entry_hasher::hash): Likewise.
31605 (tree_entry_hasher::equal): Likewise.
31606 (lto_streamer_init): Likewise.
31607 (lto_orig_address_map): Likewise.
31608 (lto_orig_address_get): Likewise.
31609 (lto_orig_address_remove): Likewise.
31610 * lto-streamer.h: Likewise.
31611 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31612 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31614 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31616 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31617 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31618 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31619 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31620 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31621 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31622 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31623 (mm_maskz_sub_ss): New intrinsics.
31624 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31625 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31626 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31627 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31628 (__builtin_ia32_subss_mask_round): New builtins.
31629 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31630 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31631 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31633 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31634 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31636 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31639 2017-05-02 Martin Jambor <mjambor@suse.cz>
31641 PR tree-optimization/78687
31642 * tree-sra.c (access): New field parent.
31643 (process_subtree_disqualification): New function.
31644 (disqualify_candidate): Call it.
31645 (build_accesses_from_assign): Reset write flag if creating an
31647 (build_access_subtree): Fill in parent field and also prpagate
31648 down grp_write flag.
31649 (create_artificial_child_access): New parameter set_grp_write, set
31650 grp_write to its value.
31651 (propagate_subaccesses_across_link): Also propagate grp_write flag
31653 (propagate_all_subaccesses): Push the closest parent back to work
31654 queue if add_access_to_work_queue returned true.
31656 2017-05-02 Richard Biener <rguenther@suse.de>
31658 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31659 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31660 -fstrict-overflow documentation.
31661 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31662 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31663 flag_strict_overflow.
31664 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31665 * lto-opts.c (lto_write_options): Do not stream it.
31666 * lto-wrapper.c (merge_and_complain): Do not handle it.
31667 * opts.c (default_options_table): Do not set -fstrict-overflow.
31668 (finish_options): Likewise do not clear it when sanitizing.
31669 * simplify-rtx.c (simplify_const_relational_operation): Do not
31670 test flag_strict_overflow.
31672 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31674 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31675 using enabled attribute.
31676 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31677 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31678 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31679 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31680 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31681 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31682 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31683 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31684 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31685 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31687 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31689 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31691 2017-05-02 Richard Biener <rguenther@suse.de>
31693 PR tree-optimization/80591
31695 2017-04-10 Richard Biener <rguenther@suse.de>
31697 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31700 2017-05-02 Richard Biener <rguenther@suse.de>
31702 PR tree-optimization/80549
31703 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31704 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31705 headers if they do not have a preheader.
31707 2017-05-02 Martin Liska <mliska@suse.cz>
31710 * common.opt: Fix typo.
31711 * doc/invoke.texi: Likewise.
31713 2017-05-01 Jan Beulich <jbeulich@suse.com>
31715 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31716 swapping, add (x,x,m,x,n) alternative.
31718 2017-05-01 Nathan Sidwell <nathan@acm.org>
31720 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31721 unnecessary unadjusted_alignment check.
31723 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31726 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31728 * gimplify.c (gimplify_cilk_detach): New function.
31729 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31730 * tree-core.h: Document EXPR_CILK_SPAWN.
31731 * tree.h (EXPR_CILK_SPAWN): Define.
31733 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31735 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31736 to use new fixit_hint representation, using the "replace" logic.
31737 (get_line_span_for_fixit_hint): Likewise.
31738 (layout::print_any_fixits): Likewise.
31739 (selftest::test_one_liner_many_fixits): Rename to...
31740 (selftest::test_one_liner_many_fixits_1): ...this, and update
31741 comment and expected output to reflect that the multiple fix-it
31742 hints are now consolidated into one insertion.
31743 (selftest::test_one_liner_many_fixits_2): New test.
31744 (selftest::test_diagnostic_show_locus_one_liner): Update for
31746 (selftest::test_fixit_consolidation): Update for fix-it API
31748 * diagnostic.c (print_parseable_fixits): Likewise.
31749 * edit-context.c (edited_line::m_line_events): Convert from
31750 auto_vec <line_event *> to auto_vec <line_event>.
31751 (class line_event): Convert from abstract base class to a concrete
31752 class, taking over the role of replace_event.
31753 (class insert_event): Delete.
31754 (class replace_event): Rename to class line_event. Convert to
31756 (edit_context::add_fixits): Reimplement.
31757 (edit_context::apply_insert): Delete.
31758 (edit_context::apply_replace): Rename to...
31759 (edit_context::apply_fixit): ...this. Convert to half-open range.
31760 (edited_file::apply_insert): Delete.
31761 (edited_file::apply_replace): Rename to...
31762 (edited_file::apply_fixit): ...this.
31763 (edited_line::~edited_line): Drop deletion of events.
31764 (edited_line::apply_insert): Delete.
31765 (edited_line::apply_replace): Rename to...
31766 (edited_line::apply_fixit): ...this. Convert to half-open range.
31767 Update for change to type of m_line_events.
31768 * edit-context.h (edit_context::apply_insert): Delete.
31769 (edit_context::apply_replace): Rename to...
31770 (edit_context::apply_fixit): ...this.
31772 2017-05-01 Martin Sebor <msebor@redhat.com>
31774 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31777 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31780 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31781 __get_cpuid_max returns 0.
31782 (__get_cpuid_count): Ditto.
31784 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31786 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31787 replacement expression is another instance of one of its arguments.
31789 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31792 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31793 check for stack push/pop autoinc.
31794 * config/i386/i386.c (ix86_agi_dependent): Return false
31795 if the only reason why modified_in_p returned true is that
31796 addr is SP based and set_insn is a push or pop.
31798 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31800 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31803 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31806 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31807 (account_size_time): Use two predicates - exec_pred and
31809 (evaluate_conditions_for_known_args): Compute both clause and
31811 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31812 (inline_summary_t::duplicate): Update.
31813 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31815 (compute_inline_parameters): Likewise.
31816 (estimate_edge_size_and_time): Update caluclation of time.
31817 (estimate_node_size_and_time): Compute both time and nonspecialized
31819 (estimate_ipcp_clone_size_and_time): Update.
31820 (inline_merge_summary): Update.
31821 (do_estimate_edge_time): Update.
31822 (do_estimate_edge_size): Update.
31823 (do_estimate_edge_hints): Update.
31824 (inline_read_section, inline_write_summary): Stream both new predicates.
31825 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31827 (compute_inlined_call_time): Cleanup.
31828 (big_speedup_p): Update.
31829 (edge_badness): Update.
31830 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31831 (size_time_entry): Replace predicate by exec_predicate and
31832 nonconst_predicate.
31833 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31834 (estimate_edge_time): Return also nonspec_time.
31835 (reset_edge_growth_cache): Update.
31837 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31839 PR rtl-optimization/80491
31840 * ifcvt.c (noce_process_if_block): When looking for x setter
31841 with missing else_bb, don't check only the insn right before
31842 cond_earliest, but look for the last insn that x is modified in
31843 within the same bb.
31845 PR rtl-optimization/80491
31846 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31848 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31850 PR tree-optimization/80487
31851 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31853 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31855 PR tree-optimization/79697
31856 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31857 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31858 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31860 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31861 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31863 2017-04-28 Martin Sebor <msebor@redhat.com>
31865 PR tree-optimization/80523
31866 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31867 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31869 (maybe_warn, format_directive, parse_directive): Use new functions.
31870 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31872 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31874 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31876 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31878 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31879 target_header_dir): Set correctly.
31880 * configure: Regenerated.
31881 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31882 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31883 instead of SYSTEM_HEADER_DIR.
31885 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31887 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31888 (estimate_local_effects): Likewise.
31889 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31890 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31891 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31892 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31893 do_estimate_edge_time, estimate_edge_time): Likewise.
31894 * ipa-inline-analysis.c (estimate_node_size_and_time,
31895 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31896 (estimate_time_after_inlining): Remove.
31898 2017-04-28 Martin Liska <mliska@suse.cz>
31900 * doc/gcov.texi: Enhance documentation of gcov.
31902 2017-04-28 Martin Liska <mliska@suse.cz>
31904 * doc/gcov.texi: Sort options in alphabetic order.
31905 * doc/gcov-dump.texi: Likewise.
31906 * doc/gcov-tool.texi: Likewise.
31907 * gcov.c (print_usage): Likewise.
31908 * gcov-dump.c (print_usage): Likewise.
31909 * gcov-tool.c (print_merge_usage_message): Likewise.
31910 (print_rewrite_usage_message): Likewise.
31911 (print_overlap_usage_message): Likewise.
31913 2017-04-28 Martin Liska <mliska@suse.cz>
31915 PR gcov-profile/53915
31916 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31918 2017-04-28 Martin Liska <mliska@suse.cz>
31920 PR gcov-profile/79891
31921 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31922 is marked by compiler as living on a line.
31923 (get_cycles_count): Remove usage of the union.
31924 (output_intermediate_file): Likewise.
31925 (find_source): Fix GNU coding style.
31926 (accumulate_line_counts): Remove old non-all block mode.
31927 (output_lines): Remove usage of the union.
31928 * profile.c (output_location): Include all BBs, even if
31929 belonging to a same line (and file) as a previous BB.
31931 2017-04-28 Martin Liska <mliska@suse.cz>
31933 * gcov.c (process_args): Handle new argument 'w'.
31934 (read_graph_file): Assign ID to BBs.
31935 (output_branch_count): Display BB # if verbose flag is set.
31936 (output_lines): Likewise for arcs.
31937 (print_usage): Add '--verbose' option help.
31938 * doc/gcov.texi: Document --verbose (-w) option.
31940 2017-04-28 Martin Liska <mliska@suse.cz>
31942 * gcov.c (struct block_location_info): New struct.
31943 (process_file): Fill up the new structure.
31944 (read_graph_file): Replace usage of encoding by the newly added
31946 (add_line_counts): Likewise.
31947 (accumulate_line_counts): Remove usage of the union.
31948 (function_info::function_info): New function.
31949 (function_info::~function_info): Likewise.
31950 (process_file): Call delete instead of release_function.
31951 (release_function): Release the function.
31952 (release_structures): Call delete instead of release_function.
31953 (solve_flow_graph): Replace usage of num_blocks.
31954 (find_exception_blocks): Likewise.
31955 (output_lines): Fix GNU coding style.
31957 2017-04-28 Martin Liska <mliska@suse.cz>
31960 * coverage.c (coverage_remove_note_file): New function.
31961 * coverage.h: Declare the function.
31962 * toplev.c (finalize): Clean if an error has been seen.
31964 2017-04-28 Martin Liska <mliska@suse.cz>
31966 PR gcov-profile/80031
31967 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31968 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31969 * gcov.c (read_graph_file): Read just number of blocks.
31970 * profile.c (branch_prob): Do not stream 0 flags per a basic
31973 2017-04-28 Martin Liska <mliska@suse.cz>
31975 * gcov-dump.c (tag_*): Add new argument to declarations.
31976 (dump_gcov_file): Likewise.
31977 (tag_blocks): Add and use new argument depth.
31978 (tag_arcs): Likewise.
31979 (tag_lines): Likewise.
31980 (tag_counters): Likewise.
31981 (tag_summary): Likewise.
31982 (dump_working_sets): Use depth to do a proper indentation.
31984 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31987 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31988 * symtab.c (symtab_node::debug_symtab): Move definition here.
31990 2017-04-28 Richard Biener <rguenther@suse.de>
31992 * lto-streamer.h (LTO_major_version): Bump to 7.
31994 2017-04-28 Richard Biener <rguenther@suse.de>
31996 * tree-vrp.c (assert_info): New struct.
31997 (add_assert_info): New helper.
31998 (register_edge_assert_for_2): Refactor to add asserts to a vector
32000 (register_edge_assert_for_1): Likewise.
32001 (register_edge_assert_for): Likewise.
32002 (finish_register_edge_assert_for): New helper actually registering
32003 asserts where live on edge.
32004 (find_conditional_asserts): Adjust.
32005 (find_switch_asserts): Likewise.
32006 (evrp_dom_walker::try_find_new_range): Generalize.
32007 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32009 2017-04-27 Marek Polacek <polacek@redhat.com>
32012 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32013 arg10 and arg11 to itype.
32015 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32017 * doc/extend.texi (Object Size Checking): Improve grammar.
32019 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32022 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32023 that the logic for permitting reciprocal estimates matches that
32026 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32029 * tree.c (type_cache_hasher::equal): Only compare
32030 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32031 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32032 non-aggregate element types.
32033 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32034 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32036 2017-04-27 Richard Biener <rguenther@suse.de>
32038 PR middle-end/80533
32039 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32040 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32041 keeping a reference to a trailing array.
32043 2017-04-27 Richard Biener <rguenther@suse.de>
32045 PR middle-end/80539
32046 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32047 being in loop-closed SSA form conservatively.
32048 (chrec_fold_multiply_poly_poly): Likewise.
32050 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32052 PR middle-end/79665
32053 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32054 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32056 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32059 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32060 (aarch64_function_arg_alignment): Return unsigned int again, but still
32061 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32062 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32063 Don't emit -Wpsabi note.
32064 (aarch64_function_arg_boundary): Likewise.
32065 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32068 2017-04-26 Nathan Sidwell <nathan@acm.org>
32070 * tree.h (crc32_unsigned_n): Declare.
32071 (crc32_unsigned, crc32_unsigned): Make inline.
32072 * tree.c (crc32_unsigned_bits): Replace with ...
32073 (crc32_unsigned_n): ... this.
32074 (crc32_unsigned, crc32_byte): Remove.
32075 (crc32_string): Remove unnecessary braces.
32077 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32079 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32080 * ipa-inline-analysis.c (MAX_TIME): Remove.
32081 (account_size_time): Use sreal for time.
32082 (dump_inline_summary): Update.
32083 (estimate_function_body_sizes): Update.
32084 (estimate_edge_size_and_time): Update.
32085 (estimate_calls_size_and_time): Update.
32086 (estimate_node_size_and_time): Update.
32087 (inline_merge_summary): Update.
32088 (inline_update_overall_summary): Update.
32089 (estimate_time_after_inlining): Update.
32090 (inline_read_section): Update.
32091 (inline_write_summary): Update.
32092 * ipa-inline.c (compute_uninlined_call_time): Update.
32093 (compute_inlined_call_time): Update.
32094 (recursive_inlining): Update.
32095 (inline_small_functions): Update.
32096 (dump_overall_stats): Update.
32097 * ipa-inline.h: Include sreal.h.
32098 (size_time_entry): Turn time to sreal.
32099 (inline_summary): Turn self_time nad time to sreal.
32101 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32103 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32105 (sreal::stream_out, sreal::stream_in): New.
32106 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32108 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32110 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32113 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32116 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32118 (dimode_scalar_chain::compute_convert_gain): Ditto.
32119 (dimode_scalar_chain::make_vector_copies): Ditto.
32120 (dimode_scalar_chain::convert_reg): Ditto.
32121 (dimode_scalar_chain::convert_insn): Ditto.
32122 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32123 (VI248_AVX512BW_1): New mode iterator.
32124 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32125 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32128 2017-04-25 Martin Sebor <msebor@redhat.com>
32130 PR tree-optimization/80497
32131 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32132 constants are representable in HOST_WIDE_INT.
32133 (parse_directive): Ditto.
32135 2017-04-25 Martin Sebor <msebor@redhat.com>
32138 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32139 (new_zero_array): Adjust signature.
32140 (dom_info::dom_init): Used unsigned rather that size_t.
32141 (dom_info::dom_info): Same.
32143 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32144 Jakub Jelinek <jakub@redhat.com>
32147 * config/arm/arm.c: Include gimple.h.
32148 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32149 returns negative, increment ncrn only if it returned positive.
32150 (arm_needs_doubleword_align): Return int instead of bool,
32151 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32152 members, but if there is any such non-FIELD_DECL
32153 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32154 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32155 returns negative, increment nregs only if it returned positive.
32156 (arm_setup_incoming_varargs): Likewise.
32157 (arm_function_arg_boundary): Emit -Wpsabi note if
32158 arm_needs_doubleword_align returns negative, return
32159 DOUBLEWORD_ALIGNMENT only if it returned positive.
32161 2017-04-25 Marek Polacek <polacek@redhat.com>
32164 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32165 first argument to type.
32167 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32170 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32171 type checks to test for compatibility instead of equality.
32173 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32174 Jakub Jelinek <jakub@redhat.com>
32177 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32179 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32180 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32181 the alignment computation, but return their maximum in warn_alignment.
32182 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32183 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32185 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32186 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32189 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32191 * config/arc/simdext.md (dmpyh): Fix typo.
32193 2017-04-25 Richard Biener <rguenther@suse.de>
32195 PR tree-optimization/80492
32196 * alias.c (compare_base_decls): Handle registers with asm
32197 specification conservatively.
32198 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32199 compare_base_decls returning dont-know properly.
32201 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32203 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32204 (legitimate_offset_address_p): New function.
32205 (arc_legitimate_address_p): Use above function.
32207 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32209 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32211 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32213 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32214 ACCH registers whenever they are available.
32216 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32218 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32219 double regs fix when not used.
32221 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32223 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32225 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32226 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32228 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32230 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32231 when emitting short ADD instructions.
32233 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32235 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32237 (cmpsi_cc_c_insn): Likewise.
32238 (cbranchsi4_scratch): Compute proper instruction length using
32239 compact_hreg_operand.
32240 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32242 2017-04-25 Richard Biener <rguenther@suse.de>
32244 PR middle-end/80509
32245 * passes.c (pass_manager::pass_manager): Initialize
32246 m_name_to_pass_map.
32248 2017-04-25 Richard Biener <rguenther@suse.de>
32250 PR tree-optimization/79201
32251 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32253 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32256 * config/s390/vector.md: Split MEM->GPR vector moves for
32257 non-s_operand addresses.
32259 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32262 * config/s390/predicates.md (reload_const_wide_int_operand): New
32264 * config/s390/s390.md ("movti"): Remove d/P alternative.
32265 ("movti_bigconst"): New pattern definition.
32267 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32270 * s390-protos.h (s390_expand_cs_hqi): Removed.
32271 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32272 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32273 modes as well as CCZ1mode and CCZmode.
32274 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32275 signature of s390_emit_compare_and_swap.
32276 (s390_expand_cs_hqi): Likewise, make static.
32277 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32278 compare-and-swap, in some cases.
32279 (s390_expand_cs): Wrapper function.
32280 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32282 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32283 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32284 patterns for small and large integers. Forbid symref memory operands.
32285 Move expander to s390.c. Require cc register.
32286 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32287 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32288 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32289 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32290 symref memory operands. Remove CC mode and call s390_match_ccmode
32292 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32294 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32296 * config/s390/s390.md (define_peephole2): New peephole to help
32297 combining the load-and-test pattern with volatile memory.
32299 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32301 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32302 with CCZmode for TARGET_Z196.
32304 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32306 PR rtl-optimization/80501
32307 * combine.c (make_compound_operation_int): Set subreg_code to SET
32308 even for AND with mask of the sign bit of mode.
32310 PR rtl-optimization/80500
32311 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32312 sum's initial value.
32314 2017-04-25 Julian Brown <julian@codesourcery.com>
32315 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32317 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32319 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32321 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32323 2017-04-25 Julian Brown <julian@codesourcery.com>
32324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32326 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32327 (thunderx2t99_sha): New Reservation.
32329 2017-04-25 Julian Brown <julian@codesourcery.com>
32330 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32332 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32333 type for 1-element load.
32335 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32337 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32339 2017-04-24 Martin Jambor <mjambor@suse.cz>
32341 PR tree-optimization/80293
32342 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32343 char arrays not totally scalarizable if it is false.
32344 (analyze_all_variable_accesses): Pass correct value in the new
32345 parameter. Add a statistics counter.
32347 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32349 PR middle-end/79931
32350 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32352 2017-04-24 Richard Biener <rguenther@suse.de>
32354 PR tree-optimization/80494
32355 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32356 out for complex types.
32358 2017-04-24 Richard Biener <rguenther@suse.de>
32360 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32361 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32362 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32363 (DFS): Adjust and never fail.
32364 (sccvn_dom_walker::fail): Remove.
32365 (sccvn_dom_walker::before_dom_children): Adjust.
32366 (run_scc_vn): Likewise and never fail.
32367 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32368 (pass_fre::execute): Likewise.
32370 2017-04-24 Richard Biener <rguenther@suse.de>
32372 PR tree-optimization/79725
32373 * tree-ssa-sink.c (statement_sink_location): Return whether
32374 failure reason was zero uses. Move that check later.
32375 (sink_code_in_bb): Deal with zero uses by removing the stmt
32378 2017-04-24 Richard Biener <rguenther@suse.de>
32381 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32382 pointer-based references.
32384 2017-04-24 Richard Biener <rguenther@suse.de>
32387 * pass_manager.h (pass_manager::operator new): Remove.
32388 (pass_manager::operator delete): Likewise.
32389 * passes.c (pass_manager::operator new): Remove.
32390 (pass_manager::operator delete): Likewise.
32391 (pass_manager::pass_manager): Zero individual pass members.
32393 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32396 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32397 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32398 Check "XEXP (src, 1)" operand here.
32399 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32400 Check "XEXP (src, 1)" operand here.
32401 (dimode_scalar_chain::make_vector_copies): Detect count register
32402 of a shift instruction. Zero extend count register from QImode
32403 to DImode to satisfy vector shift pattern count operand predicate.
32404 Substitute vector shift count operand with a DImode copy.
32405 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32408 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32410 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32411 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32412 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32413 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32414 (UNSPEC_NOREX_MEM): Remove definition.
32416 2017-04-21 Richard Biener <rguenther@suse.de>
32418 PR tree-optimization/79547
32419 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32420 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32421 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32422 without any constraints.
32424 2017-04-21 Richard Biener <rguenther@suse.de>
32426 PR tree-optimization/78847
32427 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32429 2017-04-21 Richard Biener <rguenther@suse.de>
32431 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32432 (build_distinct_type_copy): Likewise.
32433 (build_variant_type_copy): Likewise.
32434 * tree.c (build_qualified_type): Pass down mem-stat info.
32435 (build_distinct_type_copy): Likewise.
32436 (build_variant_type_copy): Likewise.
32438 2017-04-21 Richard Biener <rguenther@suse.de>
32440 PR tree-optimization/80237
32441 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32443 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32444 for a simplified result.
32446 2016-04-21 Richard Biener <rguenther@suse.de>
32448 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32449 sth as strict as a simple_iv but a chrec without symbols and an
32450 operand defined in the loop we are peeling (and not some subloop).
32451 (propagate_constants_for_unrolling): Propagate all constants.
32453 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32456 * config/i386/i386.c (print_reg): Remove assert for disalowed
32457 regno values, call output_operand_lossage instead.
32459 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32462 * config/i386/constraints.md (Yc): New register constraint.
32463 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32464 Use Yc constraint for alternative 2 of operand 0. Remove
32465 preferred_for_speed attribute.
32467 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32469 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32470 lastprivate clauses in SIMT case.
32472 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32474 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32476 2017-04-20 Richard Biener <rguenther@suse.de>
32478 PR tree-optimization/57796
32479 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32480 as N scalar stores.
32481 (vect_model_load_cost): Cost gathers as N scalar loads.
32483 2017-04-20 Richard Biener <rguenther@suse.de>
32485 * ggc-page.c (ggc_allocated_p): Rename to ...
32486 (safe_lookup_page_table_entry): ... this and return the lookup
32488 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32490 2017-04-20 Richard Biener <rguenther@suse.de>
32492 PR tree-optimization/80453
32493 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32494 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32495 from the conditions.
32496 (vn_phi_eq): Pass them down.
32497 (vn_phi_lookup): Record them.
32498 (vn_phi_insert): Likewise.
32500 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32502 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32503 uninitialized variable warning to avoid buffer overrun.
32505 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32508 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32509 is suppressed for '{ 0 }' in C.
32511 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32513 * BASE-VER: Set to 8.0.0.
32515 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32517 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32518 priority .init_array and .fini_array section with SECTION_NOTYPE
32521 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32523 PR middle-end/80423
32524 * tree.h (build_array_type): Add typeless_storage default argument.
32525 * tree.c (type_cache_hasher::equal): Also compare
32526 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32527 (build_array_type): Add typeless_storage argument, set
32528 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32530 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32531 (build_array_type): Likewise. Add typeless_storage argument.
32533 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32534 Jakub Jelinek <jakub@redhat.com>
32536 PR tree-optimization/80426
32537 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32538 operation on symbolic operands, also compute the overflow for the
32539 invariant part when the operation degenerates into a negation.
32541 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32544 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32545 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32548 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32550 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32553 * config/avr/avr.c (tree.h): Include it.
32554 (cgraph.h): Include it.
32555 (avr_encode_section_info): Don't warn for uninitialized progmem
32556 variable if it's just an alias.
32558 2017-04-19 Richard Biener <rguenther@suse.de>
32561 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32562 when needed by AutoPGO.
32564 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32567 * doc/lto.texi: Remove an extra 'that'.
32569 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32571 PR rtl-optimization/80429
32572 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32573 are only used in debug insns.
32575 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32576 Vladimir Makarov <vmakarov@redhat.com>
32578 * config/sparc/predicates.md (input_operand): Add comment. Return
32579 true for any memory operand when LRA is in progress.
32580 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32582 2017-04-18 Jeff Law <law@redhat.com>
32585 * mips.md ({return,simple_return}_internal): Do not overwrite
32588 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32590 PR tree-optimization/80443
32591 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32592 instead of adding 1, subtract -1 and similarly instead of subtracting
32595 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32597 PR rtl-optimization/80357
32598 * haifa-sched.c (tmp_bitmap): New variable.
32599 (model_recompute): Handle duplicate use records.
32600 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32601 (free_global_sched_pressure_data): Free it.
32603 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32606 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32607 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32608 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32609 instead of SYSTEM_HEADER_DIR.
32611 2017-04-18 Jeff Law <law@redhat.com>
32613 PR middle-end/80422
32614 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32615 predecessors after walking up the insn chain.
32617 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32620 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32621 sizetype type into debug info.
32623 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32626 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32627 unneeded test for TARGET_UPPER_REGS_SF.
32628 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32630 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32633 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32634 instead of gsi_after_labels.
32636 2017-04-18 Jeff Law <law@redhat.com>
32638 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32642 2017-04-13 Jeff Law <law@redhat.com>
32643 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32644 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32646 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32649 * config/avr/avr.c (intl.h): Include it.
32650 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32652 2017-04-18 Martin Liska <mliska@suse.cz>
32654 PR gcov-profile/78783
32655 * gcov-tool.c (gcov_output_files): Validate that destination
32656 file is either removed by the tool or by a user.
32658 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32659 Guy Benyei <guybe@mellanox.com>
32661 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32662 block, and do not negate it, the stored id is already negative.
32664 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32666 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32668 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32671 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32672 masks of options that should be turned off if the VSX vector
32673 options are turned off.
32674 (OTHER_P8_VECTOR_MASKS): Likewise.
32675 (OTHER_VSX_VECTOR_MASKS): Likewise.
32676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32677 rs6000_disable_incompatible_switches to validate no type switches
32679 (rs6000_incompatible_switch): New function to disallow turning on
32680 other vector options if -mno-vsx, -mno-power8-vector, or
32681 -mno-power9-vector are specified.
32683 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32685 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32687 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32689 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32690 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32691 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32692 (ARG_POINTER_CFA_OFFSET): Likewise.
32694 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32696 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32697 conditions to take advantage of various optimizations.
32699 2017-04-13 Jeff Law <law@redhat.com>
32701 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32702 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32703 (zero_extendsidi2_dext): Likewise.
32705 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32708 * fold-const.c (fold_ternary_loc): Revert
32709 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32712 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32714 PR rtl-optimization/80343
32715 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32716 new scratch pseudo.
32718 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32721 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32722 to ubsan_encode_value.
32724 2017-04-13 Jeff Law <law@redhat.com>
32726 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32727 appearing in DEBUG_INSNs.
32729 2017-04-13 Martin Liska <mliska@suse.cz>
32731 PR gcov-profile/80413
32732 * gcov-io.c (gcov_write_string): Copy to buffer just when
32733 allocated size is greater than zero.
32735 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32738 * dwarf2out.c (decls_for_scope): Ignore declarations of
32739 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32741 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32744 * ipa-visibility.c (non_local_p): Fix typos.
32745 (localize_node): When localizing symbol in same comdat group,
32746 dissolve the group only when we know external symbols are going
32748 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32750 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32752 PR tree-optimization/79390
32753 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32754 order does not result in usable sequence, retry with reversed operand
32760 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32761 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32762 op0 instead of fold_convert_loc (loc, type, arg0).
32764 2017-04-12 Jeff Law <law@redhat.com>
32766 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32767 has a delay slot in the generated code.
32769 * config/cris/cris.md (cris_preferred_reload_class): Return
32770 GENNONACR_REGS rather than GENERAL_REGS.
32772 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32775 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32776 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32777 signedness of the result type.
32779 2017-04-12 Richard Biener <rguenther@suse.de>
32780 Jeff Law <law@redhat.com>
32782 PR tree-optimization/80359
32783 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32784 trim stores to TARGET_MEM_REFs.
32786 2017-04-12 Richard Biener <rguenther@suse.de>
32788 PR tree-optimization/79390
32789 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32790 threading case even more.
32792 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32795 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32796 for quad_address_p for TImode, instead of just not indexed_address.
32798 2017-04-12 Richard Biener <rguenther@suse.de>
32799 Bernd Edlinger <bernd.edlinger@hotmail.de>
32801 PR middle-end/79671
32802 * alias.c (component_uses_parent_alias_set_from): Handle
32803 TYPE_TYPELESS_STORAGE.
32804 (get_alias_set): Likewise.
32805 * tree-core.h (tree_type_common): Add typeless_storage flag.
32806 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32807 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32808 for types containing members with TYPE_TYPELESS_STORAGE.
32809 (place_field): Likewise.
32810 (layout_type): Likewise for ARRAY_TYPE.
32811 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32812 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32813 TYPE_TYPELESS_STORAGE.
32814 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32816 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32819 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32820 first argument to type.
32822 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32826 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32827 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32828 (rs6000_expand_binop_builtin): Likewise.
32829 (rs6000_expand_ternop_builtin): Likewise; also add missing
32830 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32831 vshasigma built-ins.
32832 * doc/extend.texi: Document that vec_xxpermdi's third argument
32833 must be a constant.
32835 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32837 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32838 Use shift_const cost parameter when calculating gain of STV shifts.
32840 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32842 PR rtl-optimization/70478
32843 * lra-constraints.c (process_alt_operands): Check memory for
32844 disfavoring memory insn operand.
32846 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32848 PR middle-end/80100
32849 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32850 left shift in unsigned HOST_WIDE_INT type.
32852 PR rtl-optimization/80385
32853 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32854 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32857 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32858 if they have any depend clauses.
32860 2017-04-11 Martin Liska <mliska@suse.cz>
32863 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32864 * ipa-split.c (split_function): Create a local comdat symbol
32865 if caller is in a comdat group.
32867 2017-04-11 Martin Liska <mliska@suse.cz>
32870 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32873 2017-04-11 Martin Sebor <msebor@redhat.com>
32875 PR middle-end/80364
32876 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32877 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32879 (directive::set_width, directive::set_precision, format_character):
32881 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32884 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32887 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32888 conflict, set target->arch_name instead of target->cpu_name.
32890 2017-04-11 Richard Biener <rguenther@suse.de>
32892 PR tree-optimization/80374
32893 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32894 build_zero_cst, remove fold_convertible_p check again.
32896 2017-04-11 Martin Liska <mliska@suse.cz>
32899 * ubsan.c (instrument_object_size): Do not instrument register
32902 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32905 * config/i386/i386-builtin-types.def
32906 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32907 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32908 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32909 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32910 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32911 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32912 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32913 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32914 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32915 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32916 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32917 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32918 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32919 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32920 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32921 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32922 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32923 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32924 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32925 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32926 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32927 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32928 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32929 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32930 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32931 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32932 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32933 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32934 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32935 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32936 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32937 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32938 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32939 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32940 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32941 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32942 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32943 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32944 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32945 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32946 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32947 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32948 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32949 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32950 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32951 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32953 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32954 flag to second_arg_count, handle 4 argument function type _COUNT
32955 aliases, handle second_arg_count on second argument rather than last.
32957 2017-04-10 Jeff Law <law@redhat.com>
32959 PR tree-optimization/80374
32960 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32961 record anything if we can not convert integer_zero_node to the
32964 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32967 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32968 Enhance special handling given to the TARGET_P9_MINMAX option in
32969 relation to certain other options.
32971 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32973 PR tree-optimization/80153
32974 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32975 remove POINTER_PLUS_EXPR's base part directly, rather than through
32978 2017-04-10 Richard Biener <rguenther@suse.de>
32979 Bin Cheng <bin.cheng@arm.com>
32981 PR tree-optimization/80153
32982 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32983 the first element of pointer type aff_tree. Build result expr in
32985 (add_elt_to_tree): Convert to type unconditionally. Remove other
32986 fold_convert calls.
32987 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32988 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32990 2017-04-10 Richard Biener <rguenther@suse.de>
32992 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32995 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32997 PR rtl-optimization/70478
32998 * lra-constraints.c (curr_small_class_check): New.
32999 (update_and_check_small_class_inputs): New.
33000 (process_alt_operands): Update curr_small_class_check. Disfavor
33001 alternative insn memory operands. Check available regs for small
33004 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33007 * config/mips/mips.opt (-mvirt): Update description.
33008 * doc/invoke.texi (-mvirt): Likewise.
33010 2017-04-10 Richard Biener <rguenther@suse.de>
33012 PR middle-end/80362
33013 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33014 looking for NEGATE_EXPR in -A / -B to A / B folding.
33016 2017-04-10 Martin Liska <mliska@suse.cz>
33018 PR gcov-profile/80224
33019 * gcov.c (print_usage): Fix usage string.
33020 (get_gcov_intermediate_filename): Remove.
33021 (output_gcov_file): Use both for normal and intermediate format.
33022 (generate_results): Do not initialize special file for
33023 intermediate format.
33025 2017-04-10 Richard Biener <rguenther@suse.de>
33027 PR tree-optimization/80304
33028 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33031 2017-04-10 Nathan Sidwell <nathan@acm.org>
33034 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33035 (rs6000_init_builtins): Use it.
33037 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33039 * config/arm/arm.md (<mrc>): Add mode to SET source.
33040 (<mrrc>): Likewise.
33042 2017-04-10 Richard Biener <rguenther@suse.de>
33044 PR middle-end/80344
33045 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33047 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33050 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33051 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33052 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33053 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33054 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33055 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33056 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33057 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33058 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33059 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33060 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33061 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33062 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33063 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33064 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33065 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33066 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33067 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33068 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33069 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33070 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33071 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33072 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33074 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33076 PR rtl-optimization/70478
33077 * lra-constraints.c: Reverse the last patch.
33079 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33081 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33082 Add comment for WCHAR_T.
33084 2017-04-08 Martin Liska <mliska@suse.cz>
33087 2017-04-07 Martin Liska <mliska@suse.cz>
33090 * ipa-split.c (split_function): Add function part to a same comdat
33093 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33096 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33098 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33100 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33101 (vec_pairing): ...to this.
33102 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33103 (rs6000_sched_init): Adjust for name change.
33104 (struct rs6000_sched_context): Likewise.
33105 (rs6000_init_sched_context): Likewise.
33106 (rs6000_set_sched_context): Likewise.
33108 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33114 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33116 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33117 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33118 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33120 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33122 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33124 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33126 PR rtl-optimization/70703
33127 * ira-color.c (update_conflict_hard_regno_costs): Use
33128 int64_t instead of HOST_WIDE_INT.
33130 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33132 PR rtl-optimization/70478
33133 * lra-constraints.c (process_alt_operands): Disfavor alternative
33134 insn memory operands.
33136 2017-04-07 Jeff Law <law@redhat.com>
33138 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33139 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33141 2017-04-07 Martin Liska <mliska@suse.cz>
33144 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33145 Show error message instead of an ICE.
33147 2017-04-07 Martin Liska <mliska@suse.cz>
33150 * ipa-split.c (split_function): Add function part to a same comdat
33153 2017-04-07 Richard Biener <rguenther@suse.de>
33155 PR middle-end/80341
33156 * tree.c (get_unwidened): Also handle ! for_type case for
33158 * convert.c (do_narrow): Split out from ...
33159 (convert_to_integer_1): ... here. Do not pass final truncation
33160 type to get_unwidened for TRUNC_DIV_EXPR.
33162 2017-04-07 Richard Biener <rguenther@suse.de>
33164 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33166 (aff_combination_const): Adjust.
33167 (aff_combination_scale): Likewise.
33168 (aff_combination_add_elt): Likewise.
33169 (aff_combination_add_cst): Likewise.
33170 (aff_combination_convert): Likewise.
33171 (add_elt_to_tree): Likewise. Remove unused argument.
33172 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33174 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33176 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33178 * config/arm/arm.c (arm_default_short_enums): Use
33179 ARM_DEFAULT_SHORT_ENUMS.
33180 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33182 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33185 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33186 members with redundant out-of-class redeclaration.
33188 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33191 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33192 * config/i386/i386.md (*zero_extendsidi2):
33193 Add (?*x,*x) and (?*v,*v) alternatives.
33195 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33198 * config/i386/i386.c (ix86_expand_builtin)
33199 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33200 mode from insn data. Convert operands to insn operand mode.
33201 Copy operands that don't satisfy insn predicate to a register.
33203 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33205 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33208 2017-04-06 Richard Biener <rguenther@suse.de>
33210 PR tree-optimization/80334
33211 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33212 preserve alignment of accesses.
33214 2017-04-06 Richard Biener <rguenther@suse.de>
33216 PR tree-optimization/80262
33217 * tree-sra.c (build_ref_for_offset): Preserve address-space
33219 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33220 Drop useless address-space information on MEM_REF offsets.
33222 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33224 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33226 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33228 PR rtl-optimization/70703
33229 * ira-color.c (update_conflict_hard_regno_costs): Use
33230 HOST_WIDE_INT instead of long.
33232 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33235 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33236 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33238 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33239 for x86_64 target. Handle -m3dnowa option.
33241 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33243 PR rtl-optimization/70703
33244 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33245 (update_conflict_hard_regno_costs): Use long instead of unsigned
33246 arithmetic for cost calculation.
33248 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33249 Bernd Edlinger <bernd.edlinger@hotmail.de>
33252 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33255 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33258 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33259 ptr_mode with Pmode throughout.
33260 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33261 into probe_stack_range and use DImode.
33263 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33266 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33267 call_eh_return is true.
33269 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33271 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33272 Initialize last_match_fntype_index.
33274 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33277 * tree-nvr.c: Include internal-fn.h.
33278 (pass_return_slot::execute): Ignore internal calls without
33281 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33282 Richard Biener <rguenther@suse.de>
33285 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33286 captures used multiple times, except for the last use.
33287 * generic-match-head.c: Include gimplify.h.
33289 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33291 PR tree-optimization/79390
33292 * target.h (struct noce_if_info): Declare.
33293 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33294 * target.def (noce_conversion_profitable_p): New target hook.
33295 * ifcvt.h (struct noce_if_info): New type, moved from ...
33296 * ifcvt.c (struct noce_if_info): ... here.
33297 (noce_conversion_profitable_p): Renamed to ...
33298 (default_noce_conversion_profitable_p): ... this. No longer
33300 (noce_try_store_flag_constants, noce_try_addcc,
33301 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33302 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33303 instead of noce_conversion_profitable_p.
33304 * config/i386/i386.c: Include ifcvt.h.
33305 (ix86_option_override_internal): Don't override
33306 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33307 (ix86_noce_conversion_profitable_p): New function.
33308 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33309 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33310 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33311 * doc/tm.texi: Regenerated.
33313 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33315 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33318 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33321 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33322 instructions for small multiply cores.
33324 2017-04-04 Jeff Law <law@redhat.com>
33326 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33328 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33329 that are not set by the loop over the elements.
33331 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33334 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33335 int mode, convert_modes it to mode as unsigned, otherwise use
33336 lowpart_subreg to mode rather than SImode.
33337 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33338 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33339 Use DImode instead of SImode for the shift count operand.
33340 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33343 2017-04-04 Richard Biener <rguenther@suse.de>
33345 PR middle-end/80281
33346 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33347 arithmetic done for the negate or the plus. Simplify.
33348 (A - (-B) -> A + B): Likewise.
33349 * fold-const.c (split_tree): Make sure to not negate pointers.
33351 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33353 PR rtl-optimization/60818
33354 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33355 a compare of comparisons with the thing compared if this results
33356 in a different machine mode.
33358 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33360 * alias.c (base_alias_check): Fix typo in comment.
33361 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33362 * cgraphunit.c (symbol_table::compile): Likewise.
33363 * collect2.c (maybe_run_lto_and_relink): Likewise.
33364 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33365 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33366 * config/avr/avr.c (avr_map_op_t): Likewise.
33367 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33368 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33369 * config/epiphany/epiphany.md (movcc): Likewise.
33370 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33371 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33373 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33374 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33375 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33376 * config/sh/sh.c (sh_rtx_costs): Likewise.
33377 * fold-const.c (fold_truth_andor): Likewise.
33378 * genautomata.c (collapse_flag): Likewise.
33379 * gengtype.h (struct type::u::s): Likewise.
33380 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33381 * input.c (FORMAT_AMOUNT): Likewise.
33382 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33383 (known_aggs_to_agg_replacement_list): Likewise.
33384 * ipa-inline-analysis.c: Likewise.
33385 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33386 * ipa-polymorphic-call.c
33387 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33388 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33389 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33391 * modulo-sched.c (apply_reg_moves): Likewise.
33392 * omp-expand.c (build_omp_regions_1): Likewise.
33393 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33394 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33395 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33396 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33397 * value-prof.c: Likewise.
33398 * var-tracking.c (val_reset): Likewise.
33400 2017-04-03 Richard Biener <rguenther@suse.de>
33402 PR tree-optimization/80275
33403 * fold-const.c (split_address_to_core_and_offset): Handle
33406 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33408 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33409 descriptors is at least equal to that of functions.
33411 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33413 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33415 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33418 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33419 (mov<IMOD4:mode>): New expander.
33420 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33422 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33424 PR rtl-optimization/79405
33425 * fwprop.c (propagations_left): New variable.
33426 (forward_propagate_into): Decrement it.
33427 (fwprop_init): Initialize it.
33428 (fw_prop): If the variable has reached zero, stop propagating.
33429 (fwprop_addr): Ditto.
33431 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33434 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33435 a FUNCTION_DECL, pass it as decl instead of origin to
33438 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33440 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33443 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33446 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33447 TARGET_VSX_SMALL_INTEGER.
33449 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33451 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33452 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33454 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33456 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33457 extraction from odd-numbered MSA register.
33459 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33461 PR middle-end/80173
33462 * expmed.c (store_bit_field_1): Don't attempt to create
33463 a word subreg out of hard registers wider than word if they
33464 have HARD_REGNO_NREGS of 1 for their mode.
33466 PR middle-end/80163
33467 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33468 conversions to integer types wider than word and pointer.
33471 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33472 (rtx_equal_for_cselib_p): Pass 0 to it.
33473 * cselib.c (cselib_hasher::equal): Likewise.
33474 (rtx_equal_for_cselib_1): Add depth argument. If depth
33475 is 128, don't look up VALUE locs and punt. Increment
33476 depth in recursive calls when walking VALUE locs.
33478 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33480 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33481 (make_gcov_file_name): Use the canonical path name for generating
33483 (read_line): Fix handling of files with ascii null bytes.
33485 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33487 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33488 to initialise a vector register instead
33489 of using a const_int.
33491 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33493 PR translation/80189
33494 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33495 diagnostic messages.
33497 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33500 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33501 (dfp_diex_<mode>): Update mode of operand 1.
33502 * doc/extend.texi (dxex, dxexq): Document change to return type.
33503 (diex, diexq): Document change to argument type.
33505 2017-03-30 Martin Jambor <mjambor@suse.cz>
33508 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33509 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33510 it reflects the signature changes performed at the callee side.
33511 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33512 to cgraph_build_function_type_skip_args.
33513 (build_function_decl_skip_args): Adjust call to the above function.
33515 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33518 * config/i386/sse.md
33519 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33520 register as dest whenever it is a MEM not rtx_equal_p to the
33521 corresponding dup operand, and when forcing into reg move the
33522 reg into the memory afterwards.
33523 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33524 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33525 for the force_reg mode.
33526 (avx512vl_vextractf128<mode>): Use register as dest either
33527 always when a MEM, or when it is a MEM not rtx_equal_p to the
33528 corresponding dup operand, or even not when it is a CONST_VECTOR
33529 depending on the mode and lo vs. hi.
33530 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33532 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33533 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33534 Likewise. Require that operands[2] is even.
33535 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33536 Remove extraneous parens. Require that operands[2] is a multiple
33538 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33539 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33540 disallow memory then.
33542 2017-03-30 Richard Biener <rguenther@suse.de>
33544 PR tree-optimization/77498
33545 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33546 to non-constants over backedges.
33548 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33550 PR rtl-optimization/80233
33551 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33552 as last_combined_insn. Do not test for BARRIER_P separately.
33554 2017-03-29 Andreas Schwab <schwab@suse.de>
33557 * calls.c (prepare_call_address): Convert funexp to Pmode before
33558 copying to temp reg.
33560 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33562 PR tree-optimization/80158
33563 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33564 Handle possible future case of more than one alternate
33566 (replace_rhs_if_not_dup): Likewise.
33567 (replace_one_candidate): Likewise.
33569 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33571 PR rtl-optimization/80193
33572 * ira.c (ira): Do not check allocation for LRA.
33574 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33576 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33577 (nvptx_output_simt_exit): Declare.
33578 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33579 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33580 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33581 (nvptx_declare_function_name): ...here. Emit declaration of local
33582 memory space buffer for omp_simt_enter insn.
33583 (nvptx_output_unisimt_switch): New.
33584 (nvptx_output_softstack_switch): New.
33585 (nvptx_output_simt_enter): New.
33586 (nvptx_output_simt_exit): New.
33587 * config/nvptx/nvptx.h (struct machine_function): New fields
33588 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33589 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33590 (UNSPECV_SIMT_EXIT): Ditto.
33591 (omp_simt_enter_insn): New insn.
33592 (omp_simt_enter): New expansion.
33593 (omp_simt_exit): New insn.
33594 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33596 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33597 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33598 (expand_GOMP_SIMT_EXIT): New.
33599 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33600 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33601 (GOMP_SIMT_EXIT): Ditto.
33602 * target-insns.def (omp_simt_enter): New insn.
33603 (omp_simt_exit): Ditto.
33604 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33606 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33607 (lower_rec_input_clauses): Likewise.
33608 (lower_lastprivate_clauses): Handle SIMT privatization.
33610 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33611 (ompdevlow_adjust_simt_enter): New.
33612 (find_simtpriv_var_op): New.
33613 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33614 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33616 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33617 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33618 (copy_decl_for_dup_finish): Ditto.
33620 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33622 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33625 * config/i386/i386.c (ix86_option_override_internal): Always
33626 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33628 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33630 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33632 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33634 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33635 mark new edge's irreducible flag accordign to it.
33636 (vect_do_peeling): Check loop preheader edge's irreducible flag
33637 and pass it to function slpeel_add_loop_guard.
33639 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33641 PR tree-optimization/80218
33642 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33643 Update block frequencies and counts.
33645 2017-03-28 Richard Biener <rguenther@suse.de>
33647 PR tree-optimization/78644
33648 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33649 of a simplification result we may not use it at all.
33651 2017-03-28 Richard Biener <rguenther@suse.de>
33654 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33655 without arguments, generate default definition of a SSA name.
33657 2017-03-28 Richard Biener <rguenther@suse.de>
33659 PR middle-end/80222
33660 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33661 TYPE_REF_CAN_ALIAS_ALL references.
33662 * fold-const.c (fold_indirect_ref_1): Likewise.
33664 2017-03-28 Martin Liska <mliska@suse.cz>
33667 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33668 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33670 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33671 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33673 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33674 (EXTRA_SPECS): Define.
33675 (SUBTARGET_EXTRA_SPECS): Likewise.
33676 (SUBTARGET_CPP_SPEC): Likewise.
33677 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33678 SUBTARGET_EXTRA_SPECS.
33679 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33681 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33683 * config/arc/simdext.md (vst64_insn): Update pattern.
33684 (vld32wh_insn): Likewise.
33685 (vld32wl_insn): Likewise.
33686 (vld64_insn): Likewise.
33687 (vld32_insn): Likewise.
33689 2017-03-28 Marek Polacek <polacek@redhat.com>
33692 * fold-const.c (fold_comparison): Use protected_set_expr_location
33693 instead of SET_EXPR_LOCATION.
33695 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33697 * tree.c (add_expr): Avoid name lookup warning.
33699 2017-03-27 Jeff Law <law@redhat.com>
33701 PR tree-optimization/80216
33702 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33703 function name. Limit recursion depth.
33704 (record_temporary_equivalences): Corresponding changes.
33706 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33708 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33711 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33714 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33716 * cfgcleanup.c (reg_note_cfa_p): New array.
33717 (insns_have_identical_cfa_notes): New function.
33718 (old_insns_match_p): Don't cross-jump in between /f
33719 and non-/f instructions. If both i1 and i2 are frame related,
33720 verify all CFA notes, their order and content.
33722 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33725 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33726 HImode and SImode with zero extend to DImode to one insn.
33727 (bswap<mode>2_extenddi): Likewise.
33728 (bswapsi2_extenddi): Likewise.
33729 (bswaphi2_extendsi): Likewise.
33730 (bswaphi2): Combine bswap HImode and SImode into one insn.
33731 Separate memory insns from swapping register.
33732 (bswapsi2): Likewise.
33733 (bswap<mode>2): Likewise.
33734 (bswaphi2_internal): Delete, no longer used.
33735 (bswapsi2_internal): Likewise.
33736 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33737 store, and gpr<-gpr swap insns.
33738 (bswap<mode>2_store): Likewise.
33739 (bswaphi2_reg): Register only splitter, combine with the splitter.
33740 (bswaphi2 splitter): Likewise.
33741 (bswapsi2_reg): Likewise.
33742 (bswapsi2 splitter): Likewise.
33743 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33744 the insns into load, store, and register/register insns.
33745 (bswapdi2_ldbrx): Likewise.
33746 (bswapdi2_load): Likewise.
33747 (bswapdi2_store): Likewise.
33748 (bswapdi2_reg): Likewise.
33750 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33752 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33753 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33755 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33758 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33761 special handling for target option conflicts between dform
33762 options (-mpower9-dform, -mpower9-dform-vector,
33763 -mpower9-dform-scalar) and -mno-direct-move.
33765 2017-03-27 Richard Biener <rguenther@suse.de>
33767 PR tree-optimization/80181
33768 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33770 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33772 * config/arc/predicates.md (move_double_src_operand): Replace the
33773 call to move_double_src_operand with a call to address_operand.
33775 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33777 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33778 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33779 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33781 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33783 * config/arc/predicates.md (long_immediate_loadstore_operand):
33784 Consider scaled addresses cases.
33786 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33788 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33789 restored when in interrupt.
33790 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33791 doesn't have delay slot.
33793 2017-03-27 Richard Biener <rguenther@suse.de>
33796 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33797 inlined thunk clones.
33799 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33802 * asan.c (instrument_derefs): Copy over last operand from
33803 original COMPONENT_REF to the new COMPONENT_REF with
33804 DECL_BIT_FIELD_REPRESENTATIVE.
33805 * ubsan.c (instrument_object_size): Likewise.
33807 2017-03-27 Richard Biener <rguenther@suse.de>
33809 PR tree-optimization/80170
33810 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33811 sure DR/SCEV didnt fold in constants we do not see when looking
33812 at the reference base alignment.
33814 2017-03-27 Richard Biener <rguenther@suse.de>
33816 PR middle-end/80171
33817 * gimple-fold.c (fold_ctor_reference): Properly guard against
33818 NULL return value from canonicalize_constructor_val.
33820 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33823 * config/i386/i386.c (ix86_expand_builtin)
33824 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33825 flags reg setting and flags reg using instructions.
33826 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33827 clobbering instructions to zero extend op2.
33829 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33831 * doc/install.texi (Configuration) <--with-aix-soname>:
33832 Update link to AIX ld.
33834 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33836 PR rtl-optimization/80160
33837 PR rtl-optimization/80159
33838 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33839 reg_alternate_class into account.
33841 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33844 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33845 to consider in curr_insn_transform.
33847 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33849 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33850 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33851 and emit_mode_inner.
33853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33855 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33856 argument to the overloaded builtin variants. Use the new flag to
33857 deprecate certain builtin variants.
33858 * config/s390/s390-builtin-types.def: Add new builtin types.
33859 * config/s390/s390-builtins.h: Support new flags field for
33860 overloaded builtins.
33861 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33862 (s390_macro_to_expand): Enable vector float data type.
33863 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33864 builtins by incrementing the __VEC__ version number.
33865 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33867 (s390_resolve_overloaded_builtin): Emit error messages depending
33868 on the builtin flags.
33869 * config/s390/s390.c (s390_expand_builtin): Support additional
33870 flags argument. Change error message to match the messages
33871 emitted in s390-c.c.
33872 * config/s390/s390.md: New UNSPEC_* constants.
33873 (op_type): Add new instruction types.
33874 * config/s390/vecintrin.h: Add new builtins and test data class
33876 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33877 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33878 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33879 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33880 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33881 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33883 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33884 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33885 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33886 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33888 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33889 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33890 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33891 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33893 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33894 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33895 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33896 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33897 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33898 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33899 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33901 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33902 ("vec_scatter_element<V_HW_4:mode>_DI")
33903 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33904 ("vec_fpint<mode>", "vflls")
33905 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33906 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33907 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33908 ("*vec_cmphe<mode>_cc"): ... these.
33910 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33911 mode constant instead of magic value.
33913 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33915 * config/s390/s390.c (s390_expand_vec_compare): Support other
33916 vector floating point modes than just V2DF.
33917 (s390_expand_vcond): Likewise.
33918 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33919 (s390_cannot_change_mode_class): Prevent mode changes between TF
33920 and V1TF in vector registers.
33921 * config/s390/s390.md (DF, SF): New mode attributes.
33922 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33923 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33924 SFmode support for VRs.
33925 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33927 (VFT, VF_HW): New mode iterators.
33928 (vw, sdx): New mode attributes.
33929 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33930 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33931 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33932 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33933 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33934 also the new vector floating point modes. Renaming to ...
33936 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33937 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33938 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33939 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33940 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33941 ("vec_unordered<mode>"): ... these.
33943 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33944 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33945 ("*vec_extendv2df"): New insn definitions.
33947 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33949 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33950 ("mulditi3_2", "*muldi3_sign"): New patterns.
33951 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33952 rename the pattern definition.
33954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33956 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33958 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33960 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33962 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33963 instruction if possible.
33964 * config/s390/vector.md (vec_halfnumelts): New mode
33966 ("*vec_vllezlf<mode>"): New pattern.
33968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33970 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33971 ("popcountv4si2", "popcountv2di2"): Rename to ...
33972 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33973 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33975 ("popcount<mode>2_vxe"): New pattern.
33977 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33979 * common/config/s390/s390-common.c (processor_flags_table): Add
33981 * config.gcc: Add arch12.
33982 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33983 Default to arch12 for unknown CPU model numbers.
33984 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33985 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33986 PROCESSOR_max sanity check.
33987 * config/s390/s390-opts.h (enum processor_type): Add
33989 * config/s390/s390.c (processor_table): Add arch12.
33990 (s390_expand_builtin): Add check for B_VXE flag.
33991 (s390_issue_rate): Add PROCESSOR_ARCH12.
33992 (s390_get_sched_attrmask): Likewise.
33993 (s390_get_unit_mask): Likewise.
33994 (s390_sched_score): Enable z13 scheduling for arch12.
33995 (s390_sched_reorder): Likewise.
33996 (s390_sched_variable_issue): Likewise.
33997 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33999 (s390_tune_attr): Use z13 scheduling also for arch12.
34000 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34001 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34002 (TARGET_VXE_P): New macros.
34003 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34004 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34005 * config/s390/s390.opt: Add arch12 as processor_type.
34007 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34009 * config/s390/s390.md
34010 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34011 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34012 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34014 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34015 Rename expanders to ...
34017 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34018 ("fixuns_truncdddi2_emu"): ... these.
34020 ("fixuns_trunc<mode>si2_emu"): New expander.
34022 ("*fixuns_truncdfdi2_z13"): Rename to ...
34023 ("*fixuns_truncdfdi2_vx"): ... this.
34025 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34027 * config/s390/2964.md: Remove the single element vector compare
34028 instructions which are no longer used.
34029 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34031 (s390_canonicalize_comparison): Remove handling of DFmode
34033 (s390_expand_vec_compare_scalar): Remove function.
34034 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34035 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34037 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34039 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34041 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34043 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34044 will anyway by matched by mov<mode>_64dfp.
34046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34048 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34049 vlef/vstef. Add missing operand to vleif.
34051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34053 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34054 pair for all vector types with 64 bit elements.
34055 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34056 * config/s390/vector.md (V_HW_64): ... here.
34057 (V_128_NOSINGLE): New mode iterator.
34058 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34059 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34060 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34061 ("*vec_load_pairv2di"): Change to ...
34062 ("*vec_load_pair<mode>"): ... this one.
34064 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34066 * config/s390/constraints.md: Add comments.
34067 (jKK): Reject element sizes > 8 bytes.
34068 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34070 * config/s390/s390.md: Add the s_operand checks formerly in
34071 s390_split_ok_p to various splitters where they are still
34073 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34074 for 128 bit vectors. Plus two splitters.
34076 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34078 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34081 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34084 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34085 error if the boundary argument is not constant.
34087 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34089 PR rtl-optimization/80112
34090 * loop-doloop.c (doloop_condition_get): Don't check condition
34091 if cmp isn't SET with IF_THEN_ELSE src.
34093 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34095 PR tree-optimization/80158
34096 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34097 replacing a candidate statement, also replace it for the
34098 candidate's alternate interpretation.
34099 (replace_rhs_if_not_dup): Likewise.
34100 (replace_one_candidate): Likewise.
34102 2017-03-24 Richard Biener <rguenther@suse.de>
34104 PR tree-optimization/80167
34105 * graphite-isl-ast-to-gimple.c
34106 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34108 (translate_isl_ast_to_gimple::get_rename): Likewise.
34110 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34112 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34113 handling of certain combinations of target options, including the
34114 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34115 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34117 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34120 * config/arm/arm.md (*load_multiple): Add reload_completed to
34121 matching condition.
34123 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34124 Richard Biener <rguenth@suse.de>
34126 PR tree-optimization/79908
34127 PR tree-optimization/80136
34128 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34129 been cast away, gimplify_and_add suffices.
34131 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34133 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34135 2017-03-23 Richard Biener <rguenther@suse.de>
34137 PR tree-optimization/80032
34138 * gimplify.c (gimple_push_cleanup): Forced unconditional
34139 cleanups still have to go to the conditional_cleanups
34142 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34144 PR tree-optimization/80072
34145 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34147 (next_operand_entry_id): Change type to unsigned int.
34148 (sort_by_operand_rank): Make sure to return the right return value
34149 even if unsigned fields are bigger than INT_MAX.
34150 (struct oecount): Change cnt and id type to unsigned int.
34151 (oecount_hasher::equal): Formatting fix.
34152 (oecount_cmp): Make sure to return the right return value
34153 even if unsigned fields are bigger than INT_MAX.
34154 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34157 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34158 TREE_READONLY on result if writing it more than once.
34161 * doc/invoke.texi (-fsanitize=thread): Document that with
34162 -fnon-call-exceptions atomics are not able to throw
34166 * tsan.c: Include tree-eh.h.
34167 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34168 maybe_clean_or_replace_eh_stmt where needed.
34169 (instrument_memory_accesses): Add cfg_changed argument.
34170 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34171 if it returned true.
34172 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34174 PR rtl-optimization/63191
34175 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34176 wrapper function, moved the whole old content into ...
34177 (ix86_delegitimize_address_1): ... this. New inline function.
34178 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34179 true as last argument instead of ix86_delegitimize_address.
34181 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34183 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34184 cortexa57_branch_cost.
34186 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34188 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34190 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34193 * doc/md.texi (Constraints): Document wA constraint.
34194 * config/rs6000/constraints.md (wA): New.
34195 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34196 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34197 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34198 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34200 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34203 * gimplify.c (is_oacc_declared): New function.
34204 (oacc_default_clause): Use it to set default flags for acc declared
34205 variables inside parallel regions.
34206 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34207 declared variables.
34208 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34209 declare attribute to any decl as necessary.
34211 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34214 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34215 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34216 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34217 (arm_arch_lpae): This.
34218 * config/arm/arm.c (arm_arch7ve): Rename into ...
34219 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34220 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34223 2017-03-22 Martin Liska <mliska@suse.cz>
34226 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34227 error message instead of an ICE.
34229 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34231 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34233 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34235 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34237 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34240 2017-03-21 Martin Sebor <msebor@redhat.com>
34242 * doc/extend.texi: Use "cannot" instead of "can't."
34243 * doc/hostconfig.texi: Same.
34244 * doc/install.texi: Same.
34245 * doc/invoke.texi: Same.
34246 * doc/loop.texi: Same.
34247 * doc/md.texi: Same.
34248 * doc/objc.texi: Same.
34249 * doc/rtl.texi: Same.
34250 * doc/tm.texi: Same.
34251 * doc/tm.texi.in: Same.
34252 * doc/trouble.texi: Same.
34254 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34257 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34258 (collect_checksum_attributes): Set it.
34259 (die_checksum_ordered): Use it.
34261 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34263 PR tree-optimization/79908
34264 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34265 change: For a VA_ARG whose LHS has been cast away, use
34266 force_gimple_operand to construct the side effects.
34268 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34270 PR translation/80001
34271 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34272 more amenable to translation.
34273 (oacc_loop_auto_partitions): Likewise.
34275 2017-03-21 Marek Polacek <polacek@redhat.com>
34276 Martin Sebor <msebor@redhat.com>
34278 PR tree-optimization/80109
34279 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34280 on INTEGRAL_TYPE_P.
34282 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34283 Segher Boessenkool <segher@kernel.crashing.org>
34286 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34287 check reg_used_between_p between insn and one of succ or succ2
34288 depending on if succ is artificial insn not inserted into insn
34291 2017-03-21 Martin Liska <mliska@suse.cz>
34293 PR gcov-profile/80081
34294 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34295 * doc/gcc.texi: Include gcov-dump stuff.
34296 * doc/gcov-dump.texi: New file.
34298 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34300 PR rtl-optimization/79150
34301 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34302 conditional jump, if the jump is the last insn of the loop.
34304 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34305 Richard Biener <rguenth@suse.de>
34307 PR tree-optimization/79908
34308 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34309 been cast away, use force_gimple_operand to construct the side
34312 2017-03-21 Martin Liska <mliska@suse.cz>
34314 PR libfortran/79956
34315 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34318 2017-03-21 Brad Spengler <spender@grsecurity.net>
34321 * plugin.c (htab_hash_plugin): New function.
34322 (add_new_plugin): Use it and adjust.
34323 (parse_plugin_arg_opt): Adjust.
34324 (init_one_plugin): Likewise.
34326 2017-03-21 Richard Biener <rguenther@suse.de>
34328 PR tree-optimization/80032
34329 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34330 if set force the cleanup to happen unconditionally.
34331 (gimplify_target_expr): Push inserted clobbers with force_uncond
34332 to avoid them being removed by control-dependent DCE.
34334 2017-03-21 Richard Biener <rguenther@suse.de>
34336 PR tree-optimization/80122
34337 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34338 va_arg_pack_len when the inlined call stmt requires pack
34340 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34342 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34345 * tsan.c (instrument_builtin_call): If the memory model argument
34346 is not a constant, assume it is valid.
34349 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34352 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34354 PR rtl-optimization/79910
34355 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34356 if its dest is used by an insn before I2 (other than the combined
34357 insns themselves, which are properly handled already).
34359 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34362 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34364 * combine.c (record_used_regs): New static function.
34365 (try_combine): Handle situations where there is an additional
34366 instruction between I2 and I3 which needs to have a LOG_LINK
34370 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34372 * combine.c (try_combine): Delete redundant i1 test. Call
34373 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34375 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34378 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34379 alternatives 13/14.
34381 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34383 PR tree-optimization/80054
34384 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34385 the optimization if a PHI or any of its arguments is not dominated
34386 by the candidate's basis. Use gphi* rather than gimple* as
34388 (replace_profitable_candidates): Clean up a gimple* variable that
34389 should be a gphi* variable.
34391 2017-03-20 Martin Sebor <msebor@redhat.com>
34394 * doc/extend.texi (attribute constructor): Document present limitation.
34396 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34399 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34400 __POWER9_VECTOR__ #ifdef control, change template definition to
34401 use Power9-specific built-in function.
34402 (vec_any_eq): Likewise.
34403 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34404 to control outcomes from this test.
34405 (vector_ae_<mode>p): For VEC_F modes, likewise.
34407 2017-03-20 Ian Lance Taylor <iant@google.com>
34409 * config/i386/i386.c (ix86_function_regparm): Save an extra
34410 register for -fsplit-stack with DECL_STATIC_CHAIN.
34412 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34415 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34416 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34418 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34420 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34422 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34425 2017-03-20 Marek Polacek <polacek@redhat.com>
34428 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34430 2017-03-20 Richard Biener <rguenther@suse.de>
34432 PR tree-optimization/80113
34433 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34434 allocate extra SSA name for PHI def.
34435 (add_close_phis_to_outer_loops): Likewise.
34436 (add_close_phis_to_merge_points): Likewise.
34437 (copy_loop_close_phi_args): Likewise.
34438 (copy_cond_phi_nodes): Likewise.
34440 2017-03-20 Martin Liska <mliska@suse.cz>
34442 PR middle-end/79753
34443 * tree-chkp.c (chkp_build_returned_bound): Do not build
34444 returned bounds for a LHS that's not a BOUNDED_P type.
34446 2017-03-20 Martin Liska <mliska@suse.cz>
34450 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34451 COMPLEX_CST and VECTOR_CST.
34453 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34456 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34457 target operand. A new splitter adds the clobber statement in case
34458 the target operand is dead anyway.
34460 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34462 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34463 to age-old versions of binutils and glibc.
34465 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34467 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34469 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34471 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34473 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34475 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34476 requirement for binutils 2.13.
34478 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34480 * combine.c (try_combine): Delete redundant i1 test. Call
34481 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34483 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34485 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34486 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34488 <riscv64-*-elf>: Re-arrange section
34489 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34490 <riscv32-*-linux>: Likewise.
34491 <riscv64-*-elf>: Likewise
34492 <riscv64-*-linux>: Likewise.
34494 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34497 * aarch64.opt(verbose-cost-dump): Fix typo.
34499 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34502 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34503 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34505 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34507 * reload.c (find_reloads): When reloading a nonoffsettable address,
34508 use RELOAD_OTHER for it and its address reloads.
34510 PR rtl-optimization/79910
34511 * combine.c (record_used_regs): New static function.
34512 (try_combine): Handle situations where there is an additional
34513 instruction between I2 and I3 which needs to have a LOG_LINK
34516 2017-03-17 Jeff Law <law@redhat.com>
34518 PR tree-optimization/71437
34519 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34520 conditional in the hash table first.
34521 (vrp_dom_walker::before_dom_children): Extract condition from
34522 ASSERT_EXPR. Record condition, its inverion and any implied
34523 conditions as well.
34525 2017-03-17 Marek Polacek <polacek@redhat.com>
34526 Markus Trippelsdorf <markus@trippelsdorf.de>
34528 PR tree-optimization/80079
34529 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34532 2017-03-17 Richard Biener <rguenther@suse.de>
34534 PR middle-end/80075
34535 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34536 Properly verify the LHS before the RHS possibly claims to be
34538 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34541 2017-03-17 Martin Jambor <mjambor@suse.cz>
34543 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34544 (List of -O2 options): Likewise.
34545 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34548 2017-03-17 Tom de Vries <tom@codesourcery.com>
34550 * gcov-dump.c (print_usage): Print bug_report_url.
34552 2017-03-17 Richard Biener <rguenther@suse.de>
34554 PR middle-end/80050
34555 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34556 (parser::peek): Likewise.
34558 2017-03-17 Richard Biener <rguenther@suse.de>
34560 PR tree-optimization/80048
34561 * sese.c (free_sese_info): Properly release rename_map and
34562 copied_bb_map elements.
34564 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34566 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34567 Add linked-list forward and backlinks. Insert on
34568 construction, remove on destruction.
34569 (class pass_store_merging): Add m_stores_head field.
34570 (pass_store_merging::terminate_and_process_all_chains):
34571 Iterate over m_stores_head list.
34572 (pass_store_merging::terminate_all_aliasing_chains):
34574 (pass_store_merging::execute): Check for debug stmts first.
34575 Push new chains onto the m_stores_head stack.
34577 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34580 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34581 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34582 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34584 2017-03-16 Jeff Law <law@redhat.com>
34586 PR tree-optimization/71437
34587 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34588 member function. Implementation moved into after_dom_children
34589 member function and into the threader's thread_outgoing_edges
34591 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34592 some code into new thread_outgoing_edges.
34593 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34594 definition. Simplify marker handling (do it here). Assume we always
34595 have the available expression and the const/copies tables.
34596 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34598 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34599 * tree-vrp.c (equiv_stack): No longer file scoped.
34600 (vrp_dom_walker): New class.
34601 (vrp_dom_walker::before_dom_children): New member function.
34602 (vrp_dom_walker::after_dom_children): Likewise.
34603 (identify_jump_threads): Setup domwalker. Use it rather than
34604 walking edges in a random order by hand. Simplify setup/finalization.
34605 (finalize_jump_threads): Remove.
34606 (vrp_finalize): Do not call identify_jump_threads here.
34607 (execute_vrp): Do it here instead and call thread_through_all_blocks
34610 PR tree-optimization/71437
34611 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34613 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34615 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34616 (dom_opt_dom_walker::thread_across_edge): Remove
34617 handle_dominating_asserts argument. All callers changed.
34618 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34619 changes. Remove calls to lhs_of_dominating_assert. Other
34620 uses of handle_dominating_asserts turn into unconditional code
34621 (simplify_control_stmt_condition_1): Likewise.
34622 (simplify_control_stmt_condition): Likewise.
34623 (thread_through_normal_block, thread_across_edge): Likewise.
34624 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34625 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34626 object if it is not an SSA_NAME.
34627 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34628 before calling into the VRP specific simplifiers.
34629 (identify_jump_threads): Remove handle_dominating_asserts
34632 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34635 * tree-diagnostic.c (default_tree_printer): No longer static.
34636 * tree-diagnostic.h (default_tree_printer): New prototype.
34638 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34640 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34641 Change ins into fmov.
34643 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34645 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34646 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34647 Use h_con constraint for operand 1.
34648 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34649 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34651 2017-03-15 Jeff Law <law@redhat.com>
34653 PR tree-optimization/71437
34654 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34655 (record_temporary_equivalences): Use it.
34657 PR tree-optimization/71437
34658 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34659 tree-ssa-scopedtables.
34660 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34661 (record_conditions, record_cond, vuse_eq): Likewise.
34662 (record_edge_info): Adjust to API tweak of record_conditions.
34663 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34664 (record_temporary_equivalences, optimize_stmt): Likewise.
34665 (eliminate_redundant_computations): Likewise.
34666 (record_equivalences_from_stmt): Likewise.
34667 * tree-ssa-scopedtables.c: Include options.h and params.h.
34668 (vuse_eq): New function, moved from tree-ssa-dom.c
34669 (build_and_record_new_cond): Likewise.
34670 (record_conditions): Likewise. Accept vector of conditions rather
34671 than edge_equivalence structure for first argument.
34672 for the first argument.
34673 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34674 from tree-ssa-dom.c.
34675 (avail_exprs_stack::record_cond): Likewise.
34676 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34677 from tree-ssa-dom.c.
34678 (avail_exprs_stack): Add new member functions lookup_avail_expr
34680 (record_conditions): Declare.
34682 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34685 * lra-constraints.c (process_alt_operands): Increase reject for
34686 reloading an input/output operand.
34688 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34691 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34692 insns to convert from signed/unsigned char/short to IEEE 128-bit
34694 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34696 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34699 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34700 subreg of inner mode for values already in registers.
34702 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34704 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34705 iteration reg is used after the loop.
34707 2017-03-14 Martin Sebor <msebor@redhat.com>
34709 PR tree-optimization/79800
34710 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34711 precision in negative-positive range.
34712 (format_floating): Call non-const overload with adjusted precision.
34714 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34717 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34720 2017-03-14 Martin Sebor <msebor@redhat.com>
34722 PR middle-end/80020
34723 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34724 * builtins.def (aligned_alloc): Use it.
34727 * Makefile.in (GTFILES): Add calls.c.
34728 * calls.c: Include "gt-calls.h".
34730 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34732 PR rtl-optimization/79728
34733 * regs.h (struct target_regs): New field
34734 x_contains_allocatable_regs_of_mode.
34735 (contains_allocatable_regs_of_mode): New macro.
34736 * reginfo.c (init_reg_sets_1): Initialize it, and change
34737 contains_reg_of_mode so it includes global regs as well.
34738 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34739 rather than contains_regs_of_mode.
34741 2017-03-14 Martin Liska <mliska@suse.cz>
34743 * doc/invoke.texi: Document options that can't be combined with
34744 -fcheck-pointer-bounds.
34746 2017-03-14 Martin Liska <mliska@suse.cz>
34748 PR middle-end/79831
34749 * doc/invoke.texi (-Wchkp): Document the option.
34751 2017-03-14 Martin Liska <mliska@suse.cz>
34753 * Makefile.in: Install gcov-dump.
34755 2017-03-14 Martin Liska <mliska@suse.cz>
34757 * multiple_target.c (expand_target_clones): Bail out for
34758 an invalid attribute.
34760 2017-03-14 Richard Biener <rguenther@suse.de>
34762 * alias.c (struct alias_set_entry): Pack properly.
34763 * cfgloop.h (struct loop): Likewise.
34764 * cse.c (struct set): Likewise.
34765 * ipa-utils.c (struct searchc_env): Likewise.
34766 * loop-invariant.c (struct invariant): Likewise.
34767 * lra-remat.c (struct cand): Likewise.
34768 * recog.c (struct change_t): Likewise.
34769 * rtl.h (struct address_info): Likewise.
34770 * symbol-summary.h (function_summary): Likewise.
34771 * tree-loop-distribution.c (struct partition): Likewise.
34772 * tree-object-size.c (struct object_size_info): Likewise.
34773 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34774 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34775 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34776 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34777 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34778 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34779 (struct _stmt_vec_info): Likewise.
34781 2017-03-14 Martin Liska <mliska@suse.cz>
34784 * multiple_target.c (create_dispatcher_calls): Check that
34785 a target can create a function dispatcher.
34787 2017-03-14 Martin Liska <mliska@suse.cz>
34790 * multiple_target.c (expand_target_clones): Drop local.local
34791 flag for default implementation.
34793 2017-03-14 Richard Biener <rguenther@suse.de>
34795 PR tree-optimization/80030
34796 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34798 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34800 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34801 gcc_fallthrough() instead of __attribute__((fallthrough));
34803 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34805 * doc/gcc.texi: Remove "up" link to (DIR).
34806 * doc/gccint.texi: Ditto.
34808 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34810 * doc/install.texi (Specific) <avr>: Remove reference to
34813 2017-03-13 Jeff Law <law@redhat.com>
34815 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34816 attribute rather than comments.
34818 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34819 match_scratch operand is highest.
34821 2017-03-13 Martin Liska <mliska@suse.cz>
34823 PR middle-end/78339
34824 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34825 is a CHKP clone, use original declaration.
34827 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34829 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34830 (arc_conditional_register_usage): Use a different allocation order
34831 when optimizing for size.
34832 * common/config/arc/arc-common.c (arc_option_optimization_table):
34833 Section anchors default on when optimizing for size.
34835 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34837 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34839 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34841 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34842 * config/arc/arc.md (cpu_facility): Add cd variant.
34843 (*movqi_insn): Add code density variant.
34844 (*movhi_insn): Likewise.
34845 (*movqi_insn): Likewise.
34846 (*addsi3_mixed): Likewise.
34847 (subsi3_insn): Likewise.
34849 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34851 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34853 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34855 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34856 expressions with MINUS and UNARY ops.
34858 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34861 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34863 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34864 between vec_select and vector argument.
34865 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34866 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34867 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34868 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34869 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34870 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34872 2017-03-13 Richard Biener <rguenther@suse.de>
34875 * params.def (vect-max-peeling-for-alignment): Fix typo.
34877 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34879 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34880 issue that only occurred with binutils below 2.18.
34882 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34884 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34885 refer to binutils 2.11/2.12 minimum.
34887 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34889 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34890 ftp.kernel.org and simplify binutils requirement.
34892 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34894 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34896 (Optimize Options): Ditto. Also remove redundancy.
34898 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34900 PR translation/79848
34901 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34903 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34904 to G_ to avoid double translation.
34906 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34908 PR translation/79923
34909 * auto-profile.c (get_combined_location): Convert leading
34910 character of diagnostics to lower case and remove trailing period.
34911 (read_profile): Likewise for various diagnostics.
34912 * config/arm/arm.c (arm_option_override): Remove trailing period
34913 from various diagnostics.
34914 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34915 (msp430_expand_delay_cycles): Likewise.
34917 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34920 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34921 full command-line argument, rather than just "str".
34922 (aarch64_validate_march): Likewise.
34923 (aarch64_validate_mtune): Likewise.
34925 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34927 PR rtl-optimization/78911
34928 * lra-assigns.c (must_not_spill_p): New function.
34929 (spill_for): Use it.
34931 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34933 PR tree-optimization/79981
34934 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34935 ATOMIC_COMPARE_EXCHANGE ifn result.
34936 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34937 IFN_ATOMIC_COMPARE_EXCHANGE.
34939 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34942 * opts.c (parse_sanitizer_options): Add missing question mark to
34943 "did you mean" message.
34945 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34947 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34949 (VMULEUH_UNS): Likewise.
34950 (VMULOUB_UNS): Likewise.
34951 (VMULOUH_UNS): Likewise.
34952 * config/rs6000/rs6000.c (builtin_function_type): Remove
34953 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34955 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34958 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34959 x with result of extra_parsing_for_operand_code_0.
34960 (function_reader::extra_parsing_for_operand_code_0): Convert
34961 return type from void to rtx, returning x. When reading
34962 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34963 larger size containing struct block_symbol.
34965 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34967 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34968 -mfloat128-hardware without -m64.
34970 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34973 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34974 entries to the case statement that marks unsigned arguments to
34975 overloaded functions.
34977 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34979 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34980 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34982 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34985 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34986 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34988 2017-03-10 Martin Liska <mliska@suse.cz>
34992 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34993 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34996 2017-03-10 Olivier Hainque <hainque@adacore.com>
34998 * tree-switch-conversion (array_value_type): Start by resetting
34999 candidate type to it's main variant.
35001 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35003 PR rtl-optimization/79909
35004 * combine.c (try_combine): Use simplify_replace_rtx on individual
35005 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35006 of the whole CALL_INSN_FUNCTION_USAGE.
35008 PR tree-optimization/79972
35009 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35010 get_range_info on SSA_NAMEs. Formatting fixes.
35012 2017-03-10 Richard Biener <rguenther@suse.de>
35013 Jakub Jelinek <jakub@redhat.com>
35015 PR tree-optimization/77975
35016 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35017 edge to be constant.
35018 (get_val_for): For constant x return it. Formatting fix.
35019 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35020 would use the same bases as the current one.
35022 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35024 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35025 instead of vec_select for V1TImode.
35026 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35028 (VSX_LE_128): Add V1TI to this mode iterator.
35029 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35030 (*vsx_le_perm_store_<mode>): Likewise.
35031 (pre-reload splitter for VSX stores): Likewise.
35032 (post-reload splitter for VSX stores): Likewise.
35033 (*vsx_xxpermdi2_le_<mode>): Likewise.
35034 (*vsx_lxvd2x2_le_<mode>): Likewise.
35035 (*vsx_stxvd2x2_le_<mode>): Likewise.
35037 2017-03-09 Michael Eager <eager@eagercon.com>
35039 Correct failures with --enable-checking=yes,rtl.
35041 * config/microblaze/microblaze.c (microblaze_expand_shift):
35042 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35043 test for const0_rtx.
35044 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35045 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35047 2017-03-09 Richard Biener <rguenther@suse.de>
35049 PR tree-optimization/79977
35050 * graphite-scop-detection.c (scop_detection::merge_sese):
35051 Handle the case of extra exits to blocks dominating the entry.
35053 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35055 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35058 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35060 PR rtl-optimization/79949
35061 * lra-constraints.c (process_alt_operands): Check memory when
35062 trying to predict a cycle. Print about the overall increase.
35064 2017-03-09 Richard Biener <rguenther@suse.de>
35066 PR middle-end/79971
35067 * gimple-expr.c (useless_type_conversion_p): Preserve
35068 TYPE_SATURATING for fixed-point types.
35070 2017-03-09 Richard Biener <rguenther@suse.de>
35073 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35074 alignment of BLKmode params.
35076 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35079 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35080 (VALL_NO_V2Q): Likewise.
35082 * config/aarch64/aarch64-simd.md
35083 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35085 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35086 VALL_NO_V2Q mode iterator.
35087 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35089 2017-03-09 Martin Liska <mliska@suse.cz>
35091 PR tree-optimization/79631
35092 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35093 tree_int_cst_sign_bit just for INTEGER constants.
35095 2017-03-09 Martin Liska <mliska@suse.cz>
35099 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35102 2017-03-09 Marek Polacek <polacek@redhat.com>
35105 * tree.c (inchash::add_expr): Handle TREE_VEC.
35107 2017-03-09 Martin Liska <mliska@suse.cz>
35110 (chkp_narrow_size_and_offset): New function.
35111 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35112 (void chkp_parse_bit_field_ref): New function.
35113 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35114 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35116 2017-03-09 Martin Liska <mliska@suse.cz>
35119 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35120 (chkp_find_bounds_1): Remove gcc_unreachable.
35122 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35125 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35126 BUILT_IN_SYNC*, determine the access type from the size suffix and
35127 always build a MEM_REF with that type. Handle forgotten
35128 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35131 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35132 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35133 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35134 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35135 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35136 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35137 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35138 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35139 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35140 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35141 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35142 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35143 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35144 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35145 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35146 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35147 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35148 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35149 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35150 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35151 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35152 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35153 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35154 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35155 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35156 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35157 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35158 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35159 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35160 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35161 definitions outside of __OPTIMIZE__ guarded section.
35164 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35165 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35166 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35167 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35170 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35172 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35173 ("vfenez<mode>"): Add missing constraints.
35175 2017-03-08 Martin Sebor <msebor@redhat.com>
35178 * config/nds32/nds32.c (nds32_option_override):
35179 Fix misspelled diagnostic.
35181 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35184 * gimplify.c (gimplify_omp_for): Replace index var in outer
35185 taskloop statement with an artificial variable and add
35186 OMP_CLAUSE_PRIVATE clause for it.
35188 2017-03-08 Richard Biener <rguenther@suse.de>
35190 PR tree-optimization/79955
35191 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35192 for accesses that are completely outside of the variable.
35194 2017-03-08 Andrew Haley <aph@redhat.com>
35196 PR tree-optimization/79943
35197 * tree-ssa-loop-split.c (compute_new_first_bound): When
35198 calculating the new upper bound, (END-BEG) should be added, not
35201 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35203 * config/avr/avr.md (setmemhi): Make sure match_dup
35204 operand number comes before match_scratch.
35206 2017-03-08 Richard Biener <rguenther@suse.de>
35208 PR tree-optimization/79920
35209 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35210 with ncopies == 1 to ...
35211 (vect_transform_slp_perm_load): ... here. Properly compute
35212 all element loads by iterating VF times over the group. Do
35213 not handle ncopies (computed in a broken way) in
35214 vect_create_mask_and_perm.
35216 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35219 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35220 is a uniform vector, use uniform_vector_p return value instead of
35221 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35223 2017-03-07 Marek Polacek <polacek@redhat.com>
35225 PR middle-end/79809
35226 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35227 (alloca_call_type): Likewise.
35229 2017-03-07 Martin Liska <mliska@suse.cz>
35231 * gcov.c (process_args): Put comment to correct location.
35233 2017-03-07 Martin Liska <mliska@suse.cz>
35235 PR middle-end/68270
35236 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35237 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35238 (chkp_narrow_bounds_for_field): Likewise.
35239 (chkp_parse_array_and_component_ref): Pass one more argument to
35242 2017-03-07 Richard Biener <rguenther@suse.de>
35244 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35247 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35249 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35250 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35252 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35255 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35256 to end of description.
35257 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35259 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35261 PR rtl-optimization/79901
35262 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35264 (*avx512f_<code><mode>3<mask_name>): ... this.
35265 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35266 iterator instead of VI8_AVX2_AVX512BW.
35268 PR rtl-optimization/79901
35269 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35270 min/max expander, expand it using expand_vec_cond_expr.
35273 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35276 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35279 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35280 to void * for PCH reasons.
35281 * dwarf2out.c (output_loc_operands, output_die): Cast
35282 v.val_vec.array to unsigned char *.
35284 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35287 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35290 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35292 PR rtl-optimization/79571
35293 * lra-constraints.c (process_alt_operands): Calculate static
35294 reject and subtract it from overall when only addresses will be
35297 2017-03-06 Julia Koval <julia.koval@intel.com>
35300 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35301 incoming stack boundary to 128 for 64-bit targets.
35303 2017-03-06 Richard Biener <rguenther@suse.de>
35305 PR tree-optimization/79894
35306 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35307 to NULL after folding it.
35309 2017-03-06 Richard Biener <rguenther@suse.de>
35311 PR tree-optimization/79824
35312 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35313 check disabling peeling for gaps.
35315 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35317 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35318 attributes): Document gettimeofday.
35320 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35322 * config/s390/s390.c (s390_option_override_internal): Set
35323 PARAM_MIN_VECT_LOOP_BOUND
35325 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35327 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35328 * config/s390/s390.md: Likewise.
35330 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35333 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35334 (<avx2_avx512>_perm<mode>): Rename to ...
35335 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35337 (<avx512>_perm<mode>_mask): Rename to ...
35338 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35340 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35341 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35342 instead of VI8F_256_512.
35343 (avx512f_perm<mode>): New define_expand.
35344 (avx512f_perm<mode>_mask): Likewise.
35345 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35346 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35348 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35350 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35351 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35353 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35355 2017-03-06 Martin Liska <mliska@suse.cz>
35358 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35359 when having a SSA NAME w/o VAR_DECL assigned to it.
35361 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35363 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35364 msa_dpsub_<su>_d): Fix MODE for vec_select.
35366 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35368 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35370 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35372 2017-03-06 Richard Biener <rguenther@suse.de>
35374 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35375 * plugin.c (register_plugin_info): Likewise.
35376 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35378 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35380 * config/i386/sse.md (sse_storehps, sse_storelps,
35381 avx_<castmode><avxsizesuffix>_<castmode>,
35382 avx512f_<castmode><avxsizesuffix>_<castmode>,
35383 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35384 in condition that at least one operand is not a MEM.
35386 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35388 PR middle-end/79805
35389 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35390 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35392 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35393 gimple_call_nothrow_p flag based on whether original builtin can throw.
35394 If it can, emit following stmts on the fallthrough edge.
35395 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35396 don't create new bb if inserting just debug stmts on the edge, try to
35397 insert them on the fallthru bb or just reset debug stmts.
35399 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35402 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35403 restore recog_data (including the operand rtxes inside it) around
35404 the call to get_insn_template.
35406 2017-03-03 Martin Sebor <msebor@redhat.com>
35408 PR tree-optimization/79699
35409 * context.c (context::~context): Free MPFR caches to avoid
35410 a memory leak on program exit.
35412 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35414 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35415 Use wide_int::ulow () instead of .elt (0).
35417 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35419 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35420 (*pushxf): Limit oF constraint to 32bit targets and add oC
35421 constraint for 64bit targets.
35422 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35423 (*pushdf): Change rmF constraint to rmC.
35425 2017-03-03 Martin Liska <mliska@suse.cz>
35427 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35428 Remove unused variable.
35430 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35433 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35434 is a memory operand, increase num_memory.
35435 (ix86_expand_args_builtin): Likewise.
35437 2017-03-03 Jan Hubicka <jh@suse.cz>
35440 * ipa-devirt.c (maybe_record_node): Properly handle
35441 __cxa_pure_virtual visibility.
35443 2017-03-03 Martin Liska <mliska@suse.cz>
35445 PR tree-optimization/79803
35446 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35448 (pass_loop_prefetch::execute): Disabled optimization if an
35449 assumption about L1 cache size is not met.
35451 2017-03-03 Martin Liska <mliska@suse.cz>
35453 PR rtl-optimization/79574
35454 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35455 (hash_scan_set): Likewise.
35456 (dump_hash_table): Likewise.
35457 (hoist_code): Likewise.
35459 2017-03-03 Richard Biener <rguenther@suse.de>
35461 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35462 in place of uhigh ().
35463 (fixed_convert_from_real): Likewise.
35465 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35468 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35470 2017-03-03 Richard Biener <rguenther@suse.de>
35472 PR middle-end/79818
35473 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35474 TYPE_OVERFLOW_UNDEFINED check.
35476 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35478 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35480 (vector_ae_<mode>_p): Likewise.
35481 (vector_nez_<mode>_p): Likewise.
35482 (vector_ne_v2di_p): Likewise.
35483 (vector_ae_v2di_p): Likewise.
35484 (vector_ne_<mode>_p): Likewise.
35485 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35487 (vsx_tsqrt<mode>2_fe): Likewise.
35489 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35492 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35494 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35496 PR rtl-optimization/79780
35497 * cprop.c (one_cprop_pass): When second and further conditional trap
35498 in a single basic block is turned into an unconditional trap, turn it
35499 into a deleted note to avoid RTL verification failures.
35501 2017-03-02 Richard Biener <rguenther@suse.de>
35503 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35505 2017-03-02 Richard Biener <rguenther@suse.de>
35507 PR tree-optimization/79345
35509 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35510 param and abort the walk, returning -1 if it is hit.
35511 (walk_aliased_vdefs): Take a limit param and pass it on.
35512 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35513 defaulting to 0 and return a signed int.
35514 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35515 (check_defs): New helper.
35516 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35517 about uninitialized memory.
35518 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35519 bogus uninitialized warning.
35520 (fixed_convert_from_real): Likewise.
35522 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35524 PR tree-optimization/66768
35525 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35526 iv_use if base object can't be determined.
35528 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35530 PR tree-optimization/79345
35531 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35532 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35533 (get_pattern_stats): Initialize it.
35534 * genemit.c (gen_expand): Verify match_scratch numbers come after
35535 match_operand/match_dup numbers.
35536 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35537 match_scratch numbers.
35538 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35540 * config/s390/s390.md (trunctdsd2): Likewise.
35542 2017-03-02 Richard Biener <rguenther@suse.de>
35544 * wide-int.h (wide_int_storage::operator=): Implement in terms
35547 2017-03-02 Richard Biener <rguenther@suse.de>
35549 PR tree-optimization/79777
35550 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35551 the to insert expression to sth existing.
35553 2017-03-01 Martin Sebor <msebor@redhat.com>
35555 PR middle-end/79692
35556 * gimple-ssa-sprintf.c
35557 (directive::known_width_and_precision): New function.
35558 (format_integer): Use it.
35559 (get_mpfr_format_length): Consider the full range of precision
35560 when computing %g output with the # flag. Set the likely byte
35561 count to 3 rather than 1 when precision is indeterminate.
35562 (format_floating): Correct the lower bound of precision.
35564 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35566 * doc/invoke.texi: Document default code model for 64-bit Linux.
35568 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35571 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35572 udiv rather than div since input pattern is unsigned.
35574 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35576 * config/i386/i386.c (print_reg): Warn for values of
35577 unsupported size in integer register.
35579 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35582 * config/rs6000/predicates.md (current_file_function_operand): Do
35583 not allow self calls to be local if the function is replaceable.
35585 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35588 * config/rs6000/altivec.h (vec_ctz and others): Change the
35589 preprocessor macro that controls conditional compilation from
35590 _ARCH_PWR9 to __POWER9_VECTOR__.
35591 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35592 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35593 control (instead of _ARCH_PWR9 control) so that template
35594 definition uses power9-specific function.
35595 (vec_any_eq): Likewise.
35596 (vec_all_ne): Change macro definition to use a power9-specific
35597 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35598 _ARCH_PWR9 control).
35599 (vec_any_eq) Likewise.
35600 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35601 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35602 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35603 support for xvcmpnedp instruction.
35604 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35605 macro expansion so that Power9 implementation of vec_all_ne does
35606 not use the AltiVec predicate framework.
35607 (VCMPNEH_P): Likewise.
35608 (VCMPNEW_P): Likewise.
35609 (VCMPNED_P): Likewise.
35610 (VCMPNEFP_P): Likewise.
35611 (VCMPNEDP_P): Likewise.
35612 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35613 implementation of vec_any_eq to not use AltiVec predicate
35615 (VCMPAEH_P): Likewise.
35616 (VCMPAEW_P): Likewise.
35617 (VCMPAED_P): Likewise.
35618 (VCMPAEFP_P): Likewise.
35619 (VCMPAEDP_P): Likewise.
35620 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35621 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35622 not use the AltiVec predicate framework.
35623 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35624 of vec_any_eq to not use AltiVec predicate framework.
35625 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35626 support for predefined __POWER9_VECTOR__ macro to indicate that
35627 Power9 instruction selection is enabled.
35628 (altivec_overloaded_builtins): Remove extraneous
35629 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35630 function argument types RS6000_BTI_bool_V16QI and
35631 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35632 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35633 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35634 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35635 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35636 Power9 for implementations of vec_cmpne. Change the signature for
35637 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35638 (representing vec_all_ne) to remove the previously described first
35639 argument of type RS6000_BTI_INTSI, as this was an artifact of
35640 reliance on the AltiVec predicate framework, which is no longer
35641 used in the implementation of these functions. Add
35642 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35643 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35644 since, unlike the AltiVec predicate framework implementation, we
35645 do not share function descriptors between vec_alle and vec_anyeq.
35646 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35647 set of modes that receive special treatment even when
35648 TARGET_P9_VECTOR is true. The special treatment emits code that
35649 does not depend on Power9 instructions.
35650 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35651 define_expand to not rely on AltiVec predicate framework.
35652 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35654 (vector_ne_v2di_p): Change this define_expand to not rely on
35655 AltiVec predicate framework.
35656 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35658 (vector_ne_<mode>_p): Change this define_expand to not rely on
35659 AltiVec predicate framework.
35660 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35662 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35663 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35664 define_insn pattern.
35665 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35666 define_insn pattern because the xvcmpne<VSs>. instruction is not
35668 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35669 instruction is not supported.
35671 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35673 * config/nvptx/nvptx.c: Include intl.h.
35675 2017-03-01 Martin Jambor <mjambor@suse.cz>
35678 * ipa-prop.h (ipa_bits): Removed field known.
35679 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35680 to pointers. Adjusted their comments to warn about their sharing.
35681 (ipcp_transformation_summary): Change bits to a vector of pointers.
35682 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35683 (ipa_get_ipa_bits_for_value): Declare.
35684 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35685 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35686 (ipa_bits_hash_table): Likewise.
35687 (ipa_vr_ggc_hash_traits): Likewise.
35688 (ipa_vr_hash_table): Likewise.
35689 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35690 being pointers and vr_known being removed.
35691 (ipa_set_jf_unknown): Likewise.
35692 (ipa_get_ipa_bits_for_value): New function.
35693 (ipa_set_jfunc_bits): Likewise.
35694 (ipa_get_value_range): New overloaded functions.
35695 (ipa_set_jfunc_vr): Likewise.
35696 (ipa_compute_jump_functions_for_edge): Use the above functions to
35697 construct bits and vr parts of jump functions.
35698 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35699 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35701 (ipcp_grow_transformations_if_necessary): Also allocate
35702 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35704 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35705 them. Fix too long lines.
35706 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35707 vr_known being removed.
35708 (ipa_read_jump_function): Use new setter functions to construct bits
35709 and vr parts of jump functions or set them to NULL.
35710 (write_ipcp_transformation_info): Adjust for bits being pointers.
35711 (read_ipcp_transformation_info): Likewise.
35712 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35714 Include gt-ipa-prop.h.
35715 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35717 (ipcp_store_bits_results): Likewise.
35718 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35719 Do not write to existing jump functions but use a temporary instead.
35721 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35724 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35725 attempt to use its first operand as BIT_FIELD_REF base.
35727 2017-03-01 Richard Biener <rguenther@suse.de>
35729 PR middle-end/79721
35730 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35731 interpolating formula in wrapping arithmetic.
35732 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35734 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35736 PR tree-optimization/79734
35737 * tree-vect-generic.c (expand_vector_condition): Optimize
35738 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35739 Handle VEC_COND_EXPR where comparison has different inner width from
35740 type's inner width.
35742 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35744 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35745 markup, and similar issues. Remove @opindex entries for things
35746 that aren't options. Add missing -mmpy-option entries.
35748 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35750 PR tree-optimization/79737
35751 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35752 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35753 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35754 instead of byte_size. Formatting fix.
35755 (shift_bytes_in_array_right): Formatting fix.
35757 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35760 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35761 condition on optimize for the leaf function test.
35763 2017-02-28 Martin Liska <mliska@suse.cz>
35766 * read-rtl-function.c (function_reader::handle_unknown_directive):
35767 Bail out when one uses -flto.
35769 2017-02-28 Martin Liska <mliska@suse.cz>
35771 * common.opt: Replace space with tabular for options of <number>
35773 * config/i386/i386.opt: Show <number> value for
35774 -mlarge-data-threshold.
35775 * opts.c (print_filtered_help): Do not display number in hexadecimal
35778 2017-02-28 Martin Liska <mliska@suse.cz>
35780 * common.opt: Fix --help=option -Q for options which are of
35783 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35785 * config/i386/i386.c (print_reg): Error out for values
35786 of 8-bit size in invalid integer register.
35788 2017-02-28 Martin Sebor <msebor@redhat.com>
35790 PR tree-optimization/79691
35791 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35793 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35796 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35797 gcc_unreachable with output_operand_lossage.
35799 2017-02-28 Richard Biener <rguenther@suse.de>
35801 PR tree-optimization/79740
35802 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35804 (visit_nary_op): Insert the nary into the hashtable if we
35805 pattern-matched sth.
35806 * tree-ssa-pre.c (eliminate_insert): Robustify.
35808 2017-02-28 Richard Biener <rguenther@suse.de>
35810 PR middle-end/79731
35811 * fold-const.c (decode_field_reference): Reject out-of-bound
35814 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35816 * config/i386/i386.c: Include intl.h.
35817 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35818 instead of just cond ? "..." : "...".
35819 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35820 * coverage.c (read_counts_file): Likewise.
35821 * omp-offload.c: Include intl.h.
35822 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35823 of just cond ? "..." : "...".
35824 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35825 of just cond ? "..." : "...".
35827 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35830 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35832 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35833 'tune for' CPU name.
35834 * config/arm/arm-cpu-data.h: Regenerated.
35836 2017-02-28 Richard Biener <rguenther@suse.de>
35838 PR tree-optimization/79732
35839 * tree-inline.c (expand_call_inline): Do not shadow var.
35841 2017-02-28 Richard Biener <rguenther@suse.de>
35843 PR tree-optimization/79723
35844 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35845 address-space properly.
35847 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35849 * doc/optinfo.texi (Optimization groups): Fix option used for
35851 * doc/invoke.texi (-fopt-info): Document "omp".
35852 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35853 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35854 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35855 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35856 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35858 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35860 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35861 OPTGROUP_OMP with "omp".
35863 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35866 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35867 for arithmetic shift of unsigned V2DI.
35869 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35871 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35872 arc/linux.h headers.
35873 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35874 (LINK_SPEC): Likewise.
35875 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35876 (EXTRA_SPECS): Likewise.
35877 (STARTFILE_SPEC): Likewise.
35878 (ENDFILE_SPEC): Likewise.
35879 (LIB_SPEC): Likewise.
35880 (TARGET_SDATA_DEFAULT): Likewise.
35881 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35882 (MULTILIB_DEFAULTS): Likewise.
35883 (DWARF2_UNWIND_INFO): Likewise.
35884 * config/arc/big.h: New file.
35885 * config/arc/elf.h: Likewise.
35886 * config/arc/linux.h: Likewise.
35887 * config/arc/t-uClibc: Remove.
35889 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35891 PR tree-optimization/77536
35892 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35893 (tree_transform_and_unroll_loop): Use above function to compute the
35894 estimated niter of unrolled loop and use it when scaling profile.
35895 Also use count info rather than frequency if it's non-zero.
35896 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35897 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35898 (vect_transform_loop): Call above function.
35900 2017-02-27 Richard Biener <rguenther@suse.de>
35902 PR tree-optimization/45397
35903 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35904 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35905 (visit_nary_op): Add pattern matching for CSEing sign-changed
35906 or truncated operations with wider ones.
35908 2017-02-27 Richard Biener <rguenther@suse.de>
35910 PR tree-optimization/79690
35911 * tree-vect-stmts.c (vectorizable_store): Use vector type
35912 built from the DR with address-space.
35914 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35916 * doc/invoke.texi (Optimize Options): Refine the description
35917 of asan-use-after-return.
35919 2017-02-25 Alan Modra <amodra@gmail.com>
35921 PR rtl-optimization/79584
35922 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35923 base, not ad->base_term, the reg within base. Remove assertion
35924 that ad->base == ad->base_term. Replace gen_int_mode using
35925 bogus mode with const0_rtx.
35927 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35929 PR middle-end/79396
35930 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35931 FMA_EXPR like tcc_binary or tcc_unary.
35933 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35936 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35938 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35939 (output_loc_operands): Handle DW_OP_call_ref and
35940 DW_OP_GNU_variable_value.
35941 (struct variable_value_struct): New type.
35942 (struct variable_value_hasher): Likewise.
35943 (variable_value_hash): New variable.
35944 (string_types): Remove.
35945 (copy_loc_descr): New function.
35946 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35947 (prepend_loc_descr_to_each): New function.
35948 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35949 instead of add_loc_descr_to_each if the first argument is single
35950 location list and the second has multiple.
35951 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35952 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35953 when looking for variable value which doesn't have other location info.
35954 (loc_list_from_tree): Formatting fix.
35955 (gen_array_type_die): Simplify DW_AT_string_length handling.
35956 (adjust_string_types): Remove.
35957 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35958 string_types. Call resolve_variable_values.
35959 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35960 (resolve_addr_in_expr): Likewise. Add A argument.
35961 (copy_deref_exprloc): Remove deref argument. Adjust for the
35962 original expression being DW_OP_GNU_variable_value with optionally
35963 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35964 optionally after it.
35965 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35966 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35967 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35968 (variable_value_hasher::hash, variable_value_hasher::equal): New
35970 (resolve_variable_value_in_expr, resolve_variable_value,
35971 resolve_variable_values, note_variable_value_in_expr,
35972 note_variable_value): New functions.
35973 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35976 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35979 * opts.h (handle_generated_option): Add GENERATED_P argument.
35980 * opts-common.c (handle_option): Adjust function comment.
35981 (handle_generated_option): Add GENERATED_P argument, pass it to
35983 (control_warning_option): Pass false to handle_generated_option
35985 * opts.c (maybe_default_option): Pass true to handle_generated_option
35987 * optc-gen.awk: Likewise.
35989 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35991 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35992 a REG, look at the REG it is a SUBREG of.
35993 (splitter for cmpeqsi_t): Ditto.
35995 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35997 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35998 the special USEs with the pattern of the insn, not the insn itself.
36000 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36003 * doc/invoke.texi: Document -mload-store-pairs.
36005 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36006 Sandra Loosemore <sandra@codesourcery.com>
36008 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36009 argument isn't a CONST_INT.
36010 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36011 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36012 (nios2_validate_compare): Bypass alternate compare logic if *op2
36013 is not a CONST_INT.
36014 (ldstwm_operation_p): Return false if first_base is not a REG or
36015 if first_offset is not a CONST_INT.
36017 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36019 * config/cris/cris.md: Use correct operand in a define_peephole2.
36021 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36023 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36025 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36027 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36028 this_insn if it is an INSN or JUMP_INSN.
36029 (force_offsettable): Look at base, not at addr.
36030 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36031 on things that aren't necessarily CONST_INTs.
36033 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36035 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36036 -mfpmath=sse is the default also for x86-32 targets with SSE2
36037 instruction set when @option{-ffast-math} is enabled
36039 2017-02-24 Jeff Law <law@redhat.com>
36041 PR rtl-optimizatoin/79286
36042 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36045 2017-02-24 Richard Biener <rguenther@suse.de>
36047 PR tree-optimization/79389
36048 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36051 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36053 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36054 function comment to reflect reality.
36055 (loop_exits_before_overflow): Fix typo in function description.
36057 2017-02-24 Richard Biener <rguenther@suse.de>
36059 PR tree-optimization/79389
36060 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36061 properly that a threading opportunity exists. Detect conditional
36062 copy/constant propagation opportunities.
36064 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36066 * config/visium/visium.md (type): Add trap.
36067 (b): New mode attribute.
36068 (*btst): Rename into...
36069 (*btst<mode>): ...this and adjust.
36070 (*cbranchsi4_btst_insn): Rename into...
36071 (*cbranch<mode>4_btst_insn): ...this and adjust.
36072 (trap): New define_insn.
36074 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36076 PR tree-optimization/79389
36077 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36078 (noce_reversed_cond_code): New function.
36079 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36080 reversed_comparison_code. Formatting fix.
36081 (noce_try_store_flag): Test rev_cond != NULL in addition to
36082 reversed_comparison_code.
36083 (noce_try_store_flag_constants): Likewise.
36084 (noce_try_store_flag_mask): Likewise.
36085 (noce_try_addcc): Use rev_cond if non-NULL instead of
36086 reversed_comparison_code.
36087 (noce_try_cmove_arith): Likewise. Formatting fixes.
36088 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36089 (noce_find_if_block): Initialize rev_cond.
36090 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36091 instead of false as last argument never attempt to reverse it
36094 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36096 PR tree-optimization/79663
36097 * tree-predcom.c (combine_chains): Process refs in reverse order
36098 only for ZERO length chains, and add explaining comment.
36100 2017-02-23 Jeff Law <law@redhat.com>
36102 PR tree-optimization/79578
36103 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36104 in call to operand_equal_p.
36106 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36109 * config/i386/cpuid.h: Fix another undefined behavior.
36111 2017-02-23 Richard Biener <rguenther@suse.de>
36113 PR tree-optimization/79683
36114 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36115 vector types for data-refs.
36117 2017-02-23 Martin Liska <mliska@suse.cz>
36119 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36121 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36123 PR middle-end/79665
36124 * internal-fn.c (get_range_pos_neg): Moved to ...
36125 * tree.c (get_range_pos_neg): ... here. No longer static.
36126 * tree.h (get_range_pos_neg): New prototype.
36127 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36128 are known to be in between 0 and signed maximum inclusive, try to
36129 expand both unsigned and signed divmod and use the cheaper one from
36132 2017-02-22 Jeff Law <law@redhat.com>
36134 PR tree-optimization/79578
36135 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36136 to compare base operands.
36138 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36141 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36142 gpc_reg_operand instead of fpr_reg_operand.
36144 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36146 * config/mips/mips.c (mips_return_in_memory): Force FP
36147 vector types to be returned in memory for o32 ABI.
36149 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36151 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36152 instead of DW_TAG_member for static data member declarations and don't
36153 set no_linkage_name for static inline data members.
36154 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36157 2017-02-22 Martin Liska <mliska@suse.cz>
36159 * doc/invoke.texi: Replace inequality signs with square brackets
36162 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36165 * lra-constraints.c (simplify_operand_subreg): Handle
36166 WORD_REGISTER_OPERATIONS targets.
36168 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36171 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36172 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36173 elimination by swapping fld*.
36175 2017-02-22 Richard Biener <rguenther@suse.de>
36177 PR tree-optimization/79673
36178 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36179 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36180 irrelevant address-space qualifiers and avoiding a
36181 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36183 2017-02-22 Richard Biener <rguenther@suse.de>
36185 PR tree-optimization/79666
36186 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36187 to not symbolically negate if that may introduce undefined
36190 2017-02-22 Martin Liska <mliska@suse.cz>
36193 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36194 * data-streamer-out.c (streamer_write_gcov_count_stream):
36196 * value-prof.c (stream_out_histogram_value): Make assert more
36197 precise based on type of counter.
36199 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36202 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36203 Use nonimmediate_operand instead of memory_operand for operand 1.
36204 (float-extend standard_x87sse_constant_load splitter): Ditto.
36206 2017-02-21 Jeff Law <law@redhat.com>
36208 PR tree-optimization/79621
36209 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36210 blocks with edges to themselves.
36212 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36215 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36216 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36217 Use gimple_call_builtin_p.
36220 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36221 on temporarily removed DEBUG_INSNs.
36223 PR tree-optimization/79649
36224 * tree-loop-distribution.c (classify_partition): Give up on
36225 non-generic address space loads/stores.
36227 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36229 * doc/loop.texi (Loop manipulation): Remove nonexistent
36230 tree_ssa_loop_version from the documentation.
36231 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36233 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36236 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36237 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36238 * config/rs6000/rs6000.c: Include except.h.
36239 (rs6000_expand_split_stack_prologue): Call
36240 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36242 2017-02-21 Martin Jambor <mjambor@suse.cz>
36245 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36246 have been analyzed.
36248 2017-02-21 Martin Jambor <mjambor@suse.cz>
36250 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36251 for backward compatibility only.
36252 * doc/invoke.texi (Option Summary): Remove all references to
36253 -fipa-cp-alignment.
36255 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36259 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36261 * lra-constraints.c (curr_insn_transform): Handle
36262 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36264 2017-02-21 Martin Liska <mliska@suse.cz>
36266 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36268 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36270 PR translation/79638
36271 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36273 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36276 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36277 (arm_function_ok_for_sibcall): Return false for an indirect call by
36278 descriptor if all the argument registers are used.
36279 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36280 alignment of the function.
36282 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36284 PR tree-optimization/61441
36285 * simplify-rtx.c (simplify_const_unary_operation): For
36286 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36287 the sNaN unmodified.
36289 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36291 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36292 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36293 instead of SYSTEM_HEADER_DIR.
36295 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36296 Martin Liška <mliska@suse.cz>
36298 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36299 Fix typos and grammar, use active voice, and clarify.
36301 2017-02-20 Marek Polacek <polacek@redhat.com>
36303 PR middle-end/79537
36304 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36307 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36309 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36312 * config/i386/i386.c (ix86_expand_builtin): Handle
36313 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36314 ix86_builtins_isa[fcode].isa as a requirement of those
36315 flags and any other flag in the bitmask.
36316 (ix86_init_mmx_sse_builtins): Use 0 instead of
36317 ~OPTION_MASK_ISA_64BIT as mask.
36318 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36319 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36320 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36321 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36323 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36326 * lra-constraints.c (split_reg): Check requested split mode
36327 is supported by the register.
36329 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36331 * lra-constraints.c (simplify_operand_subreg): Remove early
36334 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36337 * lra-constraints.c (curr_insn_transform): Tighten condition
36338 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36340 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36343 * lra-constraints.c (curr_insn_transform): Handle
36344 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36346 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36349 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36351 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36353 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36356 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36359 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36361 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36364 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36366 * final.c (last_columnnum, override_columnnum): New variables.
36367 (final_start_function): Set last_columnnum, pass it to begin_prologue
36368 hook and pass 0 to dwarf2out_begin_prologue.
36369 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36370 to source_line debug hook.
36371 (notice_source_line): Compute last_columnnum and for debug_column_info
36372 return true on column changes.
36373 * debug.h (struct gcc_debug_hooks): Add column argument to
36374 source_line and begin_prologue hooks.
36375 (debug_nothing_int_charstar_int_bool): Remove prototype.
36376 (debug_nothing_int_int_charstar,
36377 debug_nothing_int_int_charstar_int_bool): New prototypes.
36378 (dwarf2out_begin_prologue): Add column argument.
36379 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36380 begin_prologue hooks.
36381 (debug_nothing_int_charstar_int_bool): Remove.
36382 (debug_nothing_int_int_charstar,
36383 debug_nothing_int_int_charstar_int_bool): New functions.
36384 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36385 through to dwarf2out_source_line.
36386 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36387 (dwarf2out_source_line): Add column argument, emit it if requested.
36388 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36390 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36391 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36392 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36393 through to dwarf2out_begin_prologue.
36394 (vmsdbgout_source_line): Add column argument, pass it through to
36395 dwarf2out_source_line.
36396 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36397 dbxout_source_line caller.
36398 (dbxout_source_line): Add column argument.
36400 * common.opt (gno-column-info, gcolumn-info): New options.
36401 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36402 (check_die): Also test for multiple DW_AT_decl_column attributes.
36403 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36404 DW_AT_decl_column if requested.
36405 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36407 (gen_variable_die): Likewise.
36408 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36409 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36412 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36413 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36414 (ix86_handle_option): Handle OPT_m3dnowa.
36415 * doc/invoke.texi (-m3dnowa): Document.
36416 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36417 -m3dnowa instead of -m3dnow -march=athlon.
36420 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36421 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36423 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36426 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36427 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36428 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36429 generator for vsx_xxpermdi_<mode>_be.
36430 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36431 force big-endian semantics.
36432 (vsx_xxpermdi_<mode>_be): New define_expand with same
36433 implementation as previous version of vsx_xxpermdi_<mode>.
36435 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36437 PR tree-optimization/79327
36438 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36439 variable, its initialization and use.
36441 2017-02-17 Julia Koval <julia.koval@intel.com>
36443 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36444 (OPTION_MASK_ISA_PKU_UNSET): New.
36445 (ix86_handle_option): Handle -mrdpid.
36446 * config/i386/cpuid.h (bit_RDPID): New.
36447 * config/i386/driver-i386.c (host_detect_local_cpu):
36448 Detect RDPID feature.
36449 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36450 * config/i386/i386-c.c (ix86_target_macros_internal):
36452 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36453 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36454 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36455 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36456 * config/i386/i386.md (define_insn "rdpid"): New.
36457 * config/i386/i386.opt Add -mrdpid.
36458 * config/i386/immintrin.h (_rdpid_u32): New.
36460 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36462 PR rtl-optimization/79541
36463 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36464 instead of transforming it into USE.
36466 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36468 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36469 If HONOR_SNANS (SFmode) force the input to a register.
36470 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36471 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36472 an frsp or similar insn.
36474 2017-02-17 Martin Liska <mliska@suse.cz>
36476 PR rtl-optimization/79577
36477 * params.def (selsched-max-sched-times): Increase minimum to 1.
36479 2017-02-17 Martin Liska <mliska@suse.cz>
36481 PR rtl-optimization/79574
36482 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36484 2017-02-17 Martin Liska <mliska@suse.cz>
36486 PR tree-optimization/79529
36487 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36488 ssa_defined_default_def_p to handle cases which are implicitly
36490 * tree-ssa.c (ssa_defined_default_def_p): New function.
36491 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36492 which are implicitly defined.
36493 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36495 2017-02-17 Richard Biener <rguenther@suse.de>
36497 PR middle-end/79576
36498 * params.def (max-ssa-name-query-depth): Limit to 10.
36500 2017-02-17 Richard Biener <rguenther@suse.de>
36502 PR tree-optimization/79552
36503 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36506 2017-02-17 Richard Biener <rguenther@suse.de>
36509 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36511 2017-02-17 Marek Polacek <polacek@redhat.com>
36513 PR middle-end/79536
36514 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36515 (fold_negate_expr): New wrapper.
36517 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36519 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36520 Correct terminology and de-emphasize pre-standard behavior.
36522 2017-02-16 Alan Modra <amodra@gmail.com>
36524 PR rtl-optimization/79286
36525 * ira.c (def_dominates_uses): New function.
36526 (update_equiv_regs): Don't create an equivalence for insns that
36527 may trap where the register def does not dominate the use.
36529 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36531 PR rtl-optimization/78127
36532 * lra.c (lra): Call lra_eliminate before finish the loop after
36535 2017-02-16 Richard Biener <rguenther@suse.de>
36537 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36539 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36540 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36541 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36542 (isl_val_int_from_wi): New function.
36543 (extract_affine_gmp): Rename to ...
36544 (extract_affine_wi): ... this, take a widest_int.
36545 (extract_affine_int): Just wrap extract_affine_wi.
36546 (add_param_constraints): Use isl_val_int_from_wi.
36547 (add_loop_constraints): Likewise, and extract_affine_wi.
36549 2017-02-15 Jeff Law <law@redhat.com>
36551 PR middle-end/79521
36552 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36553 ira_init_register_move_cost_if_necessary.
36555 2017-02-15 Martin Sebor <msebor@redhat.com>
36557 PR middle-end/32003
36558 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36559 removed in a prior commit.
36561 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36563 PR tree-optimization/79347
36564 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36565 counters during peeling.
36567 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36569 * Makefile.in (site.exp): Remove "set ISLVER".
36571 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36574 * real.c (real_from_integer): Call real_convert even for decimal.
36576 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36579 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36581 2017-02-14 Andrew Pinski <apinski@cavium.com>
36583 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36584 cores and change the partno/implementer to be correct.
36585 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36586 the 'B" as the implementer.
36587 * config/aarch64/aarch64-tune.md: Regenerate.
36589 2017-02-14 Carl Love <cel@us.ibm.com>
36591 * config/rs6000/rs6000.c: Add case statement entry to make the
36592 xvcvuxdsp built-in argument unsigned.
36593 * config/rs6000/vsx.md: Fix the source and return operand types so they
36594 match the instruction definitions from the ISA document. Fix typo
36595 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36598 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36601 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36602 member early_clobber_alts.
36603 * lra-lives.c (reg_early_clobber_p): New.
36604 (process_bb_lives): Use it.
36605 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36606 (debug_operand_data): Initialize early_clobber_alts.
36607 (setup_operand_alternative): Set up early_clobber_alts.
36608 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36609 alternatives to new_insn_reg.
36610 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36612 (lra_update_insn_regno_info): Pass the new arg.
36614 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36616 PR middle-end/79505
36617 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36618 (new_oacc_loop_raw): Don't clear already cleared fields.
36621 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36622 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36623 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36625 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36628 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36630 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36633 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36634 the extra instruction to the right place to store 128-bit constant
36637 2017-02-14 Martin Sebor <msebor@redhat.com>
36639 PR middle-end/79448
36640 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36641 warning for strings of unknown length.
36643 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36645 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36647 2017-02-14 Jeff Law <law@redhat.com>
36650 * ira-costs.c (scan_one_insn): Initialize register move costs
36651 for pseudos seen in USE/CLOBBER insns.
36653 PR tree-optimization/79095
36654 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36655 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36656 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36657 if the operands are known to be not equal, then the resulting range
36659 (intersect_ranges): If the new range is ~[0,0] and the old range is
36660 wide, then prefer ~[0,0].
36661 * tree-vrp.c (overflow_comparison_p_1): New function.
36662 (overflow_comparison_p): New function.
36663 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36664 if NAME is used in an overflow test.
36665 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36666 overflow check that can be expressed as an equality test, then adjust
36667 ops to be that equality test.
36669 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36671 * config/s390/s390-builtin-types.def: Remove flags argument.
36672 * config/s390/s390.c (s390_init_builtins): Likewise.
36674 2017-02-14 Martin Liska <mliska@suse.cz>
36676 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36677 vector. Fix trailing white spaces.
36679 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36681 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36684 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36686 PR rtl-optimization/68664
36687 * config/arm/arm.c (arm_sched_can_speculate_insn):
36688 New function. Declare prototype.
36689 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36691 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36693 PR rtl-optimization/68664
36694 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36696 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36698 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36700 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36701 max skip bytes for function, loop and jump.
36703 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36705 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36706 ABS_EXPR for gimple dump.
36708 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36711 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36713 PR tree-optimization/79408
36714 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36715 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36716 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36717 also if rhs1 is INTEGER_CST.
36719 2017-02-14 Richard Biener <rguenther@suse.de>
36721 PR middle-end/79432
36722 * tree-into-ssa.c (insert_phi_nodes): When the function can
36723 have abnormal edges rewrite SSA names with broken use-def
36724 dominance out of SSA and register them for PHI insertion.
36726 2017-02-13 Martin Sebor <msebor@redhat.com>
36728 PR middle-end/79496
36729 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36730 clearing info.nowrite flag when snprintf size argument is a range.
36732 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36734 * cprop.c (cprop_jump): Add missing space in string literal.
36735 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36736 (get_constraint_for_component_ref): Likewise.
36737 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36738 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36739 * lra-constraints.c (process_alt_operands): Likewise.
36740 * ipa-inline.c (inline_small_functions): Likewise.
36741 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36742 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36743 * trans-mem.c (diagnose_tm_1_op): Likewise.
36744 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36745 (grid_parallel_clauses_gridifiable): Likewise.
36747 * config/nvptx/mkoffload.c (process): Add space in between
36750 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36751 "MOD4_SSE_REGS" and "ALL_REGS".
36753 * spellcheck.c (test_data): Add , in between "foo" and "food".
36755 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36758 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36759 boundary crossing check and subsequent code generation agree.
36761 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36763 * config/aarch64/aarch64.c (has_memory_op): Delete.
36764 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36767 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36769 PR rtl-optimization/79388
36770 PR rtl-optimization/79450
36771 * combine.c (distribute_notes): When removing TEM_INSN for which
36772 corresponding dest has last value recorded, invalidate that last
36775 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36777 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36778 of explicit '@'. Add missing assembly comment marker on branch costs
36781 2017-02-13 Nathan Sidwell <nathan@acm.org>
36783 * gengtype-lex.l (<in_struct>): Add '/'.
36785 2017-02-13 Martin Liska <mliska@suse.cz>
36788 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36790 2017-02-13 Richard Biener <rguenther@suse.de>
36792 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36794 * configure: Re-generate.
36795 * config.in: Likewise.
36796 * graphite-dependences.c: Simplify as if
36797 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36798 * graphite-isl-ast-to-gimple.c: Likewise.
36799 * graphite-optimize-isl.c: Likewise.
36800 * graphite-poly.c: Likewise.
36801 * graphite-sese-to-poly.c: Likewise.
36802 * graphite.h: Likewise.
36803 * toplev.c: Include isl/version.h and use isl_version () for
36804 printing the ISL version.
36805 * doc/install.texi: Update ISL requirement.
36807 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36809 * doc/standards.texi (Standards): Update reference to
36812 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36814 * doc/extend.texi (Named Address Spaces): sourceware.org now
36816 * doc/install.texi (Binaries): Ditto.
36819 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36821 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36822 standard terminology "stringize"/"stringizing" throughout.
36823 * doc/cppinternals.texi: Likewise.
36825 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36827 * doc/extend.texi: Fix some spelling mistakes and typos.
36828 * doc/invoke.texi: Likewise.
36830 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36833 * params.def (inline-min-speedup) Change from 10 to 8.
36835 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36837 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36840 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36843 * ipa-inline-analysis.c (get_minimal_bb): New function.
36844 (record_modified): Use it.
36845 (remap_edge_change_prob): Handle also ancestor functions.
36847 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36849 * doc/contrib.texi (Contributors): Remove broken link into
36850 the Mauve CVS repository.
36852 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36854 PR middle-end/79454
36855 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36856 result computation whenever lhs doesn't have vector mode, not
36857 just when it has BLKmode.
36859 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36861 * doc/makefile.texi (profiledbootstrap): Refer to the
36862 installation instructions only in textual form.
36864 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36867 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36869 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36871 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36872 (Specific): Update mingw-w64 reference.
36874 (Specific): Remove broken link to Renesas RX processor.
36876 2017-02-10 Richard Biener <rguenther@suse.de>
36878 * toplev.c (process_options): Do not mention obsolete graphite
36879 options when printing sorry message about missing graphite support.
36880 Mention -floop-nest-optimize.
36882 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36884 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36885 (vtst_p16): Likewise.
36886 (vtstq_p8): Likewise.
36887 (vtstq_p16): Likewise.
36889 (vtstq_p64): Likewise.
36890 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36891 (vset_lane_p64): New.
36892 (vsetq_lane_p64): New.
36894 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36896 PR tree-optimization/79411
36897 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36898 stmt operands are SSA_NAMEs used in abnormal phis.
36899 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36902 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36905 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36908 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36910 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36912 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36914 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36918 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36919 not arbitrary TREE_CONSTANT.
36922 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36923 "omp declare target link" attribute unless is_global_var.
36924 * omp-offload.c (find_link_var_op): Likewise.
36926 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36927 Chung-Lin Tang <cltang@codesourcery.com>
36929 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36931 (gimplify_adjust_omp_clauses): Don't delete TILE.
36932 (gimplify_omp_for): Deal with TILE.
36933 * internal-fn.c (expand_GOACC_TILE): New function.
36934 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36936 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36937 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36939 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36940 avoid DIV for outermost collapse var.
36941 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36942 Remove out of date comments, fix whitespace.
36943 * omp-general.c (omp_extract_for_data): Deal with tiling.
36944 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36945 adjust OLF_DIM_BASE value.
36946 (struct omp_for_data): Add tiling field.
36947 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36948 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36949 for auto loops. Remove default auto determining, moved to
36950 oacc_loop_fixed_partitions.
36951 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36952 stmts, add e_mask field.
36953 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36954 (oacc_thread_numbers): Use oacc_dim_call.
36955 (oacc_xform_tile): New.
36956 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36957 (finish_oacc_loop): Adjust for ifns vector.
36958 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36959 add case for GOACC_TILE fns.
36960 (oacc_loop_xform_loop): Delete.
36961 (oacc_loop_process): Iterate over call list directly, and add
36962 handling for GOACC_TILE fns.
36963 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36965 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36966 vector partitioning to outer loops. Assign 2 partitions to loops
36967 when available. Add TILE handling.
36968 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36969 (execite_oacc_device_lower): Process GOACC_TILE fns,
36970 ignore unknown specs.
36971 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36972 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36973 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36975 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36977 * configure.ac (ACX_BUGURL): Update.
36978 * configure: Regenerate.
36980 2017-02-09 Richard Biener <rguenther@suse.de>
36982 PR tree-optimization/69823
36983 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36984 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36986 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36988 * config/arc/arc-c.def: Add __NPS400__ definition.
36989 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36990 (TARGET_NPS400): Define.
36992 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36994 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36996 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36997 pointer, arch_info.
36998 (arc_cpu_types): Fill the arch_info field with a pointer into the
36999 arc_arch_types table.
37000 (arc_selected_cpu): Declare.
37001 * config/arc/arc.c (arc_selected_cpu): Make global.
37002 (arc_selected_arch): Delete.
37003 (arc_base_cpu): Delete.
37004 (arc_override_options): Remove references to deleted variables,
37005 update access to arch information.
37006 (ARC_OPT): Update access to arch information.
37007 (ARC_OPTX): Likewise.
37008 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37009 (TARGET_ARC600): Update access to arch information.
37010 (TARGET_ARC601): Likewise.
37011 (TARGET_ARC700): Likewise.
37012 (TARGET_EM): Likewise.
37013 (TARGET_HS): Likewise.
37014 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37017 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37020 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37021 condition/operands for integer GE/LE/GEU/LEU operations.
37023 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37025 PR translation/79397
37026 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37029 2017-02-08 Martin Jambor <mjambor@suse.cz>
37032 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37033 whether allocation happened.
37034 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37035 nothing was allocated.
37037 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37039 PR tree-optimization/79408
37040 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37041 constant, but SSA_NAME with a known integer range, use the minimum
37042 of that range instead of op1 to determine if modulo can be replaced
37043 with its first operand.
37045 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37047 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37049 2017-02-08 Richard Biener <rguenther@suse.de>
37051 PR tree-optimization/71824
37052 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37053 Check all loops contained in the merged region.
37055 2017-02-07 Andrew Pinski <apinski@cavium.com>
37057 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37059 2017-02-07 Andrew Pinski <apinski@cavium.com>
37061 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37062 (thunderxt88): Likewise.
37063 (thunderxt81): Disable LSE and change v8.1 to v8.
37064 (thunderxt83): Likewise.
37066 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37067 Richard Biener <rguenther@suse.de>
37069 PR middle-end/79399
37070 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37071 type from int to size_t.
37072 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37074 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37076 PR rtl-optimization/79386
37077 * cprop.c (bypass_conditional_jumps): Initialize
37078 bypass_last_basic_block already before splitting bbs after
37079 unconditional traps...
37080 (bypass_conditional_jumps): ... rather than here.
37083 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37084 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37085 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37086 fix -masm=intel patterns.
37088 2017-02-07 Richard Biener <rguenther@suse.de>
37090 PR tree-optimization/79256
37091 PR middle-end/79278
37092 * builtins.c (get_object_alignment_2): Use min_align_of_type
37093 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37094 and ADJUST_FIELD_ALIGN.
37096 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37098 * doc/tm.texi: Regenerate.
37099 * stor-layout.c (layout_decl): Adjust.
37100 (update_alignment_for_field): Likewise.
37101 (place_field): Likewise.
37102 (min_align_of_type): Likewise.
37103 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37104 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37105 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37106 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37107 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37108 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37109 * config/i386/i386.c (x86_field_alignment): Likewise.
37110 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37111 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37112 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37113 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37114 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37115 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37119 2017-01-30 Richard Biener <rguenther@suse.de>
37121 PR tree-optimization/79256
37122 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37123 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37126 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37128 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37129 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37130 builtins to SImode and emit a zero-extend, if necessary.
37132 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37134 * docs/invoke.texi (RISC-V Options): Alphabetize.
37136 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37138 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37141 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37143 * config/riscv/riscv.c: New file.
37144 * common/config/riscv/riscv-common.c: Likewise.
37145 * config.gcc: Likewise.
37146 * config/riscv/constraints.md: Likewise.
37147 * config/riscv/elf.h: Likewise.
37148 * config/riscv/generic.md: Likewise.
37149 * config/riscv/linux.h: Likewise.
37150 * config/riscv/multilib-generator: Likewise.
37151 * config/riscv/peephole.md: Likewise.
37152 * config/riscv/pic.md: Likewise.
37153 * config/riscv/predicates.md: Likewise.
37154 * config/riscv/riscv-builtins.c: Likewise.
37155 * config/riscv/riscv-c.c: Likewise.
37156 * config/riscv/riscv-ftypes.def: Likewise.
37157 * config/riscv/riscv-modes.def: Likewise.
37158 * config/riscv/riscv-opts.h: Likewise.
37159 * config/riscv/riscv-protos.h: Likewise.
37160 * config/riscv/riscv.h: Likewise.
37161 * config/riscv/riscv.md: Likewise.
37162 * config/riscv/riscv.opt: Likewise.
37163 * config/riscv/sync.md: Likewise.
37164 * config/riscv/t-elf-multilib: Likewise.
37165 * config/riscv/t-linux: Likewise.
37166 * config/riscv/t-linux-multilib: Likewise.
37167 * config/riscv/t-riscv: Likewise.
37168 * configure.ac: Likewise.
37169 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37170 Waterman as RISC-V maintainers.
37171 * doc/install.texi: Add RISC-V entries.
37172 * doc/invoke.texi: Add RISC-V options section.
37173 * doc/md.texi: Add RISC-V constraints section.
37174 * configure: Regenerated.
37176 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37179 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37180 false values to be constant vectors with all 0 or all 1 bits set.
37181 (vcondu<mode><mode>): Likewise.
37182 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37184 (fpmask_comparison_operator): Update comment.
37185 (vecint_comparison_operator): New predicate.
37186 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37187 vector conditionals when the true and false values are constant
37188 vectors with all 0 bits or all 1 bits set.
37190 2017-02-06 Martin Sebor <msebor@redhat.com>
37192 PR tree-optimization/79376
37193 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37195 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37197 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37198 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37199 to simplify split condition.
37201 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37203 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37204 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37207 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37209 PR rtl-optimization/68664
37210 * target.def (can_speculate_insn): New hook.
37211 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37212 * doc/tm.texi: Regenerate.
37213 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37214 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37215 (rs6000_sched_can_speculate_insn): New function.
37217 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37219 PR tree-optimization/79284
37220 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37221 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37222 vectorizable_mask_load_store, vectorizable_operation,
37223 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37224 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37225 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37226 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37227 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37228 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37229 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37230 is_gimple_assign (stmt). Replace another such test with
37231 is_gimple_assign (stmt).
37233 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37236 * config/avr/avr.c (rtl-iter.h): Include it.
37237 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37238 (avr_legitimate_combined_insn): ...and implementation.
37240 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37242 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37243 * config/s390/s390.c (s390_const_operand_ok)
37244 (s390_canonicalize_comparison, s390_extract_part)
37245 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37246 (s390_contiguous_bitmask_p, s390_rtx_costs)
37247 (legitimize_pic_address): Likewise.
37248 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37249 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37250 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37251 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37252 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37254 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37256 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37257 REGNO($0) == REGNO($1).
37259 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37261 * config/s390/linux.h(SIZE_TYPE): Add comment.
37263 2017-02-06 Julian Brown <julian@codesourcery.com>
37264 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37265 Virendra Pathak <virendra.pathak@broadcom.com>
37267 * config/aarch64/aarch64-cores.def: Change the scheduler
37269 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37270 * config/aarch64/thunderx2t99.md: New file.
37272 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37274 * doc/standards.texi (Go Language): Update link to language
37277 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37279 * tree-eh.c (lower_resx): Sanitize profile.
37280 (cleanup_empty_eh_move_lp): Likewise.
37282 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37285 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37287 * cfgloopmanip.h (loop_version): Update prototype.
37288 * modulo-sched.c (sms_schedule): Update call of loop_version.
37289 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37290 * tree-parloops.c (gen_parallel_loop): Likewise.
37291 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37292 * tree-ssa-loop-split.c (split_loop): Likewise.
37293 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37294 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37296 2017-02-05 Martin Liska <mliska@suse.cz>
37299 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37301 (print_operand_address): Initialize a struct to zero.
37303 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37305 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37306 garbage collector only in textual form.
37308 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37310 * doc/extend.texi (x86 specific memory model extensions for
37311 transactional memory): Simplify a phrase.
37313 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37316 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37317 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37318 (atomic_storedi_1): Likewise.
37320 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37322 PR tree-optimization/79338
37323 * tree-parloops.c (gather_scalar_reductions): Don't call
37324 vect_analyze_loop_form for loop->inner before destroying loop's
37327 2017-02-03 Martin Sebor <msebor@redhat.com>
37329 PR tree-optimization/79327
37330 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37331 when precision has resulted in leading zeros.
37332 (format_integer): Adjust the likely counter to assume an unknown
37333 argument that may be zero is non-zero.
37335 2017-02-03 Jason Merrill <jason@redhat.com>
37338 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37339 avoid copying non-taken branch.
37341 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37343 PR tree-optimization/79340
37344 * tree-vect-loop.c (vectorizable_reduction): Release
37345 vec_defs elements after safe_splicing them into other vectors.
37348 PR tree-optimization/79327
37349 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37350 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37352 (format_integer): Use wide_int_to_tree instead of build_int_cst
37353 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37354 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37355 of shortest and longest sequence.
37357 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37359 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37360 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37362 2017-02-03 Walter Lee <walt@tilera.com>
37365 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37366 after initial stackframe link reg save.
37367 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37369 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37372 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37373 wu for stxssp alternative.
37375 2017-02-03 Martin Sebor <msebor@redhat.com>
37377 PR tree-optimization/79352
37378 * gimple-fold.c (get_range_strlen): Add argument.
37379 (get_range_strlen): Change return type to bool.
37380 (get_maxval_strlen): Pass in a dummy argument.
37381 * gimple-fold.h (get_range_strlen): Change return type to bool.
37382 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37383 * tree.h (array_at_struct_end_p): Add argument.
37384 * tree.c (array_at_struct_end_p): Handle it.
37386 2017-02-03 Martin Liska <mliska@suse.cz>
37389 * multiple_target.c (create_dispatcher_calls): Redirect edge
37390 from a caller of a dispatcher.
37391 (expand_target_clones): Make the clones local.
37392 (ipa_target_clone): Do both target clones and resolvers.
37393 (ipa_dispatcher_calls): Remove the pass.
37394 (pass_dispatcher_calls::gate): Likewise.
37395 (make_pass_dispatcher_calls): Likewise.
37396 * passes.def (pass_target_clone): Put as very first IPA early
37399 2017-02-03 Martin Liska <mliska@suse.cz>
37401 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37402 in case of a function with ifunc attribute.
37404 2017-02-03 Martin Liska <mliska@suse.cz>
37406 * cgraph.c (cgraph_node::dump): Dump function version info.
37407 * symtab.c (symtab_node::dump_base): Add missing new line.
37409 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37411 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37412 (ifcombine_ifandif): Use it.
37414 2017-02-03 Martin Liska <mliska@suse.cz>
37416 * doc/invoke.texi: Document default value for
37417 use-after-scope-direct-emission-threshold.
37419 2017-02-03 Martin Liska <mliska@suse.cz>
37421 PR tree-optimization/79339
37422 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37423 (format_floating): Likewise.
37425 2017-02-03 Martin Liska <mliska@suse.cz>
37428 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37430 (ipa_node_params_t::remove): Likewise.
37431 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37432 initialization from removed ipa_node_params_t::insert.
37433 (ipa_node_params::~ipa_node_params): Move from removed
37434 ipa_node_params_t::release.
37435 * symbol-summary.h (symbol_summary::m_released): New member.
37436 Do not release a summary twice. Do not allow to call finalizer
37437 for types of a summary that live in GGC memory.
37439 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37441 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37444 2017-02-02 Martin Sebor <msebor@redhat.com>
37446 PR middle-end/79275
37447 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37448 (format_string): Tighten up the range of output for non-constant
37449 strings and correct the expected range for wide non-constant strings.
37451 2017-02-02 Martin Sebor <msebor@redhat.com>
37453 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37455 PR middle-end/32003
37456 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37458 (-fdump-tree-@var): Add to index and document how to come up
37459 with pass-specific option and dump file names.
37460 (-fdump-passes): Clarify where to look for output.
37462 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37464 PR middle-end/77445
37465 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37466 statistics of the analyzed path; allow threading for speed when
37467 any of BBs along the path are optimized for speed.
37469 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37471 PR middle-end/78468
37472 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37473 settings of the virtual registers.
37476 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37478 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37479 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37482 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37484 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37485 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37487 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37489 * config/s390/s390.md: Add missing comments with the expanded
37491 * config/s390/vector.md: Likewise.
37492 * config/s390/vx-builtins.md: Likewise.
37494 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37497 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37498 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37499 conditions on a single line.
37501 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37503 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37504 __S390_VX__ to __VX__.
37506 2017-02-01 Andrew Pinski <apinski@cavium.com>
37508 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37509 stmt_info to record_stmt_cost.
37510 (vect_get_known_peeling_cost): Pass stmt_info if known to
37512 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37513 cpu_vector_cost field into
37514 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37515 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37516 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37517 splitting of scalar_stmt_cost and vec_stmt_cost.
37518 (thunderx_vector_cost): Likewise.
37519 (cortexa57_vector_cost): LIkewise.
37520 (exynosm1_vector_cost): Likewise.
37521 (xgene1_vector_cost): Likewise.
37522 (thunderx2t99_vector_cost): Improve after the splitting of the two
37524 (aarch64_builtin_vectorization_cost): Update for the splitting of
37525 scalar_stmt_cost and vec_stmt_cost.
37527 2017-02-01 Torvald Riegel <triegel@redhat.com>
37528 Richard Henderson <rth@redhat.com>
37530 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37531 conditional on existance of a fast atomic load.
37532 * optabs-query.c (can_atomic_load_p): New function.
37533 * optabs-query.h (can_atomic_load_p): Declare it.
37534 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37535 no fast atomic load is available for the particular size of access.
37536 (expand_atomic_compare_and_swap): Likewise.
37537 (expand_atomic_load): Likewise.
37538 (expand_atomic_store): Likewise.
37539 (expand_atomic_fetch_op): Likewise.
37540 * testsuite/lib/target-supports.exp
37541 (check_effective_target_sync_int_128): Remove x86 because it provides
37542 no fast atomic load.
37543 (check_effective_target_sync_int_128_runtime): Likewise.
37545 2017-02-01 Richard Biener <rguenther@suse.de>
37547 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37548 (graphite_transform_loops): Provide opt-info for optimized nests.
37549 * tree-parloop.c (parallelize_loops): Provide opt-info for
37550 parallelized loops.
37552 2017-02-01 Richard Biener <rguenther@suse.de>
37554 PR middle-end/79315
37555 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37556 was not set before.
37558 2017-02-01 Richard Biener <rguenther@suse.de>
37560 PR tree-optimization/71824
37561 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37562 Verify the loops are valid in the merged SESE region.
37563 (scop_detection::can_represent_loop_1): Check analyzing the
37564 evolution of the number of iterations in the region succeeds.
37566 2017-01-31 Ian Lance Taylor <iant@golang.org>
37568 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37569 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37571 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37573 PR preprocessor/79210
37574 * input.c (get_substring_ranges_for_loc): Replace line_width
37575 assertion with error-handling.
37577 2017-01-31 Richard Biener <rguenther@suse.de>
37579 PR tree-optimization/77318
37580 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37581 (create_pw_aff_from_tree): Take loop parameter.
37582 (add_condition_to_pbb): Pass loop of the condition to
37583 create_pw_aff_from_tree.
37585 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37587 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37588 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37590 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37594 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37596 (convert_int_to_float128): Likewise.
37597 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37598 (convert_int_to_float128): Likewise.
37599 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37600 (UNSPEC_IEEE128_CONVERT): Likewise.
37601 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37602 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37603 Use local variables for IBM extended format.
37604 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37605 (fix_trunc<mode>si2_fprs): Likewise.
37606 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37607 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37608 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37609 to know that we can now have integers of all sizes in vector
37611 (fix<uns>_<mode>di2_hw): Likewise.
37612 (float<uns>_<mode>si2_hw): Likewise.
37613 (fix_<mode>si2_hw): Likewise.
37614 (fixuns_<mode>si2_hw): Likewise.
37615 (float<uns>_<mode>di2_hw): Likewise.
37616 (float_<mode>di2_hw): Likewise.
37617 (float_<mode>si2_hw): Likewise.
37618 (floatuns_<mode>di2_hw): Likewise.
37619 (floatuns_<mode>si2_hw): Likewise.
37620 (xscvqp<su>wz_<mode>): Delete, no longer used.
37621 (xscvqp<su>dz_<mode>): Likewise.
37622 (xscv<su>dqp_<mode>): Likewise.
37623 (ieee128_mfvsrd_64bit): Likewise.
37624 (ieee128_mfvsrd_32bit): Likewise.
37625 (ieee128_mfvsrwz): Likewise.
37626 (ieee128_mtvsrw): Likewise.
37627 (ieee128_mtvsrd_64bit): Likewise.
37628 (ieee128_mtvsrd_32bit): Likewise.
37630 2017-01-31 Martin Liska <mliska@suse.cz>
37633 * ipa-prop.c (ipa_free_all_node_params): Call release method
37634 instead of ~sumbol_summary to not to trigger double times
37637 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37639 PR tree-optimization/71691
37640 * bitmap.h (class auto_bitmap): New.
37641 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37642 is_maybe_undefined instead of ssa_undefined_value_p.
37644 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37646 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37647 __S390_ARCH_LEVEL__ to __ARCH__.
37649 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37651 PR tree-optimization/79267
37652 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37653 if should_remove_lhs_p is true.
37655 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37658 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37659 (add_alignment_attribute): New.
37660 (base_type_die): Add alignment attribute.
37661 (subrange_type_die): Likewise.
37662 (modified_type_die): Likewise.
37663 (gen_array_type_die): Likewise.
37664 (gen_descr_array_type_die: Likewise.
37665 (gen_enumeration_type_die): Likewise.
37666 (gen_subprogram_die): Likewise.
37667 (gen_variable_die): Likewise.
37668 (gen_field_die): Likewise.
37669 (gen_ptr_to_mbr_type_die): Likewise.
37670 (gen_struct_or_union_type_die): Likewise.
37671 (gen_subroutine_type_die): Likewise.
37672 (gen_typedef_die): Likewise.
37673 (base_type_cmp): Compare alignment attribute.
37675 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37678 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37679 (setb_unsigned) New pattern for setb with CCUNS.
37680 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37681 subfc./subfe sequence to avoid overflow problems. Generate a
37682 shorter sequence with cmpld/setb for power9.
37683 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37684 for generating subfc. instruction.
37685 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37686 now uses this instruction.
37688 2017-01-30 Ian Lance Taylor <iant@google.com>
37691 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37692 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37694 2017-01-30 Martin Sebor <msebor@redhat.com>
37696 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37697 Move constant to the right of a relational operator.
37698 (get_mpfr_format_length, format_character, format_string): Ditto.
37699 (should_warn_p, maybe_warn): Same.
37701 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37703 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37706 * asan.c (get_translation_unit_decl): Remove function.
37707 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37709 2017-01-30 Martin Liska <mliska@suse.cz>
37711 PR gcov-profile/79259
37712 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37713 -fprofile-generate.
37715 2017-01-30 Martin Liska <mliska@suse.cz>
37718 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37719 Initialize variables with NULL value.
37721 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37724 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37726 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37728 2017-01-30 Richard Biener <rguenther@suse.de>
37730 PR tree-optimization/79276
37731 * tree-vrp.c (process_assert_insertions): Properly adjust common
37732 when removing a duplicate.
37734 2017-01-30 Richard Biener <rguenther@suse.de>
37736 PR tree-optimization/79256
37737 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37738 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37740 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37742 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37745 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37746 ("*r<noxa>sbg_<mode>_sll_bitmask")
37747 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37748 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37749 Use contiguous_bitmask_nowrap_operand.
37751 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37754 * config/rs6000/altivec.h (vec_xl): Revise #define.
37755 (vec_xst): Likewise.
37757 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37759 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37761 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37763 PR rtl-optimization/79194
37764 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37765 traps before call to bypass_conditional_jumps.
37767 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37769 PR tree-optimization/71374
37770 * lra-constraints.c (check_conflict_input_operands): New.
37771 (match_reload): Use it.
37773 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37776 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37777 account to calculate conflict_set.
37779 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37781 PR rtl-optimization/78559
37782 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37783 other_insn in combine.
37785 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37787 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37788 uint16_type_node for BT_UINT16.
37790 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37792 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37793 "RTL Tests" to menu.
37794 (GIMPLE Tests): New node.
37795 (RTL Tests): New node.
37797 2017-01-27 Richard Biener <rguenther@suse.de>
37799 PR tree-optimization/79245
37800 * tree-loop-distribution.c (distribute_loop): Apply cost
37801 modeling also to detected patterns.
37803 2017-01-27 Richard Biener <rguenther@suse.de>
37805 PR tree-optimization/71433
37806 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37807 (compare_assert_loc): New function.
37808 (process_assert_insertions): Sort and optimize assert locations
37809 to remove duplicates and push down identical assertions on
37810 edges to their destination block.
37812 2017-01-27 Richard Biener <rguenther@suse.de>
37814 PR tree-optimization/79244
37815 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37816 out SSA names even if abnormal.
37818 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37820 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37821 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37822 instead of MPFR_RNDN.
37824 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37827 * arm.c (arm_option_override): Don't call build_target_option_node
37828 until after doing all option overrides.
37829 (arm_valid_target_attribute_tree): Likewise.
37831 2017-01-27 Martin Liska <mliska@suse.cz>
37833 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37834 for {cd}tors and C++ {cd}tors.
37836 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37838 * config/s390/s390.md ("*setmem_long_and")
37839 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37841 2017-01-26 Martin Sebor <msebor@redhat.com>
37843 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37846 2017-01-26 Martin Sebor <msebor@redhat.com>
37848 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37849 HAVE_DFmode before using XFmode or DFmode.
37850 (parse_directive): Avoid using the z length modifier to avoid
37851 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37853 PR middle-end/78703
37854 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37855 to accept adjustment as an array.
37856 (get_int_range): New function.
37857 (struct directive): Make width and prec arrays.
37858 (directive::set_width, directive::set_precision): Call get_int_range.
37859 (format_integer, format_floating): Handle width and precision ranges.
37860 (format_string, parse_directive): Same.
37862 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37865 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37866 comdat_type_p set, just clone them, but keep the children in the
37870 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37871 which have direct callers with -fvar-tracking-assignments enabled
37873 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37874 inside of type units.
37876 2017-01-26 Martin Sebor <msebor@redhat.com>
37878 PR middle-end/78703
37879 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37881 (struct format_result): Replace number_chars, number_chars_min,
37882 and number_chars_max with a single member of struct result_range.
37884 (format_result::operator+=): Adjust.
37885 (struct fmtresult): Remove bounded. Handle likely and unlikely
37887 (fmtresult::adjust_for_width_or_precision): New function.
37888 (fmtresult:type_max_digits): New function.
37889 (bytes_remaining): Handle likely and unlikely counters.
37890 (min_bytes_remaining): Remove.
37891 (format_percent): Simplify.
37892 (format_integer, format_floating): Set likely and unlikely counters.
37893 (get_string_length, format_character, format_string): Same.
37894 (format_plain, should_warn_p): New function.
37895 (maybe_warn): Call should_warn_p. Update diagnostic messages
37896 and handle those for all directives, including plain strings.
37897 (format_directive): Handle likely and unlikely counters.
37898 Remove unnecessary quoting from diagnostics. Add an informational
37900 (add_bytes): Remove.
37901 (pass_sprintf_length::compute_format_length): Simplify.
37902 (try_substitute_return_value): Handle likely and unlikely counters.
37904 2017-01-26 Carl Love <cel@us.ibm.com>
37906 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37907 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37909 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37912 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37913 endianess for subregs into account.
37914 * lra-constraints.c (lra_constraints): Do risky transformations
37915 always on the first iteration.
37916 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37917 last_call_used_reg_set.
37918 (process_bb_lives): Define and use last_call_used_reg_set.
37919 * lra.c (lra): Always continue after lra_constraints on the first
37922 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37924 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37926 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37928 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37930 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37931 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37932 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37933 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37934 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37935 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37936 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37937 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37938 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37940 2017-01-26 Marek Polacek <polacek@redhat.com>
37943 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37944 for the third operand.
37946 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37948 PR middle-end/79236
37949 * omp-low.c (struct omp_context): Add simt_stmt field.
37950 (scan_omp_for): Return omp_context *.
37951 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37952 context to the _simt_ SIMD stmt.
37953 (lower_omp_for): For combined SIMD with sibling _simt_
37954 SIMD, make sure to use the same decls in _looptemp_
37955 clauses as in the sibling.
37957 2017-01-26 David Sherwood <david.sherwood@arm.com>
37959 PR middle-end/79212
37960 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37963 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37966 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37967 emit fld b; fld a; if possible.
37969 * brig-builtins.def: Update copyright years.
37970 * config/arm/arm_acle_builtins.def: Update copyright years.
37972 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37975 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37976 constraint instead of o for the stxsd instruction.
37978 2017-01-25 Carl Love <cel@us.ibm.com>
37980 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37981 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37983 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37985 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37987 2017-01-25 Richard Biener <rguenther@suse.de>
37989 PR tree-optimization/69264
37990 * target.def (vector_alignment_reachable): Improve documentation.
37991 * doc/tm.texi: Regenerate.
37992 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37994 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37995 earlier changes with respect to TYPE_USER_ALIGN.
37996 (vector_alignment_reachable_p): Likewise. Improve dumping.
37998 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38001 * config/arm/arm.md (xordi3): Force constant operand into a register
38004 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38006 * doc/invoke.texi (-fstore-merging): Correct default optimization
38007 levels at which it is enabled.
38008 (-O): Move -fstore-merging from list to...
38011 2017-01-25 Richard Biener <rguenther@suse.de>
38014 * omp-expand.c: Include debug.h.
38015 (expand_omp_taskreg): Make sure to generate early debug before
38016 outlining anything from a function.
38017 (expand_omp_target): Likewise.
38018 (grid_expand_target_grid_body): Likewise.
38020 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38023 * asan.c (get_translation_unit_decl): New function.
38024 (asan_add_global): Extract modules file name from globals
38025 TRANSLATION_UNIT_DECL name.
38027 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38030 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38031 for long calls with APCS frame and VFP.
38033 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38035 * cfg.c (original_copy_tables_initialized_p): New function.
38036 * cfg.h (original_copy_tables_initialized_p): New decl.
38037 * cfgrtl.c (relink_block_chain): Guard the call to
38038 free_original_copy_tables with a call to
38039 original_copy_tables_initialized_p.
38040 * cgraph.h (symtab_node::native_rtl_p): New decl.
38041 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38042 (symtab_node::needed_p): Don't assert for early assembly output
38043 for __RTL functions.
38044 (cgraph_node::finalize_function): Set "force_output" for __RTL
38046 (cgraph_node::analyze): Bail out early for __RTL functions.
38047 (analyze_functions): Update assertion to support __RTL functions.
38048 (cgraph_node::expand): Bail out early for __RTL functions.
38049 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38051 * function.h (struct function): Update comment for field
38053 * gimple-expr.c: Include "tree-pass.h".
38054 (gimple_has_body_p): Return false for __RTL functions.
38055 * Makefile.in (OBJS): Add run-rtl-passes.o.
38056 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38058 (gcc::pass_manager::get_clean_slate): New accessor.
38059 * passes.c: Include "insn-addr.h".
38060 (should_skip_pass_p): Add logging. Update logic for running
38061 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38062 property-provider override so it is only done for gimple passes.
38064 (skip_pass): New function.
38065 (execute_one_pass): Call skip_pass when skipping passes.
38066 * read-md.c (md_reader::read_char): Support filtering
38067 the input to a subset of line numbers.
38068 (md_reader::md_reader): Initialize fields
38069 m_first_line and m_last_line.
38070 (md_reader::read_file_fragment): New function.
38071 * read-md.h (md_reader::read_file_fragment): New decl.
38072 (md_reader::m_first_line): New field.
38073 (md_reader::m_last_line): New field.
38074 * read-rtl-function.c (function_reader::create_function): Only
38075 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38076 curr_properties. Set DECL_INITIAL to a dummy block.
38077 (read_rtl_function_body_from_file_range): New function.
38078 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38080 * run-rtl-passes.c: New file.
38081 * run-rtl-passes.h: New file.
38083 2017-01-24 Jeff Law <law@redhat.com>
38085 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38088 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38090 PR tree-optimization/79159
38091 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38092 (record_nonwrapping_iv): Improve boundary using above function if no
38093 value range information.
38095 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38096 Martin Jambor <mjambor@suse.cz>
38098 * brig-builtins.def: New file.
38099 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38100 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38101 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38102 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38103 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38104 * builtin-types.def (BT_INT8): New.
38105 (BT_INT16): Likewise.
38106 (BT_UINT8): Likewise.
38107 (BT_UINT16): Likewise.
38108 (BT_FN_ULONG): Likewise.
38109 (BT_FN_UINT_INT): Likewise.
38110 (BT_FN_UINT_ULONG): Likewise.
38111 (BT_FN_UINT_LONG): Likewise.
38112 (BT_FN_UINT_PTR): Likewise.
38113 (BT_FN_ULONG_PTR): Likewise.
38114 (BT_FN_INT8_FLOAT): Likewise.
38115 (BT_FN_INT16_FLOAT): Likewise.
38116 (BT_FN_UINT32_FLOAT): Likewise.
38117 (BT_FN_UINT16_FLOAT): Likewise.
38118 (BT_FN_UINT8_FLOAT): Likewise.
38119 (BT_FN_UINT64_FLOAT): Likewise.
38120 (BT_FN_UINT16_UINT32): Likewise.
38121 (BT_FN_UINT32_UINT16): Likewise.
38122 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38123 (BT_FN_INT_PTR_INT): Likewise.
38124 (BT_FN_UINT_PTR_UINT): Likewise.
38125 (BT_FN_LONG_PTR_LONG): Likewise.
38126 (BT_FN_ULONG_PTR_ULONG): Likewise.
38127 (BT_FN_VOID_UINT64_UINT64): Likewise.
38128 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38129 (BT_FN_INT8_INT8_INT8): Likewise.
38130 (BT_FN_INT16_INT16_INT16): Likewise.
38131 (BT_FN_INT_INT_INT): Likewise.
38132 (BT_FN_UINT_FLOAT_UINT): Likewise.
38133 (BT_FN_FLOAT_UINT_UINT): Likewise.
38134 (BT_FN_ULONG_UINT_UINT): Likewise.
38135 (BT_FN_ULONG_UINT_PTR): Likewise.
38136 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38137 (BT_FN_UINT_UINT_UINT): Likewise.
38138 (BT_FN_VOID_UINT_PTR): Likewise.
38139 (BT_FN_UINT_UINT_PTR: Likewise.
38140 (BT_FN_UINT32_UINT64_PTR): Likewise.
38141 (BT_FN_INT_INT_UINT_UINT): Likewise.
38142 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38143 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38144 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38145 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38146 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38147 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38148 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38149 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38150 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38151 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38152 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38153 * doc/frontends.texi: List BRIG FE.
38154 * doc/install.texi (Testing): Add BRIG tesring requirements.
38155 * doc/invoke.texi (Overall Options): Mention BRIG.
38156 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38158 2017-01-24 Richard Biener <rguenther@suse.de>
38160 PR translation/79208
38161 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38163 2017-01-24 Martin Jambor <mjambor@suse.cz>
38166 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38167 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38168 and known_contexts.
38170 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38172 PR middle-end/79123
38173 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38174 casts from signed to unsigned really don't have a range.
38176 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38178 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38179 GMP_RNDx for compatiblity.
38181 2017-01-24 Martin Liska <mliska@suse.cz>
38184 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38185 that would prevent us to call alloca with -1 as argument.
38187 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38189 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38190 Avoid -Wformat-security warning.
38192 2017-01-23 Andrew Pinski <apinski@cavium.com>
38194 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38197 2017-01-23 Martin Sebor <msebor@redhat.com>
38199 PR middle-end/78703
38200 * gimple-ssa-sprintf.c (warn_level): New global.
38201 (format_integer): Use it here and throughout the rest of the file.
38202 Use the same switch to compute sign as base.
38203 (maybe_warn): New function.
38204 (format_directive): Factor out warnings into maybe_warn.
38205 Add debugging output. Use warn_level.
38206 (add_bytes): Use warn_level.
38207 (pass_sprintf_length::compute_format_length): Add debugging output.
38208 (try_substitute_return_value): Same.
38209 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38211 PR middle-end/78703
38212 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38213 (struct fmtresult, format_integer, format_floating): Adjust.
38214 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38215 (get_string_length, format_string,format_directive): Same.
38216 (pass_sprintf_length::compute_format_length): Same.
38217 (try_substitute_return_value): Simplify slightly.
38219 PR middle-end/78703
38220 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38221 (fmtresult::operator+=): Outlined.
38222 (struct fmtresult): Add ctors.
38223 (struct conversion_spec): Rename...
38224 (struct directive): ...to this. Add and remove data members.
38225 (directive::set_width, directive::set_precision): New functions.
38226 (format_percent): Use fmtresult ctor.
38227 (get_width_and_precision): Remove.
38228 (format_integer): Make naming changes. Avoid computing width and
38230 (format_floating): Same. Adjust indentation.
38231 (format_character, format_none): New functions.
38232 (format_string): Moved character handling to format_character.
38233 (format_directive): Remove arguments, change return type.
38234 (parse_directive): New function.
38235 (pass_sprintf_length::compute_format_length): Move directive
38236 parsing to parse_directive.
38238 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38240 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38241 (assign_assembler_name_if_needed): ... this.
38242 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38243 (assign_assembler_name_if_needed): ... this.
38244 (free_lang_data_in_cgraph): Adjust callers.
38245 * cgraphunit.c (cgraph_node::analyze): Likewise.
38246 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38249 2017-01-23 Richard Biener <rguenther@suse.de>
38251 PR tree-optimization/79088
38252 PR tree-optimization/79188
38253 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38254 resetting loop bounds after last path deletion. Reset loop
38255 bounds of the target loop, make code match the comments.
38256 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38257 Make sure loops need no fixups.
38259 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38261 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38262 exponent support with double type for first argument.
38263 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38264 type returned by __builtin_vec_extract_sig,
38265 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38266 functions from "vector int" to "vector unsigned int" or from
38267 "vector long long int" to "vector unsigned long long int".
38268 Changed type returned by __builtin_vec_extract_exp,
38269 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38270 functions from "vector int" to "vector unsigned int" or from
38271 "vector long long int" to "vector unsigned long long int".
38272 Changed return type of __builtin_vec_test_data_class,
38273 __builtin_vec_test_data_class_sp, and
38274 __builtin_vec_test_data_class_dp from "vector int" to
38275 "vector bool int" or from "vector long long int" to "vector bool
38276 long long int" and changed second argument type from "unsigned
38277 int" to "int". Added new overloaded function forms "vector float
38278 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38279 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38280 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38281 double, vector unsigned long long int)" and "vector double
38282 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38283 long int)". Changed return type of
38284 __builtin_scalar_test_data_class and
38285 __builtin_scalar_test_data_class_sp and
38286 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38287 int" and changed second argument from "unsigned int" to "int".
38288 Changed type returned by __builtin_scalar_test_neg,
38289 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38290 from "int" to "bool int". Added new overloaded function form
38291 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38292 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38293 exponent double-precision with floating point first argument.
38294 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38295 documentation of scalar_test_data_class, scalar_test_neg,
38296 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38297 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38298 vec_test_data_class built-in functions to reflect refinements in
38299 their type signatures.
38301 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38303 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38305 (aarch64_elf_asm_destructor): Likewise.
38307 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38309 PR rtl-optimization/78634
38310 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38311 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38312 * ifcvt.c (noce_try_cmove): Add missing cost check.
38314 PR rtl-optimization/71724
38315 * combine.c (if_then_else_cond): Look for situations where it is
38316 beneficial to undo the work of one of the recursive calls.
38318 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38320 PR tree-optimization/70754
38321 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38322 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38323 combined stmt before it if not NULL.
38324 (combine_chains): Process refs reversely and compute dominance point
38327 2017-01-23 Martin Liska <mliska@suse.cz>
38329 PR tree-optimization/79196
38330 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38331 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38333 (strlen_optimize_stmt): Call the renamed function.
38335 2017-01-23 Michael Matz <matz@suse.de>
38337 PR tree-optimization/78384
38338 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38340 2017-01-23 Richard Biener <rguenther@suse.de>
38342 PR tree-optimization/79186
38343 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38344 both incoming edges before moving an assert.
38346 2017-01-23 Martin Jambor <mjambor@suse.cz>
38348 * ipa-prop.c (load_from_param_1): Removed.
38349 (load_from_unmodified_param): Bits from load_from_param_1 put back
38351 (load_from_param): Removed.
38352 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38353 with stmt. Reverted back to use of load_from_unmodified_param.
38355 2017-01-23 Martin Jambor <mjambor@suse.cz>
38358 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38359 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38360 field a pointer to garbage collected vector, mark lattices and
38361 ipcp_orig_node with GTY((skip)).
38362 (ipa_get_param_count): Adjust to descriptors being a pointer.
38363 (ipa_get_param): Likewise.
38364 (ipa_get_type): Likewise.
38365 (ipa_get_param_move_cost): Likewise.
38366 (ipa_set_param_used): Likewise.
38367 (ipa_get_controlled_uses): Likewise.
38368 (ipa_set_controlled_uses): Likewise.
38369 (ipa_is_param_used): Likewise.
38370 (ipa_node_params_t): Move into garbage collector. New methods insert
38372 (ipa_node_params_sum): Annotate wth GTY(()).
38373 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38375 (ipa_load_from_parm_agg): Adjust declaration.
38376 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38377 * ipa-profile.c (ipa_profile): Likewise.
38378 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38379 (ipa_populate_param_decls): Make descriptors parameter garbage
38381 (ipa_dump_param): Adjust to descriptors being a pointer.
38382 (ipa_alloc_node_params): Likewise.
38383 (ipa_initialize_node_params): Likewise.
38384 (load_from_param_1): Make descriptors parameter garbage collected.
38385 (load_from_unmodified_param): Likewise.
38386 (load_from_param): Likewise.
38387 (ipa_load_from_parm_agg): Likewise.
38388 (ipa_node_params::~ipa_node_params): Removed.
38389 (ipa_free_all_node_params): Remove call to delete operator.
38390 (ipa_node_params_t::insert): New.
38391 (ipa_node_params_t::remove): Likewise.
38392 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38393 copy known_csts and known_contexts vectors.
38394 (ipa_read_node_info): Adjust to descriptors being a pointer.
38395 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38397 (ipcp_transform_function): Make descriptors variable garbage
38400 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38402 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38403 * config/i386/avx512dqintrin.h: Ditto.
38404 * config/i386/avx512fintrin.h: Ditto.
38405 * config/i386/i386.c: Handle new builtins.
38406 * config/i386/i386-builtin.def: Add new builtins.
38407 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38408 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38410 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38411 Martin Liska <mliska@suse.cz>
38413 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38414 * asan.c (asan_expand_poison_ifn): Support stores and use
38415 appropriate ASAN report function.
38416 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38417 * internal-fn.def (ASAN_POISON_USE): Declare.
38418 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38419 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38420 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38421 ASAN_POISON calls w/o LHS.
38422 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38423 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38424 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38425 * gimplify.c (asan_poison_variables): Add attribute
38426 use_after_scope_memory to variables that really needs to live
38428 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38429 having the attribute.
38431 2017-01-23 Martin Liska <mliska@suse.cz>
38433 * asan.c (create_asan_shadow_var): New function.
38434 (asan_expand_poison_ifn): Likewise.
38435 * asan.h (asan_expand_poison_ifn): New declaration.
38436 * internal-fn.c (expand_ASAN_POISON): Likewise.
38437 * internal-fn.def (ASAN_POISON): New builtin.
38438 * sanopt.c (pass_sanopt::execute): Expand
38439 asan_expand_poison_ifn.
38440 * tree-inline.c (copy_decl_for_dup_finish): Make function
38442 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38443 * tree-ssa.c (is_asan_mark_p): New function.
38444 (execute_update_addresses_taken): Rewrite local variables
38445 (identified just by use-after-scope as addressable) into SSA.
38447 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38449 * doc/install.texi (Specific): opensource.apple.com uses https
38450 now. Remove trailing slash.
38452 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38454 * README.Portability: Remove note on an Irix compatibility issue.
38456 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38458 * gcov.c (INCLUDE_ALGORITHM): Define.
38459 (INCLUDE_VECTOR): Define.
38460 No longer include <vector> and <algorithm> directly.
38462 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38464 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38466 * doc/invoke.texi (Code Gen Options): Ditto.
38468 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38471 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38473 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38475 rtl-optimization/79125
38476 * cprop.c (local_cprop_pass): Handle cases where we make an
38477 unconditional trap.
38479 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38483 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38484 read from, for big endian.
38486 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38488 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38489 register pauth builtins for LP64 only.
38491 2017-01-20 Marek Polacek <polacek@redhat.com>
38494 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38497 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38499 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38501 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38502 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38503 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38505 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38508 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38509 in big-endian mode when they are not a single duplicated value.
38511 2017-01-20 Richard Biener <rguenther@suse.de>
38513 * BASE-VER: Bump to 7.0.1.
38515 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38517 * omp-low.c (omplow_simd_context): New struct. Use it...
38518 (lower_rec_simd_input_clauses): ...here and...
38519 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38520 references to idx, lane, max_vf, is_simt.
38522 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38524 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38527 2017-01-20 Martin Jambor <mjambor@suse.cz>
38529 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38530 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38532 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38533 (GTFILES): Rename hsa.c to hsa-common.c.
38534 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38535 * hsa-dump.c: Likewise.
38536 * hsa-gen.c: Likewise.
38537 * hsa-regalloc.c: Likewise.
38538 * ipa-hsa.c: Likewise.
38539 * omp-expand.c: Likewise.
38540 * omp-low.c: Likewise.
38541 * toplev.c: Likewise.
38543 2017-01-20 Marek Polacek <polacek@redhat.com>
38546 * doc/invoke.texi: Document -Wduplicated-branches.
38547 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38548 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38549 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38550 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38551 return 0 only when not OEP_LEXICOGRAPHIC.
38552 (fold_build_cleanup_point_expr): Use the expression
38553 location when building CLEANUP_POINT_EXPR.
38554 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38555 * tree.c (add_expr): Handle error_mark_node.
38557 2017-01-20 Martin Liska <mliska@suse.cz>
38560 * tree-profile.c (init_ic_make_global_vars): Do not call
38562 (gimple_init_gcov_profiler): Likewise.
38564 2017-01-20 Martin Liska <mliska@suse.cz>
38567 * cgraph.h (maybe_create_reference): Remove argument and
38569 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38571 * ipa-cp.c (create_specialized_node): Likewise.
38572 * symtab.c (symtab_node::maybe_create_reference): Handle
38573 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38575 2017-01-20 Martin Liska <mliska@suse.cz>
38577 * read-rtl-function.c (function_reader::create_function): Use
38578 build_decl instread of build_decl_stat.
38580 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38582 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38583 * config/i386/avx512dqintrin.h: Ditto.
38584 * config/i386/avx512fintrin.h: Ditto.
38585 * config/i386/i386-builtin-types.def: Add new types.
38586 * config/i386/i386.c: Handle new types.
38587 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38588 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38589 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38590 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38591 (__builtin_ia32_kshiftridi): New.
38592 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38594 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38598 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38599 define to rs6000_init_stack_protect_guard.
38600 (rs6000_init_stack_protect_guard): New function.
38602 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38603 Yunqiang Su <yunqiang.su@imgtec.com>
38605 * config.gcc (supported_defaults): Add madd4.
38606 (with_madd4): Add validation.
38607 (all_defaults): Add madd4.
38608 * config/mips/mips.opt (mmadd4): New option.
38609 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38611 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38613 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38614 (ISA_HAS_FUSED_MADD4): Likewise.
38615 * doc/invoke.texi (-mmadd4): Document the new option.
38616 * doc/install.texi (--with-madd4): Document the new option.
38618 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38620 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38621 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38622 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38623 (aarch64_init_pauth_hint_builtins): New.
38624 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38625 (aarch64_expand_builtin): Expand new builtins.
38627 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38629 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38630 * combine-stack-adj.c (no_unhandled_cfa): Handle
38631 REG_CFA_TOGGLE_RA_MANGLE.
38632 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38633 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38634 info for return address signing.
38635 (aarch64_expand_epilogue): Likewise.
38637 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38639 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38640 * config/aarch64/aarch64-protos.h
38641 (aarch64_return_address_signing_enabled): New declaration.
38642 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38644 (aarch64_expand_prologue): Sign return address before it's pushed onto
38646 (aarch64_expand_epilogue): Authenticate return address fetched from
38648 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38649 (aarch64_attributes): New function attributes for "sign-return-address".
38650 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38651 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38652 ("*do_return"): Generate combined instructions according to key index.
38653 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38654 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38656 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38657 * config/aarch64/aarch64.opt (msign-return-address=): New.
38658 * doc/extend.texi (AArch64 Function Attributes): Documents
38659 "sign-return-address=".
38660 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38662 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38664 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38665 overall option summary.
38667 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38669 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38670 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38671 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38672 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38674 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38676 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38677 -mpower9-minmax by default for -mcpu=power9.
38678 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38679 128-bit floating point.
38681 2017-01-20 Alan Modra <amodra@gmail.com>
38683 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38684 optimizing for size.
38686 2017-01-20 Alan Modra <amodra@gmail.com>
38689 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38690 for strcmp and strncmp from corresponding builtin decl.
38692 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38694 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38695 instead of i386/rtems-64.h.
38696 * config/i386/rtems-64.h: Remove.
38698 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38702 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38704 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38706 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38708 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38709 Change int to HOST_WIDE_INT.
38710 * config/aarch64/aarch64-protos.h
38711 (aarch64_simd_gen_const_vector_dup): Likewise.
38712 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38714 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38716 * langhooks-def.h (lhd_type_for_size): New decl.
38717 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38718 * langhooks.c (lhd_type_for_size): New function, taken from
38721 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38723 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38724 define_bypass for CR latency.
38725 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38726 (power9-alu2): Add define_bypass for CR latency.
38728 (power9-mul): Update insn latency.
38729 (power9-mul-compare): Update insn latency, bypass latency and remove
38732 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38734 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38736 * config/aarch64/aarch64.md
38737 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38738 aarch64_nopcrelative_literal_loads.
38739 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38741 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38743 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38744 TARGET_LOONGSON_3A.
38745 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38747 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38750 * config.gcc (supported_defaults): Add lxc1-sxc1.
38751 (with_lxc1_sxc1): Add validation.
38752 (all_defaults): Add lxc1-sxc1.
38753 * config/mips/mips.opt (mlxc1-sxc1): New option.
38754 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38756 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38757 __mips_no_lxc1_sxc1.
38758 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38759 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38760 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38762 2017-01-19 Richard Biener <rguenther@suse.de>
38764 PR tree-optimization/72488
38765 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38766 sure to restore SSA info.
38767 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38769 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38771 PR rtl-optimization/79121
38772 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38773 of the inner type when shifting an extended value.
38775 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38778 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38779 interposable aliases.
38781 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38784 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38785 Use the evmergelohi instruction.
38786 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38787 (mov_sitf_e500_subreg8_2_be): Likewise.
38788 (mov_sitf_e500_subreg12_2_le): Likewise.
38789 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38790 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38791 (mov_sitf_e500_subreg8_2_le): Likewise.
38792 (mov_sitf_e500_subreg12_2_be): Likewise.
38794 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38796 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38797 attribute from vecsimple to vecperm.
38798 (altivec_vbpermq2): Likewise.
38800 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38803 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38805 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38806 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38807 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38808 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38809 case where N arg is SIZE_MAX.
38810 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38811 (cmpstrsi): Add pattern.
38813 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38815 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38816 __builtin_vec_revb builtins.
38817 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38818 built-in functions to support generation of the ISA 3.0 XXBR<x>
38819 vector byte reverse instructions.
38820 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38821 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38822 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38823 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38824 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38825 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38826 (P9V_BUILTIN_VEC_REVB): Likewise.
38827 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38828 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38829 (p9_xxbrq_v16qi): Likewise.
38830 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38831 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38832 (p9_xxbrh_v8hi): Likewise.
38833 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38834 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38835 vec_revb built-in functions.
38837 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38839 PR rtl-optimization/78952
38840 * config/i386/i386.md (any_extract): New code iterator.
38841 (*insvqi_2): Use any_extract for source operand.
38842 (*insvqi_3): Use any_shiftrt for source operand.
38844 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38846 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38848 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38850 2017-01-18 Matthias Klose <doko@ubuntu.com>
38852 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38854 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38856 * config/rs6000/altivec.h (vec_bperm): Change #define.
38857 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38858 (altivec_vbpermq2): New define_insn.
38859 (altivec_vbpermd): Likewise.
38860 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38861 function interface.
38862 (VBPERMD): Likewise.
38863 (VBPERM): New polymorphic function interface.
38864 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38865 Add entries for P9V_BUILTIN_VEC_VBPERM.
38866 * doc/extend.texi: Add interfaces for vec_bperm.
38868 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38870 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38871 first letter of error messages.
38872 (s390_resolve_overloaded_builtin): Likewise.
38873 * config/s390/s390.c (s390_expand_builtin): Likewise.
38874 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38875 (s390_valid_target_attribute_inner_p): Likewise.
38876 * config/s390/s390.md ("tabort"): Likewise.
38878 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38880 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38881 (ISA_AVOID_DIV_HILO): New macro.
38882 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38883 (ISA_HAS_DDIV): Likewise.
38885 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38887 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38889 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38891 * doc/invoke.texi (fabi-version): Spelling fix.
38893 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38896 * doc/invoke.texi (fabi-version): Mention mangling fix for
38899 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38902 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38904 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38907 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38910 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38912 (stack_protect_test): Ditto.
38913 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38914 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38916 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38917 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38918 -mstack-protector-guard-offset=.
38919 (RS/6000 and PowerPC Options): Ditto.
38921 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38923 * config/i386/i386.h (MASK_CLASS_P): New define.
38924 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38925 there are no registers from different register sets also when
38926 mask registers are used. Update function comment.
38927 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38928 to (*k/*r) and (*k/*km) alternatives.
38930 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38932 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38933 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38934 (EH_RETURN_HANDLER_RTX): New define.
38935 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38936 Force frame pointer in EH return functions.
38937 (aarch64_expand_epilogue): Add barrier for eh_return.
38938 (aarch64_final_eh_return_addr): Remove.
38939 (aarch64_eh_return_handler_rtx): New function.
38940 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38942 (aarch64_eh_return_handler_rtx): New prototype.
38944 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38946 * config/rs6000/altivec.h (vec_rlmi): New #define.
38947 (vec_vrlnm): Likewise.
38948 (vec_rlnm): Likewise.
38949 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38950 (UNSPEC_VRLNM): Likewise.
38951 (VIlong): New mode iterator.
38952 (altivec_vrl<VI_char>mi): New define_insn.
38953 (altivec_vrl<VI_char>nm): Likewise.
38954 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38956 (VRLDNM): Likewise.
38957 (RLNM): New polymorphic function entry.
38958 (VRLWMI): New monomorphic function entry.
38959 (VRLDMI): Likewise.
38960 (RLMI): New polymorphic function entry.
38961 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38962 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38963 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38966 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38969 * dwarf2out.c (field_byte_offset): Restore the
38970 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38971 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38972 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38975 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38978 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38980 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38983 * dwarf2out.c (add_data_member_location_attribute): For constant
38984 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38985 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38986 DW_AT_byte_size attributes.
38988 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38990 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38991 after forcing to constant memory when the code model is medium.
38993 2017-01-17 Julia Koval <julia.koval@intel.com>
38996 * config/i386/avx512fintrin.h
38997 (_mm512_i32gather_ps): Change __addr type to void const*.
38998 (_mm512_mask_i32gather_ps): Ditto.
38999 (_mm512_i32gather_pd): Ditto.
39000 (_mm512_mask_i32gather_pd): Ditto.
39001 (_mm512_i64gather_ps): Ditto.
39002 (_mm512_mask_i64gather_ps): Ditto.
39003 (_mm512_i64gather_pd): Ditto.
39004 (_mm512_mask_i64gather_pd): Ditto.
39005 (_mm512_i32gather_epi32): Ditto.
39006 (_mm512_mask_i32gather_epi32): Ditto.
39007 (_mm512_i32gather_epi64): Ditto.
39008 (_mm512_mask_i32gather_epi64): Ditto.
39009 (_mm512_i64gather_epi32): Ditto.
39010 (_mm512_mask_i64gather_epi32): Ditto.
39011 (_mm512_i64gather_epi64): Ditto.
39012 (_mm512_mask_i64gather_epi64): Ditto.
39013 (_mm512_i32scatter_ps): Change __addr type to void*.
39014 (_mm512_mask_i32scatter_ps): Ditto.
39015 (_mm512_i32scatter_pd): Ditto.
39016 (_mm512_mask_i32scatter_pd): Ditto.
39017 (_mm512_i64scatter_ps): Ditto.
39018 (_mm512_mask_i64scatter_ps): Ditto.
39019 (_mm512_i64scatter_pd): Ditto.
39020 (_mm512_mask_i64scatter_pd): Ditto.
39021 (_mm512_i32scatter_epi32): Ditto.
39022 (_mm512_mask_i32scatter_epi32): Ditto.
39023 (_mm512_i32scatter_epi64): Ditto.
39024 (_mm512_mask_i32scatter_epi64): Ditto.
39025 (_mm512_i64scatter_epi32): Ditto.
39026 (_mm512_mask_i64scatter_epi32): Ditto.
39027 (_mm512_i64scatter_epi64): Ditto.
39028 (_mm512_mask_i64scatter_epi64): Ditto.
39029 * config/i386/avx512pfintrin.h
39030 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39031 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39032 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39033 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39034 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39035 (_mm512_prefetch_i32scatter_ps): Ditto.
39036 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39037 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39038 (_mm512_prefetch_i64scatter_pd): Ditto.
39039 (_mm512_prefetch_i64scatter_ps): Ditto.
39040 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39041 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39042 * config/i386/avx512vlintrin.h
39043 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39044 (_mm_mmask_i32gather_ps): Ditto.
39045 (_mm256_mmask_i32gather_pd): Ditto.
39046 (_mm_mmask_i32gather_pd): Ditto.
39047 (_mm256_mmask_i64gather_ps): Ditto.
39048 (_mm_mmask_i64gather_ps): Ditto.
39049 (_mm256_mmask_i64gather_pd): Ditto.
39050 (_mm_mmask_i64gather_pd): Ditto.
39051 (_mm256_mmask_i32gather_epi32): Ditto.
39052 (_mm_mmask_i32gather_epi32): Ditto.
39053 (_mm256_mmask_i32gather_epi64): Ditto.
39054 (_mm_mmask_i32gather_epi64): Ditto.
39055 (_mm256_mmask_i64gather_epi32): Ditto.
39056 (_mm_mmask_i64gather_epi32): Ditto.
39057 (_mm256_mmask_i64gather_epi64): Ditto.
39058 (_mm_mmask_i64gather_epi64): Ditto.
39059 (_mm256_i32scatter_ps): Change __addr type to void*.
39060 (_mm256_mask_i32scatter_ps): Ditto.
39061 (_mm_i32scatter_ps): Ditto.
39062 (_mm_mask_i32scatter_ps): Ditto.
39063 (_mm256_i32scatter_pd): Ditto.
39064 (_mm256_mask_i32scatter_pd): Ditto.
39065 (_mm_i32scatter_pd): Ditto.
39066 (_mm_mask_i32scatter_pd): Ditto.
39067 (_mm256_i64scatter_ps): Ditto.
39068 (_mm256_mask_i64scatter_ps): Ditto.
39069 (_mm_i64scatter_ps): Ditto.
39070 (_mm_mask_i64scatter_ps): Ditto.
39071 (_mm256_i64scatter_pd): Ditto.
39072 (_mm256_mask_i64scatter_pd): Ditto.
39073 (_mm_i64scatter_pd): Ditto.
39074 (_mm_mask_i64scatter_pd): Ditto.
39075 (_mm256_i32scatter_epi32): Ditto.
39076 (_mm256_mask_i32scatter_epi32): Ditto.
39077 (_mm_i32scatter_epi32): Ditto.
39078 (_mm_mask_i32scatter_epi32): Ditto.
39079 (_mm256_i32scatter_epi64): Ditto.
39080 (_mm256_mask_i32scatter_epi64): Ditto.
39081 (_mm_i32scatter_epi64): Ditto.
39082 (_mm_mask_i32scatter_epi64): Ditto.
39083 (_mm256_i64scatter_epi32): Ditto.
39084 (_mm256_mask_i64scatter_epi32): Ditto.
39085 (_mm_i64scatter_epi32): Ditto.
39086 (_mm_mask_i64scatter_epi32): Ditto.
39087 (_mm256_i64scatter_epi64): Ditto.
39088 (_mm256_mask_i64scatter_epi64): Ditto.
39089 (_mm_i64scatter_epi64): Ditto.
39090 (_mm_mask_i64scatter_epi64): Ditto.
39091 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39092 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39093 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39094 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39095 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39096 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39097 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39098 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39099 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39100 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39101 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39102 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39103 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39104 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39105 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39106 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39107 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39108 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39109 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39110 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39111 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39112 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39113 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39114 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39115 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39116 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39117 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39118 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39119 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39120 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39121 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39122 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39123 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39124 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39125 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39126 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39127 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39128 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39129 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39130 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39131 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39132 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39133 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39134 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39135 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39136 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39137 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39138 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39139 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39140 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39141 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39142 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39143 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39144 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39145 definitions accordingly.
39147 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39148 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39151 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39154 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39157 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39158 hard regs for inner regno.
39160 2017-01-17 Martin Liska <mliska@suse.cz>
39163 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39164 assumption and add comment.
39166 2017-01-17 Nathan Sidwell <nathan@acm.org>
39168 * ipa-visibility.c (localize_node): New function, broken out of ...
39169 (function_and_variable_visibility): ... here. Call it.
39171 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39173 PR middle-end/77445
39174 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39175 correctly set frequency of oudgoing edge.
39176 (duplicate_thread_path): Fix profile updating.
39178 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39181 * configure.ac: Add GCC_BASE_VER.
39182 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39183 version from BASE-VER file.
39184 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39185 (gcc.o): Depend on $(BASEVER).
39186 * common.opt (dumpfullversion): New option.
39187 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39188 * doc/invoke.texi: Document -dumpfullversion.
39189 * doc/install.texi: Document --with-gcc-major-version-only.
39190 * configure: Regenerated.
39192 2017-01-17 Richard Biener <rguenther@suse.de>
39194 PR tree-optimization/71433
39195 * tree-vrp.c (register_new_assert_for): Merge same asserts
39196 on all incoming edges.
39197 (process_assert_insertions_for): Handle insertions at the
39200 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39202 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39203 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39205 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39208 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39211 2017-01-17 Alan Modra <amodra@gmail.com>
39214 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39215 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39216 symbolic stack limit when pic.
39218 2017-01-16 Martin Sebor <msebor@redhat.com>
39220 PR tree-optimization/78608
39221 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39223 2017-01-16 Jeff Law <law@redhat.com>
39226 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39227 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39228 for several include directories that may be relative to sysroot.
39229 * config/i386/x-mingw32 (gplus_includedir): Define.
39230 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39231 (native_system_includedir): Likewise.
39232 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39233 override if TARGET_SYSTEM_ROOT is defined.
39234 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39236 PR tree-optimization/79090
39237 PR tree-optimization/33562
39238 PR tree-optimization/61912
39239 PR tree-optimization/77485
39240 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39241 and computed trims into the dump file.
39243 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39245 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39247 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39250 * gimplify.c (gimplify_init_constructor): If want_value and
39251 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39255 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39256 sequence. Formatting fixes.
39257 (doloop_optimize): Formatting fixes.
39260 * gcc.c (debug_level_greater_than_spec_func): New function.
39261 (static_spec_functions): Add debug-level-gt spec function.
39262 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39264 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39265 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39266 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39267 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39268 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39269 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39271 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39273 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39274 QImode fixups to general and mask registers only.
39276 2017-01-16 Carl Love <cel@us.ibm.com>
39278 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39279 for built-in functions
39280 vector signed char vec_nabs (vector signed char)
39281 vector signed short vec_nabs (vector signed short)
39282 vector signed int vec_nabs (vector signed int)
39283 vector signed long long vec_nabs (vector signed long long)
39284 vector float vec_nabs (vector float)
39285 vector double vec_nabs (vector double)
39286 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39288 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39289 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39290 * doc/extend.texi: Update the documentation file for the new built-in
39293 2017-01-16 Martin Sebor <msebor@redhat.com>
39295 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39298 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39300 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39301 UNSPEC_VSX__XXSPLTD to require special splat handling.
39303 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39306 * system.h: Poison strndup.
39308 2017-01-16 Alan Modra <amodra@gmail.com>
39311 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39314 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39316 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39318 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39320 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39321 call recog here. Assert that INSN_CODE (insn) is non-negative.
39323 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39326 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39328 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39329 in the currently scheduled RTL fragment.
39331 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39333 PR rtl-optimization/78751
39334 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39337 2017-01-14 Jeff Law <law@redhat.com>
39339 PR tree-optimization/79090
39340 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39341 variable length stores.
39342 (compute_trims): Delete dead assignment to *trim_tail.
39343 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39346 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39348 PR rtl-optimization/78626
39349 PR rtl-optimization/78727
39350 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39351 of a block, and split such blocks after everything else is finished.
39353 2017-01-14 Alan Modra <amodra@gmail.com>
39356 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39357 target legitimate_combined_insn.
39358 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39359 (rs6000_legitimate_combined_insn): New function.
39360 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39362 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39363 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39364 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39366 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39368 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39370 2017-01-13 Jeff Law <law@redhat.com>
39372 PR tree-optimization/33562
39373 PR tree-optimization/61912
39374 PR tree-optimization/77485
39375 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39377 (delete_dead_assignment): Likewise.
39378 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39379 statement to delete_dead_call and delete_dead_assignment.
39381 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39384 * substring-locations.c (format_warning_va): Strengthen case 1 so
39385 that both endpoints of the substring must be within the format
39386 range for just the substring to be printed.
39388 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39390 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39391 * config/i386/i386.c (ix86_target_string): Add missing options
39392 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39393 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39394 flags_other and ix86_target_other to flags2_other. Display unknown
39396 (ix86_valid_target_attribute_inner_p): Add missing options and
39397 reorder options by implied ISAs, as in ix86_target_string.
39399 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39401 * hash-table.h (hash_table::too_empty_p): New function.
39402 (hash_table::expand): Use it.
39403 (hash_table::traverse): Likewise.
39404 (hash_table::empty_slot): Use sizeof (value_type) instead of
39405 sizeof (PTR) to convert bytes to elements. Shrink the table
39406 if the current size is excessive for the current number of
39409 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39411 * ira-costs.c (record_reg_classes): Break from the inner loop
39412 early once alt_fail is known to be true. Update outer loop
39413 handling accordingly.
39415 2017-01-13 Jeff Law <law@redhat.com>
39417 * tree-ssa-dse.c (decrement_count): New function.
39418 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39419 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39420 when we know the partially dead statement is a mem* function.
39422 PR tree-optimization/61912
39423 PR tree-optimization/77485
39424 * tree-ssa-dse.c: Include expr.h.
39425 (maybe_trim_constructor_store): New function.
39426 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39428 PR tree-optimization/33562
39429 PR tree-optimization/61912
39430 PR tree-optimization/77485
39431 * doc/invoke.texi: Document new dse-max-object-size param.
39432 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39433 * tree-ssa-dse.c: Include params.h.
39434 (dse_store_status): New enum.
39435 (initialize_ao_ref_for_dse): New, partially extracted from
39437 (valid_ao_ref_for_dse, normalize_ref): New.
39438 (setup_live_bytes_from_ref, compute_trims): Likewise.
39439 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39440 (maybe_trim_partially_dead_store): Likewise.
39441 (maybe_trim_complex_store): Likewise.
39442 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39443 Track what bytes live from the original store. Return tri-state
39444 for dead, partially dead or live.
39445 (dse_dom_walker): Add constructor, destructor and new private members.
39446 (delete_dead_call, delete_dead_assignment): New extracted from
39448 (dse_optimize_stmt): Make a member of dse_dom_walker.
39449 Use initialize_ao_ref_for_dse.
39451 PR tree-optimization/33562
39452 PR tree-optimization/61912
39453 PR tree-optimization/77485
39454 * sbitmap.h (bitmap_count_bits): Prototype.
39455 (bitmap_clear_range, bitmap_set_range): Likewise.
39456 * sbitmap.c (bitmap_clear_range): New function.
39457 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39459 2017-01-13 Martin Liska <mliska@suse.cz>
39462 * function.c (set_cfun): Add new argument force.
39463 * function.h (set_cfun): Likewise.
39464 * ipa-inline-transform.c (inline_call): Use the function when
39465 strict alising from is dropped for function we inline to.
39467 2017-01-13 Richard Biener <rguenther@suse.de>
39469 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39470 for dumping GIMPLE INTEGER_CSTs.
39472 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39474 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39475 to 201112L since C++17.
39477 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39480 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39481 if -fsanitize=kernel-address is present.
39483 2017-01-13 Richard Biener <rguenther@suse.de>
39485 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39486 as _Literal ( type ) number in case usual suffixes do not
39487 preserve all information.
39489 2017-01-13 Richard Biener <rguenther@suse.de>
39491 PR tree-optimization/77283
39492 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39493 and ssa-iterators.h.
39494 (is_feasible_trace): Implement a cost model based on joiner
39497 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39500 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39501 char or short to __float128/_Float128 directly.
39503 2017-01-12 Martin Sebor <msebor@redhat.com>
39505 to -Wformat-overflow.
39506 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39507 (min_bytes_remaining): Same.
39508 (get_string_length): Same.
39509 (format_string): Same.
39510 (format_directive): Same.
39512 (pass_sprintf_length::handle_gimple_call): Same.
39514 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39516 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39517 info.nowrite calls with no lhs that can't throw. Return bool
39518 whether gsi_remove has been called or not.
39519 (pass_sprintf_length::handle_gimple_call): Return bool whether
39520 try_substitute_return_value called gsi_remove. Formatting fix.
39521 (pass_sprintf_length::execute): Don't use gsi_remove if
39522 handle_gimple_call returned true.
39525 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39526 be removed due to side-effects, don't remove following barrier nor
39527 turn the successor edge into fallthru edge.
39529 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39532 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39533 element-reversing loads and stores as not swappable.
39535 2017-01-12 Nathan Sidwell <nathan@acm.org>
39536 Nicolai Stange <nicstange@gmail.com>
39538 * combine.c (try_combine): Don't ignore result of overlap checking
39539 loop. Combine overlap & asm check into single loop.
39541 2017-01-12 Richard Biener <rguenther@suse.de>
39543 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39544 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39546 2017-01-12 Richard Biener <rguenther@suse.de>
39548 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39549 and TS_TARGET_OPTION directly derive from TS_BASE.
39550 * tree-core.h (tree_optimization_option): Derive from tree_base.
39551 (tree_target_option): Likewise.
39553 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39555 * config/i386/i386.c (memory_address_length): Increase len
39556 only when rip_relative_addr_p returns false.
39558 2017-01-11 Julia Koval <julia.koval@intel.com>
39560 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39561 (OPTION_MASK_ISA_SGX_SET): New.
39562 (ix86_handle_option): Handle OPT_msgx.
39563 * config.gcc: Added sgxintrin.h.
39564 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39565 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39566 * config/i386/i386.c (ix86_target_string): Add -msgx.
39568 (ix86_option_override_internal): Handle new options.
39569 (ix86_valid_target_attribute_inner_p): Add sgx.
39570 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39571 * config/i386/i386.opt: Add msgx.
39572 * config/i386/sgxintrin.h: New file.
39573 * config/i386/x86intrin.h: Add sgxintrin.h.
39575 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39578 * fold-const.c (maybe_nonzero_address): Return 1 for function
39580 (tree_single_nonzero_warnv_p): Don't handle function local objects
39584 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39587 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39590 * opts.c: Include "spellcheck.h"
39591 (struct string_fragment): New struct.
39592 (struct edit_distance_traits<const string_fragment &>): New
39594 (get_closest_sanitizer_option): New function.
39595 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39597 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39599 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39601 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39602 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39603 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39604 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39605 for initial die_offset if dwarf_split_debug_info.
39606 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39607 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39609 (output_skeleton_debug_sections): Formatting fix. Use
39610 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39611 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39613 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39615 * config/arm/cortex-a53.md: Add bypasses for
39616 cortex_a53_r2f_cvt.
39617 (cortex_a53_r2f): Only use for transfers.
39618 (cortex_a53_f2r): Likewise.
39619 (cortex_a53_r2f_cvt): Add reservation for conversions.
39620 (cortex_a53_f2r_cvt): Likewise.
39622 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39624 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39625 to all inlined functions, change static to extern.
39627 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39630 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39632 (arm_assemble_integer): Likewise.
39634 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39636 * config.gcc: Use new awk script to check CPU, FPU and architecture
39637 parameters for --with-... options.
39638 * config/arm/parsecpu.awk: New file
39639 * config/arm/arm-cpus.in: New file.
39640 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39642 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39644 * config/arm/t-arm: Update dependency rules.
39645 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39646 of processing .def files.
39647 * config/arm/genopt.sh: Deleted.
39648 * config/arm/gentune.sh: Deleted.
39649 * config/arm/arm-cores.def: Deleted.
39650 * config/arm/arm-arches.def: Deleted.
39651 * config/arm/arm-fpus.def: Deleted.
39652 * config/arm/arm-tune.md: Regenerated.
39653 * config/arm/arm-tables.opt: Regenerated.
39654 * config/arm/arm-cpu.h: New generated file.
39655 * config/arm/arm-cpu-data.h: New generated file.
39656 * config/arm/arm-cpu-cdata.h: New generated file.
39658 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39661 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39663 (input_varpool_node): Unpack dynamically_initialized bit.
39665 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39667 PR rtl-optimization/79032
39668 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39669 the alignment of the adjusted memory reference against that of MODE,
39670 instead of the alignment of the original memory reference.
39672 2017-01-11 Martin Jambor <mjambor@suse.cz>
39674 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39676 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39677 decorated functions.
39679 2017-01-11 Richard Biener <rguenther@suse.de>
39681 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39682 set range/nonnull info for PHI results. Do not set it on
39683 stmts marked for removal.
39685 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39687 * expr.c (store_field): In the bitfield case, fetch the return value
39688 from the registers before applying a single big-endian adjustment.
39689 Always do a final load for a BLKmode value not larger than a word.
39691 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39694 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39695 that we correctly handle column numbers greater than
39696 LINE_MAP_MAX_COLUMN_NUMBER.
39698 2017-01-10 Martin Sebor <msebor@redhat.com>
39700 PR middle-end/78245
39701 * gimple-ssa-sprintf.c (get_destination_size): Call
39702 {init,fini}object_sizes.
39703 * tree-object-size.c (addr_object_size): Adjust.
39704 (pass_through_call): Adjust.
39705 (pass_object_sizes::execute): Adjust.
39706 * tree-object-size.h (fini_object_sizes): Declare.
39708 2017-01-10 Martin Sebor <msebor@redhat.com>
39710 PR tree-optimization/78775
39711 * builtins.c (get_size_range): Move...
39712 * calls.c: ...to here.
39713 (alloc_max_size): Accept zero argument.
39714 (operand_signed_p): Remove.
39715 (maybe_warn_alloc_args_overflow): Call get_size_range.
39716 * calls.h (get_size_range): Declare.
39718 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39720 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39721 from TI's devices.csv file as of September 2016.
39722 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39724 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39726 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39727 * doc/invoke.texi: Likewise.
39728 * doc/md.texi: Likewise.
39729 * doc/objc.texi: Likewise.
39731 2017-01-10 Joshua Conner <joshconner@google.com>
39733 * config/arm/fuchsia-elf.h: New file.
39734 * config/fuchsia.h: New file.
39735 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39736 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39738 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39740 2016-01-10 Richard Biener <rguenther@suse.de>
39742 PR tree-optimization/79034
39743 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39744 Propagate out degenerate PHIs in the joiner.
39746 2017-01-10 Martin Liska <mliska@suse.cz>
39748 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39749 (sort_congruence_classes_by_decl_uid): Likewise.
39750 (sort_congruence_class_groups_by_decl_uid): Likewise.
39751 (sem_item_optimizer::merge_classes): Sort class, groups in these
39752 classes and members in the groups by DECL_UID of declarations.
39753 This would make merge operations stable.
39755 2017-01-10 Martin Liska <mliska@suse.cz>
39757 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39758 usage of m_classes_vec.
39759 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39760 (sem_item_optimizer::get_group_by_hash): Likewise.
39761 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39762 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39763 (sem_item_optimizer::verify_classes): Likewise.
39764 (sem_item_optimizer::process_cong_reduction): Likewise.
39765 (sem_item_optimizer::dump_cong_classes): Likewise.
39766 (sem_item_optimizer::merge_classes): Likewise.
39767 * ipa-icf.h (congruence_class_hash): Rename from
39768 congruence_class_group_hash. Remove declaration of m_classes_vec.
39770 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39772 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39773 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39774 * config.gcc: Add avx512vpopcntdqintrin.h.
39775 * config/i386/avx512vpopcntdqintrin.h: New.
39776 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39777 * config/i386/i386-builtin-types.def: Add new types.
39778 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39779 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39780 __builtin_ia32_vpopcountq_v8di_mask): New.
39781 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39782 __AVX512VPOPCNTDQ__.
39783 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39784 (PTA_AVX512VPOPCNTDQ): Define.
39785 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39786 TARGET_AVX512VPOPCNTDQ_P): Define.
39787 * config/i386/i386.opt: Add mavx512vpopcntdq.
39788 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39789 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39791 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39793 PR middle-end/77484
39794 * predict.def (PRED_CALL): Set to 67.
39796 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39798 * expr.c (store_field): In the bitfield case, if the value comes from
39799 a function call and is of an aggregate type returned in registers, do
39800 not modify the field mode; extract the value in all cases if the mode
39801 is BLKmode and the size is not larger than a word.
39803 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39806 * config/i386/cpuid.h: Fix undefined behavior.
39808 2017-01-04 Jeff Law <law@redhat.com>
39810 PR tree-optimization/79007
39811 PR tree-optimization/67955
39812 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39813 conservative for pt.null when flag_non_call_exceptions is on.
39815 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39817 PR translation/79019
39818 PR translation/79020
39819 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39820 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39821 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39823 * config/avr/avr.opt (maccumulate-args): Likewise.
39824 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39825 * common.opt (freport-bug): Likewise.
39826 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39827 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39828 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39829 translatable string.
39830 * config/i386/i386.c (function_value_32): Likewise.
39831 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39832 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39834 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39835 * common/config/msp430/msp430-common.c (msp430_handle_option):
39837 * symtab.c (symtab_node::verify_base): Likewise.
39838 * opts.c (set_debug_level): Likewise.
39839 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39840 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39841 missing whitespace to translatable strings.
39842 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39843 * config/sh/superh.h: Likewise.
39844 * config/i386/xopintrin.h: Likewise.
39845 * config/i386/znver1.md: Likewise.
39846 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39847 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39848 * double-int.h (struct double_int): Likewise.
39849 * double-int.c (div_and_round_double): Likewise.
39850 * wide-int.cc: Likewise.
39851 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39852 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39853 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39854 (crossjumps_occurred): ... this.
39855 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39858 PR tree-optimization/78899
39859 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39860 returning bool return struct loop *, NULL for failure and the new
39862 (versionable_outer_loop_p): Don't version outer loop if it has
39863 dont_vectorized bit set.
39864 (tree_if_conversion): When versioning outer loop, ensure
39865 tree_if_conversion is performed also on the inner loop of the
39866 non-vectorizable outer loop copy.
39867 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39868 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39869 prevent vectorization of it.
39870 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39871 the outer loop vectorization of the non-scalar version is attempted
39872 before vectorization of the inner loop in scalar version. If
39873 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39874 vectorization of its inner loop.
39875 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39876 has 2 inner loops, rename also on edges from bb whose single pred
39877 is outer_loop->header. Fix typo in function comment.
39879 2017-01-09 Martin Sebor <msebor@redhat.com>
39882 * asan.c (asan_emit_stack_protection): Increase local buffer size
39883 to avoid snprintf truncation warning.
39885 2017-01-09 Andrew Pinski <apinski@cavium.com>
39887 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39888 to reference thunderx2t99 for the tuning structure
39889 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39891 (thunderx2t99_extra_costs): This.
39892 * config/aarch64/aarch64-tune.md: Regenerate.
39893 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39894 (vulcan_addrcost_table): This.
39895 (vulcan_regmove_cost): Rename to ...
39896 (thunderx2t99_regmove_cost): This.
39897 (vulcan_vector_cost): Rename to ...
39898 (thunderx2t99_vector_cost): this.
39899 (vulcan_branch_cost): Rename to ...
39900 (thunderx2t99_branch_cost): This.
39901 (vulcan_tunings): Rename to ...
39902 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39903 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39905 2017-01-09 Martin Jambor <mjambor@suse.cz>
39909 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39910 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39911 (propagate_vr_accross_jump_function): Use the above function for all
39912 value range computations for pass-through jump functions and type
39913 converasion from explicit value range values.
39914 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39915 parameters from TYPE_ARG_TYPES.
39916 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39917 (ipa_write_node_info): Stream type of the actual argument.
39918 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39920 2017-01-09 Martin Liska <mliska@suse.cz>
39923 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39924 (lookup_compiler): Do not show error message with have_E.
39926 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39928 PR tree-optimization/78938
39929 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39930 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39931 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39932 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39935 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39937 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39940 2017-01-09 Richard Biener <rguenther@suse.de>
39942 PR tree-optimization/78997
39943 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39944 name condition properly.
39946 2017-01-09 Richard Biener <rguenther@suse.de>
39949 * dwarf2out.c (is_cxx): New overload with context.
39950 (is_naming_typedef_decl): Use it.
39952 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39954 * invoke.texi (Option Summary): Correct spacing in option lists
39955 and add line breaks to fix over-long lines.
39957 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39959 PR middle-end/17660
39961 * extend.texi (Common Variable Attributes): Add xref to GCC
39962 Internals manual to explain mode attribute keywords.
39964 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39967 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39968 and Preprocessor Options.
39969 (Options for Linking): Document -pthread here....
39970 (RS/6000 and PowerPC Options): ...not here.
39971 (Solaris 2 Options): ...or here.
39972 * doc/cppopts.texi: Document -pthread.
39974 2017-01-08 Martin Sebor <msebor@redhat.com>
39976 PR middle-end/77708
39977 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39978 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39979 New member functions.
39980 (format_directive): Used them.
39982 (pass_sprintf_length::handle_gimple_call): Same.
39983 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39984 to avoid truncation for any argument.
39985 (extract_affine_mul): Same.
39986 * tree.c (get_file_function_name): Same.
39988 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39990 PR middle-end/77484
39991 * predict.def (PRED_INDIR_CALL): Set to 86.
39993 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39995 PR preprocessor/54124
39996 * doc/cppopts.texi: Reformat -d subtable to list the full name
39997 of the options. Add cross-reference to the docs for the general
39998 compiler -d options.
39999 * doc/invoke.texi (Developer Options): Add cross-reference to the
40000 preprocessor-specific -d option documentation.
40002 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40004 PR preprocessor/13498
40005 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40006 redudant material, and reflect new command-line options.
40007 (System Headers): Likewise.
40009 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40011 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40012 -isystem, and -idirafter. Copy-edit.
40013 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40014 default for -ftrack-macro-expansion. Delete obsolete and
40015 badly-formatted implementation details about -fdebug-cpp output.
40016 * doc/cppwarnopts.texi: Copy-edit.
40018 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40021 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40022 that the transition from a max line width >= 1<<10 to narrower
40023 lines works correctly.
40025 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40027 * doc/options.texi (PerFunction): New.
40028 * opt-functions.awk (switch_flags): Map both Optimization and
40029 PerFunction to CL_OPTIMIZATION.
40030 * opth-gen.awk: Test for PerFunction flag along with
40032 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40033 it only when the latter is present. Skip those that don't in
40034 the hash function generator.
40035 * common.opt (fvar-tracking): Mark as PerFunction instead of
40037 (fvar-tracking-assignments): Likewise.
40038 (fvar-tracking-assignments-toggle): Likewise.
40039 (fvar-tracking-uninit): Likewise.
40041 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40043 PR translation/79018
40044 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40047 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40050 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40051 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40052 TARGET_LONG_JUMP_TABLE_OFFSETS.
40053 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40054 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40055 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40056 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40057 * config/m68k/m68k.md (tablejump expander): Likewise.
40058 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40059 TARGET_LONG_JUMP_TABLE_OFFSETS.
40060 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40061 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40063 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40064 David Holsgrove <david.holsgrove@xilinx.com>
40066 * common/config/microblaze/microblaze-common.c
40067 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40068 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40070 * config/microblaze/microblaze.c (microblaze_must_save_register)
40071 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40073 (microblaze_eh_return): New function.
40074 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40075 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40076 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40077 * config/microblaze/microblaze.md (eh_return): New pattern.
40079 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40081 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40082 GCC_DIAGNOSTIC_STRINGIFY): Define.
40084 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40086 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40088 * config/arm/arm.md (<mcrr>): New.
40090 * config/arm/arm.c (arm_arch5te): New.
40091 (arm_option_override): Set arm_arch5te.
40092 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40094 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40095 (arm_mcrr_qualifiers): ... this. New.
40096 (MRRC_QUALIFIERS): Define to...
40097 (arm_mrrc_qualifiers): ... this. New.
40098 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40100 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40101 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40102 (MRRCI, mrrc, MRRC): New.
40103 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40104 VUNSPEC_MRRC2): New.
40106 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40108 * config/arm/arm.md (<mcr>): New.
40110 * config/arm/arm.c (arm_coproc_builtin_available): Add
40111 support for mcr, mrc, mcr2 and mrc2.
40112 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40113 (arm_mcr_qualifiers): ... this. New.
40114 (MRC_QUALIFIERS): Define to ...
40115 (arm_mrc_qualifiers): ... this. New.
40116 (MCR_QUALIFIERS): Define to ...
40117 (arm_mcr_qualifiers): ... this. New.
40118 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40120 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40121 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40122 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40123 VUNSPEC_MRC2): New.
40125 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40127 * config/arm/arm.md (*ldc): New.
40131 * config/arm/arm.c (arm_coproc_builtin_available): Add
40132 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40133 (arm_coproc_ldc_stc_legitimate_address): New.
40134 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40135 'qualifier_const_pointer'.
40136 (LDC_QUALIFIERS): Define to...
40137 (arm_ldc_qualifiers): ... this. New.
40138 (STC_QUALIFIERS): Define to...
40139 (arm_stc_qualifiers): ... this. New.
40140 * config/arm/arm-protos.h
40141 (arm_coproc_ldc_stc_legitimate_address): New.
40142 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40143 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40144 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40145 stc2, stcl, stc2l): New.
40146 * config/arm/constraints.md (Uz): New.
40147 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40148 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40149 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40150 VUNSPEC_STC2L): New.
40152 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40154 * config/arm/arm.md (<cdp>): New.
40155 * config/arm/arm.c (neon_const_bounds): Rename this ...
40156 (arm_const_bounds): ... this.
40157 (arm_coproc_builtin_available): New.
40158 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40159 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40160 (CDP_QUALIFIERS): Define to...
40161 (arm_cdp_qualifiers): ... this. New.
40163 (arm_expand_builtin_args): Add case for 6 arguments.
40164 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40165 (arm_const_bounds): ... this.
40166 (arm_coproc_builtin_available): New.
40167 * config/arm/arm_acle.h (__arm_cdp): New.
40169 * config/arm/arm_acle_builtins.def (cdp): New.
40171 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40172 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40173 'arm_const_bounds'.
40174 * config/arm/types.md (coproc): New.
40175 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40176 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40177 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40178 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40180 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40182 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40183 (UBINOP_QUALIFIERS): New.
40185 (acle_builtin_data): New. Change comment.
40186 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40187 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40188 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40189 arm_acle_builtins.def.
40190 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40191 (arm_init_acle_builtins): New.
40192 (CRC32_BUILTIN): Remove.
40193 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40194 crc32cb, crc32ch and crc32cw.
40195 (arm_init_crc32_builtins): Remove.
40196 (arm_init_builtins): Use arm_init_acle_builtins rather
40197 than arm_init_crc32_builtins.
40198 (arm_expand_acle_builtin): New.
40199 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40200 * config/arm/arm_acle_builtins.def: New.
40202 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40204 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40205 (arm_builtin_datum): ... this.
40206 (arm_init_neon_builtin): Rename to ...
40207 (arm_init_builtin): ... this. Add a new parameters PREFIX
40208 and USE_SIG_IN_NAME.
40209 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40210 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40211 'arm_builtin_datum'.
40212 (arm_init_vfp_builtins): Likewise.
40213 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40214 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40215 (arm_expand_neon_args): Rename to ...
40216 (arm_expand_builtin_args): ... this. Rename builtin_arg
40217 enum values and differentiate between ARG_BUILTIN_MEMORY
40218 and ARG_BUILTIN_NEON_MEMORY.
40219 (arm_expand_neon_builtin_1): Rename to ...
40220 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40221 values, arm_expand_builtin_args and add bool parameter NEON.
40222 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40223 (arm_expand_vfp_builtin): Likewise.
40224 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40226 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40228 PR middle-end/77484
40229 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40230 * predict.c (tree_estimate_probability_bb): Reverse direction of
40231 polymorphic call predictor.
40233 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40235 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40236 (determine_pass_name_match): ...this new function and...
40237 (should_skip_pass_p): ...this new function.
40239 2017-01-06 Nathan Sidwell <nathan@acm.org>
40241 * ipa-visibility.c (function_and_variable_visibility): Reformat
40242 comments and long lines. Remove extrneous if.
40243 * symtab.c (symtab_node::make_decl_local): Fix code format.
40244 (symtab_node::set_section_for_node): Fix comment typo.
40246 2017-01-06 Martin Liska <mliska@suse.cz>
40249 * lra-constraints.c: Rename invariant to lra_invariant.
40250 * predict.c (set_even_probabilities): Initialize e to NULL.
40252 2017-01-05 Martin Sebor <msebor@redhat.com>
40254 PR tree-optimization/78910
40255 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40256 (format_integer): Correct off-by-one error in the handling
40257 of precision with negative numbers in signed conversions..
40259 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40261 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40263 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40265 PR tree-optimization/71016
40266 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40267 factor_out_conditional_conversion. Formatting fix.
40268 (factor_out_conditional_conversion): Add cond_stmt argument.
40269 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40270 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40273 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40275 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40276 read-rtl-function.o, and selftest-rtl.o.
40277 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40278 (selftest::aarch64_test_loading_full_dump): New function.
40279 (selftest::aarch64_run_selftests): New function.
40280 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40281 selftest::aarch64_run_selftests.
40282 * config/i386/i386.c
40283 (selftest::ix86_test_loading_dump_fragment_1): New function.
40284 (selftest::ix86_test_loading_call_insn): New function.
40285 (selftest::ix86_test_loading_full_dump): New function.
40286 (selftest::ix86_test_loading_unspec): New function.
40287 (selftest::ix86_run_selftests): Call the new functions.
40288 * emit-rtl.c (maybe_set_max_label_num): New function.
40289 * emit-rtl.h (maybe_set_max_label_num): New decl.
40290 * function.c (instantiate_decls): Guard call to
40291 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40292 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40294 * gensupport.c (gen_reader::gen_reader): Pass "false"
40295 for new "compact" param of rtx_reader.
40296 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40297 rather than an empty string for NULL strings.
40298 * read-md.c: Potentially include config.h rather than bconfig.h.
40299 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40300 (have_error): New global, copied from errors.c.
40301 (md_reader::read_name): Rename to...
40302 (md_reader::read_name_1): ...this, adding "out_loc" param,
40303 and converting "missing name or number" to returning false, rather
40305 (md_reader::read_name): Reimplement in terms of read_name_1.
40306 (md_reader::read_name_or_nil): New function.
40307 (md_reader::read_string): Handle "(nil)" by returning NULL.
40308 (md_reader::md_reader): Add new param "compact".
40309 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40310 (md_reader::read_file): New method.
40311 * read-md.h (md_reader::md_reader): Add new param "compact".
40312 (md_reader::read_file): New method.
40313 (md_reader::is_compact): New accessor.
40314 (md_reader::read_name): Convert return type from void to file_location.
40315 (md_reader::read_name_or_nil): New decl.
40316 (md_reader::read_name_1): New decl.
40317 (md_reader::m_compact): New field.
40318 (noop_reader::noop_reader): Pass "false" for new "compact" param
40320 (rtx_reader::rtx_reader): Add new "compact" param.
40321 (rtx_reader::read_rtx_operand): Make virtual and convert return
40322 type from void to rtx.
40323 (rtx_reader::read_until): New decl.
40324 (rtx_reader::handle_any_trailing_information): New virtual function.
40325 (rtx_reader::postprocess): New virtual function.
40326 (rtx_reader::finalize_string): New virtual function.
40327 (rtx_reader::m_in_call_function_usage): New field.
40328 (rtx_reader::m_reuse_rtx_by_id): New field.
40329 * read-rtl-function.c: New file.
40330 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40331 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40332 (selftest::verify_three_block_rtl_cfg): New decl.
40333 * read-rtl-function.h: New file.
40334 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40335 For host, include function.h, memmodel.h, and emit-rtl.h.
40336 (one_time_initialization): New function.
40337 (struct compact_insn_name): New struct.
40338 (compact_insn_names): New array.
40339 (find_code): Handle insn codes in compact dumps.
40340 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40341 (bind_subst_iter_and_attr): Likewise.
40342 (add_condition_to_string): Likewise.
40343 (add_condition_to_rtx): Likewise.
40344 (apply_attribute_uses): Likewise.
40345 (add_current_iterators): Likewise.
40346 (apply_iterators): Likewise.
40347 (initialize_iterators): Guard usage of apply_subst_iterator with
40348 #ifdef GENERATOR_FILE.
40349 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40350 (md_reader::read_mapping): Likewise.
40351 (add_define_attr_for_define_subst): Likewise.
40352 (add_define_subst_attr): Likewise.
40353 (read_subst_mapping): Likewise.
40354 (check_code_iterator): Likewise.
40355 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40357 (one_time_initialization): New function.
40358 (rtx_reader::read_until): New method.
40359 (read_flags): New function.
40360 (parse_reg_note_name): New function.
40361 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40362 Handle reuse_rtx ids.
40363 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40364 Add parsing support for RTL dumps, mirroring the special-cases in
40365 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40366 values, and calling handle_any_trailing_information.
40367 (rtx_reader::read_rtx_operand): Convert return type from void
40368 to rtx, returning return_rtx. Handle case 'e'. Call
40369 finalize_string on XSTR and XTMPL fields.
40370 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40371 "(nil)" values were omitted. Call the postprocess vfunc on the
40373 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40374 class ctor. Initialize m_in_call_function_usage. Call
40375 one_time_initialization.
40376 * rtl-tests.c (selftest::test_uncond_jump): Call
40377 set_new_first_and_last_insn.
40378 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40379 * selftest-rtl.c: New file.
40380 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40381 (selftest::get_insn_by_uid): New decl.
40382 * selftest-run-tests.c (selftest::run_tests): Call
40383 read_rtl_function_c_tests.
40384 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40385 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40388 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40390 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40391 operands in a special way. Assert that pos+len <= mode precision.
40393 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40395 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40396 3 argument Alias with unlimited for the negative form.
40397 (fno-vect-cost-model): Removed.
40399 2017-01-05 Martin Liska <mliska@suse.cz>
40401 * hsa-gen.c (gen_hsa_divmod): New function.
40402 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40404 2017-01-05 Martin Liska <mliska@suse.cz>
40407 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40410 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40412 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40413 small constant length operands.
40415 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40417 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40418 between loop iterations.
40420 2017-01-05 Martin Liska <mliska@suse.cz>
40423 * gimplify.c (gimplify_decl_expr): Compare to
40424 asan_poisoned_variables instread of checking flags.
40425 (gimplify_target_expr): Likewise.
40426 (gimplify_expr): Likewise.
40427 (gimplify_function_tree): Conditionally initialize
40428 asan_poisoned_variables.
40430 2017-01-04 Jeff Law <law@redhat.com>
40432 PR tree-optimizatin/78812
40433 * rtl.h (contains_mem_rtx_p): Prototype.
40434 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40435 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40436 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40437 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40438 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40440 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40442 * input.c (assert_char_at_range): Default-initialize actual_range.
40444 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40446 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40447 to match the caller.
40449 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40451 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40452 insns after final jump in test to emit dummy move.
40454 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40456 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40457 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40459 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40461 * multiple_target.c (create_dispatcher_calls): Init e_next.
40462 * tree-ssa-loop-split.c (split_loop): Init border.
40463 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40466 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40471 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40472 (altivec_register_operand): Do not return true if the operand
40473 contains a SUBREG mixing SImode and SFmode.
40474 (vsx_register_operand): Likewise.
40475 (vsx_reg_sfsubreg_ok): New predicate.
40476 (vfloat_operand): Do not return true if the operand contains a
40477 SUBREG mixing SImode and SFmode.
40478 (vint_operand): Likewise.
40479 (vlogical_operand): Likewise.
40480 (gpc_reg_operand): Likewise.
40481 (int_reg_operand): Likewise.
40482 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40483 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40484 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40486 (rs6000_emit_move_si_sf_subreg): New helper function.
40487 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40488 fixup SUBREGs involving SImode and SFmode.
40489 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40490 numbers for the new peephole2 optimization.
40491 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40492 the GLIBC math library that do AND/IOR/XOR operations on single
40493 precision floating point.
40494 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40495 target macros to say whether we need to avoid SUBREGs mixing
40497 (TARGET_ALLOW_SF_SUBREG): Likewise.
40498 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40499 (UNSPEC_SI_FROM_SF): Likewise.
40500 (iorxor): Change spacing.
40501 (and_ior_xor): New iterator for AND, IOR, and XOR.
40502 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40503 (movdi_from_sf_zero_ext): Likewise.
40504 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40505 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40506 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40507 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40508 (fms<mode>4): Likewise.
40509 (fnma<mode>4): Likewise.
40510 (fnms<mode>4): Likewise.
40511 (nfma<mode>4): Likewise.
40512 (nfms<mode>4): Likewise.
40514 2017-01-04 Marek Polacek <polacek@redhat.com>
40517 * doc/invoke.texi: Document -Wpointer-compare.
40519 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40521 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40524 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40525 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40528 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40531 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40532 documentation of the powerpc_popcntb_ok attribute.
40533 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40534 code to issue warning messages if a requested CPU configuration is
40535 not supported by the binary (assembler and loader) toolchain.
40536 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40537 made to define a built-in function that has been disabled.
40538 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40539 made to define a built-in function that has been disabled.
40540 (altivec_init_builtins): Add comment explaining why definition
40541 of the DST built-in functions is not preceded by an assertion
40542 check. Add assertions to prevent ICE if attempts are made to
40543 define an altivec predicate or an abs* built-in function that has
40545 (htm_init_builtins): Add comment explaining why definition of the
40546 htm built-in functions is not preceded by an assertion check.
40548 2017-01-04 Jeff Law <law@redhat.com>
40550 PR tree-optimizatin/67955
40551 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40552 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40553 the points-to solution does not include pt_null. Use DECL_PT_UID
40556 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40558 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40559 Use gen_int_mode instead of gen_lopwart for const_int operands.
40561 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40563 PR tree-optimization/71563
40564 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40565 out of range value - has low bits known to be zero.
40567 2017-01-04 Alan Modra <amodra@gmail.com>
40569 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40570 * configure: Regenerate.
40571 * config.in: Regenerate.
40573 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40576 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40577 a substring of the message, but strcmp with the whole message. Ifdef
40578 ENABLE_NLS, translate the message first using dgettext.
40580 2017-01-03 Jeff Law <law@redhat.com>
40582 PR tree-optimizatin/78856
40583 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40584 (mark_threaded_blocks): Remove code to truncate thread paths that
40585 cross multiple loop headers. Instead invalidate the cached loop
40586 iteration information and handle case of a thread path walking
40587 into an irreducible region.
40589 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40592 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40593 assertions. Add support for doing the signbit if the IEEE 128-bit
40594 floating point value is in a GPR.
40595 * config/rs6000/rs6000.md (Fsignbit): Delete.
40596 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40597 Update the length attribute if the value is in a GPR.
40598 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40599 the sign or zero extension instruction, since the value is always 0/1.
40600 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40603 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40604 extracting SImode to a GPR register so that we can generate a
40605 store, limit the vector to be in a traditional Altivec register
40606 for the vextuwrx instruction.
40608 2017-01-03 Ian Lance Taylor <iant@google.com>
40610 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40612 2017-01-03 Martin Sebor <msebor@redhat.com>
40614 PR tree-optimization/78696
40615 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40616 precision. Use MPFR for %f for greater fidelity. Correct handling
40618 (pass_sprintf_length::compute_format_length): Set width and precision
40619 specified by asrerisk to void_node for vararg functions.
40620 (try_substitute_return_value): Adjust dump output.
40622 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40624 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40626 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40628 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40629 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40630 -mlra/-mno-lra was passed to the compiler.
40632 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40634 PR rtl-optimization/65618
40635 * emit-rtl.c (try_split): Move initialization of "before" and
40636 "after" to just before the call to emit_insn_after_setloc.
40638 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40640 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40642 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40644 * dwarf2out.c (gen_enumeration_type_die): When
40645 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40647 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40649 PR tree-optimization/78965
40650 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40651 Change first argument from const call_info & to call_info &. For %n
40652 set info.nowrite to false.
40654 PR middle-end/78901
40655 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40656 possibly throwing calls.
40658 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40659 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40660 and fns handling, rather than in a separate case SSA_NAME.
40662 2017-01-02 Jeff Law <law@redhat.com>
40664 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40665 fixes for first_period and second_period variables.
40667 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40670 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40671 (*insvqi_1): New insn pattern.
40672 (*insvqi_1_mem_rex64): Ditto.
40673 (*insvqi_2): Ditto.
40674 (*insvqi_3): Rename from *insvqi.
40676 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40678 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40680 * doc/cfg.texi (Edges): Remove reference to Java.
40681 (Maintaining the CFG): Ditto.
40683 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40685 PR middle-end/77674
40686 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40687 transparent aliases.
40689 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40691 PR middle-end/77484
40692 * predict.def (PRED_CALL): Update hitrate.
40693 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40694 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40695 into direct/indirect/polymorphic variants.
40697 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40699 Update copyright years.
40701 * gcc.c (process_command): Update copyright notice dates.
40702 * gcov-dump.c (print_version): Ditto.
40703 * gcov.c (print_version): Ditto.
40704 * gcov-tool.c (print_version): Ditto.
40705 * gengtype.c (create_file): Ditto.
40706 * doc/cpp.texi: Bump @copying's copyright year.
40707 * doc/cppinternals.texi: Ditto.
40708 * doc/gcc.texi: Ditto.
40709 * doc/gccint.texi: Ditto.
40710 * doc/gcov.texi: Ditto.
40711 * doc/install.texi: Ditto.
40712 * doc/invoke.texi: Ditto.
40714 Copyright (C) 2017 Free Software Foundation, Inc.
40716 Copying and distribution of this file, with or without modification,
40717 are permitted in any medium without royalty provided the copyright
40718 notice and this notice are preserved.